mirror/hdf5
2
0
Fork 0
mirror of https://github.com/HDFGroup/hdf5.git synced 2024-11-27 02:10:55 +08:00
Code Issues Packages Projects Releases Wiki Activity
c424877351
hdf5/release_docs/HISTORY-1_8.txt
Larry Knox 89fbe00dec Merge pull request #426 in HDFFV/hdf5 from ~LRKNOX/hdf5_lrk:hdf5_1_10 to hdf5_1_10 
* commit '54957d37f5aa73912763dbb6e308555e863c43f4':
  Commit copyright header change for src/H5PLpkg.c which was added after running script to make changes.
  Add new files in release_docs to MANIFEST. Cimmit changes to Makefile.in(s) and H5PL.c that resulted from running autogen.sh.
  Merge pull request #407 in HDFFV/hdf5 from ~LRKNOX/hdf5_lrk:hdf5_1_10_1 to hdf5_1_10_1
  Change copyright headers to replace url referring to file to be removed and replace it with new url for COPYING file.
2017-04-25 16:05:36 -05:00

12345 lines
561 KiB
Plaintext
Raw Blame History

HDF5 History
============
This file contains development history of HDF5 1.8 branch
19. Release Information for hdf5-1.8.17
18. Release Information for hdf5-1.8.16
17. Release Information for hdf5-1.8.15
16. Release Information for hdf5-1.8.14
15. Release Information for hdf5-1.8.13
14. Release Information for hdf5-1.8.12
13. Release Information for hdf5-1.8.11
12. Release Information for hdf5-1.8.10-patch1
11. Release Information for hdf5-1.8.10
10. Release Information for hdf5-1.8.9
09. Release Information for hdf5-1.8.8
08. Release Information for hdf5-1.8.7
07. Release Information for hdf5-1.8.6
06. Release Information for hdf5-1.8.5
05. Release Information for hdf5-1.8.4
04. Release Information for hdf5-1.8.3
03. Release Information for hdf5-1.8.2
02. Release Information for hdf5-1.8.1
01. Release Information for hdf5-1.8.0
[Search on the string '%%%%' for section breaks of each release.]
%%%%1.8.17%%%%
HDF5 version 1.8.17 released on 2016-05-10
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.16 and
HDF5-1.8.17-*, and contains information on the platforms tested and
known problems in HDF5-1.8.17-*.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.17 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.17 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.17 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.17 (current
release) versus Release 1.8.16
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.16
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration
-------------
- Cmakehdf5: Added Ability to Run Multiple Make Commands
Added option --njobs to specify up to how many jobs to launch during
build (cmake) and testing (ctest).
(AKC - 2015/12/13, HDFFV-9612)
- Cmakehdf5: Added Szip Support and Verbose Option
Added --with-szlib to support the Szip library; and
--enable/disable-verbose to display all CMake process output.
(AKC - 2015/11/16, HDFFV-8932 and DAILYTEST-195)
- CMake minimum is now 3.1.0. (ADB - 2015/11/14)
- Large File System (LFS) Support has Changed in the Autotools
We assume that fseeko and ftello exist.
The *64 I/O functions and types are no longer explicitly used.
We now rely on a mapping provided by _FILE_OFFSET_BITS (or its
equivalent).
_LARGEFILE(64)_SOURCE is no longer exposed via AM_CPPFLAGS.
(DER - 2016/03/29, HDFFV-9626 and HDFFV-9541)
Library
-------
- New API Calls for Searching for External Dataset Storage
API calls that determine the search path for dataset external
storage were added. H5Pset/get_efile_prefix() API calls were added
to the library. These functions give control over the search path
for dataset external storage that has been configured with
H5Pset_external().
Additionally, the HDF5_EXTFILE_PREFIX environment variable can be
used to control the search path.
(DER - 2016/04/20, HDFFV-8740)
Parallel Library
----------------
- None
Tools
-----
- None
High-Level APIs
---------------
C Packet Table API
------------------
- Replacement of a Public Function with H5PTcreate
The existing function H5PTcreate_fl limits applications so they
can use the deflate compression only. The public function
H5PTcreate has been added to replace H5PTcreate_fl. H5PTcreate
takes a property list identifier to provide flexibility on
creation properties. This also removes the following warning:
"deprecated conversion from string constant to "char*"
[-Wwrite-strings]".
(BMR - 2016/04/25, HDFFV-9708, HDFFV-8615)
- New Public Functions: H5PTget_dataset and H5PTget_type
Two accessor functions have been added. H5PTget_dataset returns
the identifier of the dataset associated with the packet table,
and H5PTget_type returns the identifier of the datatype used by
the packet table.
(BMR - 2016/04/25, HDFFV-8623 patch 3)
- Regarding #ifdef VLPT_REMOVED
The #ifdef VLPT_REMOVED blocks have been removed from the packet
table (PT) library source except for the following functions:
+ H5PTis_varlen() has been made available again
+ H5PTfree_vlen_readbuff() is now H5PTfree_vlen_buff()
(BMR - 2016/04/25, HDFFV-442)
C++ Packet Table API
--------------------
- New Constructor in FL_PacketTable
An overloaded constructor has been added to FL_PacketTable and
takes a property list identifier to provide flexibility on
creation properties such as compression.
FL_PacketTable(hid_t fileID, const char* name, hid_t dtypeID,
hsize_t chunkSize = 0, hid_t plistID = H5P_DEFAULT)
(BMR - 2016/04/25, HDFFV-8623 patch 5)
- New Member Functions in PacketTable
Two accessor wrappers were added to class PacketTable.
PacketTable::GetDataset() returns the identifier of the dataset
associated with the packet table, and PacketTable::GetDatatype()
returns the identifier of the datatype that the packet table uses.
(BMR - 2016/04/25, HDFFV-8623 patch 4)
- New Member Functions with "char*" as an Argument
Overloaded functions were added to provide the "const char*"
argument; the existing version will be deprecated in future
releases. This also removes the following warning:
"deprecated conversion from string constant to "char*"
[-Wwrite-strings]".
(BMR - 2016/04/25, HDFFV-8623 patch 1, HDFFV-8615)
- Regarding #ifdef VLPT_REMOVED
The #ifdef VLPT_REMOVED blocks have been removed from the packet
table library source code except for the following functions:
+ VL_PacketTable::IsVariableLength() was moved to PacketTable
+ VL_PacketTable::FreeReadBuff() is now PacketTable::FreeBuff()
(BMR - 2016/04/25, HDFFV-442)
Fortran API
-----------
- None
C++ API
-------
- New Member Function in DSetCreatPropList
DSetCreatPropList::setNbit() was added to setup N-bit compression for
a dataset.
(BMR - 2016/04/25, HDFFV-8623 patch 7)
- New Overloaded "const" Member Functions in ArrayType
The two following functions were added:
ArrayType::getArrayNDims() const
ArrayType::getArrayDims() const
to provide const version, and the non-const version was marked
deprecated. In-memory array information, ArrayType::rank and
ArrayType::dimensions, were removed. This is an implementation
detail and should not affect applications.
(BMR, 2016/04/25, HDFFV-9725)
- New member function added
The assignment operator ArrayType::operator= is added because ArrayType
has pointer data members.
(BMR, 2016/03/07, HDFFV-9562)
Support for New Platforms, Languages, and Compilers
===================================================
- Mac OS X El Capitan 10.11.4 with compilers Apple clang/clang++
version 7.3.0 from Xcode 7.3, gfortran GNU Fortran (GCC) 5.2.0
and Intel icc/icpc/ifort version 16.0.2
Bug Fixes since HDF5-1.8.16
===========================
Configuration
-------------
- Updated Linux Language Level Flags to Match the Autotools. Removed
Linux-specific Flags from OS X.
An addition to the flags simply being out of sync with the Autotools,
the Linux flags were used on OS X builds which led to symbols not being
found. Although this was non-fatal and compilation continued (implicit
definitions were used by the compiler and the symbols resolved at link
time), a large number of warnings were raised.
Linux changes:
* CHANGED: _POSIX_C_SOURCE (from 199605 to 200112L)
* ADDED: _GNU_SOURCE
* REMOVED: _BSD_SOURCE
* REMOVED: _DEFAULT_SOURCE
(DER - 2015/12/08, HDFFV-9627)
- The --enable-clear-file-buffers configure Option was Non-functional
so the Feature was Always Enabled (its default value).
Regardless of the configure flag, the setting was always enabled when
the Autotools were used to configure HDF5. This was due to the "no"
option being processed after the "*" option in configure.ac so "*"
matched first. CMake was unaffected.
The option now works correctly.
NOTE that builders are always advised to leave this option enabled.
When disabled, buffers that are written to disk may contain the
memory's previous contents, which may include secure information.
The performance overhead of the feature (a single memset call per
allocation) is minimal.
(DER - 2016/02/03, HDFFV-9676)
- Added a patch to remove '"'s from arguments for MPI compilers that
were causing errors compiling H5lib_settings.c with SGI MPT.
(LRK - 2016/04/20, HDFFV-9439)
Library
-------
- Fixed shared file pointer problem which caused a crash when running a
program provided by a user.
(VC - 2016/04/01, HDFFV-9469)
- Fixed some format string warnings that prevent compiling with
-Werror=format-security on gcc.
These only appeared in error messages and would not cause problems
under normal operation.
(DER - 2016/01/13, HDFFV-9640)
- Fixed a library segmentation fault when accessing a corrupted
file provided by a user.
(MSC - 2016/02/19, HDFFV-9670)
Parallel Library
----------------
- None
Performance
-------------
- None
Tools
-----
- h5dump: Sub-setting Fixed for Dimensions Greater than Two
When a dataset has more than two dimensions, sub-setting would
incorrectly calculate the data that needed to be displayed.
Added in block and stride calculations that account for dimensions
greater than two. NOTE: lines that have line breaks inserted
because of display length calculations may have index info that
is incorrect until the next dimension break.
(ADB - 2016/03/07, HDFFV-9698)
- h5dump: Issue with Argument Segmentation Fault
When an argument with an optional value was at the end of the command
line with a value, h5dump would crash. Reworked check for remaining
arguments.
(ADB - 2016/03/07, HDFFV-9570, HDFFV-9684)
- h5dump: Issue with Default Fill Value
Added all default cases of fill value to the display of fill value.
(ADB -, 2016/03/07, HDFFV-9241)
- h5dump: Clarified Help
Clarified usage of -O F option in h5dump utility help.
(ADB - 2016/03/07, HDFFV-9066)
- h5dump: Issue with Double Free Fault
Added a check for filename not null before calling free().
(ADB - 2016/01/27, HDFFV-9639)
- VS2015 Release Changed how Timezone was Handled
Created a function, HDget_timezone, in H5system.c. Replaced
timezone variable usage with function call.
(ADB - 2015/11/02, HDFFV-9550)
Fortran API
-----------
- None
C++ API
-------
- Removal of Obsolete Methods
The overloaded methods which had parameters that should be const
but were not have been removed.
(BMR - 2016/01/13, HDFFV-9789)
High-Level APIs:
---------------
- Fixed Memory Leak in Packet Table API
Applied user's patch to fix memory leak in the creation of a
packet table.
(BMR - 2016/04/25, HDFFV-9700)
Fortran High-Level APIs:
------------------------
- None
Testing
-------
- None
Supported Platforms
===================
The following platforms are supported and have been tested for this release.
They are built with the configure process unless specified otherwise.
AIX 6.1 xlc/xlc_r 10.1.0.5
(NASA G-ADA) xlC/xlC_r 10.1.0.5
xlf90/xlf90_r 12.1.0.6
Linux 2.6.32-573.22.1.el6 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(platypus) Version 4.4.7 20120313
Version 4.9.3, Version 5.2.0
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 15.7-0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 15.0.3.187 Build 20150407
MPICH 3.1.4 compiled with GCC 4.9.3
Linux 2.6.32-504.8.1.el6.ppc64 gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-16)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-16)
(ostrich) GNU Fortran (GCC) 4.4.7 20120313 (Red Hat 4.4.7-16)
IBM XL C/C++ V13.1
IBM XL Fortran V15.1
Linux 3.10.0-229.14.1.el7 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(kituo/moohan) Version 4.8.5 20150623 (Red Hat 4.8.5-4)
Version 5.2.0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 15.0.3.187 Build 20150407
MPICH 3.1.4 compiled with GCC 4.9.3
SunOS 5.11 32- and 64-bit Sun C 5.12 SunOS_sparc
(emu) Sun Fortran 95 8.6 SunOS_sparc
Sun C++ 5.12 SunOS_sparc
Windows 7 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Visual Studio 2015 (cmake)
Cygwin(CYGWIN_NT-6.1 2.2.1(0.289/5/3) gcc(4.9.3) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Visual Studio 2015 (cmake)
Windows 8.1 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Windows 8.1 x64 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Mac OS X Mt. Lion 10.8.5 Apple clang/clang++ version 5.1 from Xcode 5.1
64-bit gfortran GNU Fortran (GCC) 4.8.2
(swallow/kite) Intel icc/icpc/ifort version 15.0.3
Mac OS X Mavericks 10.9.5 Apple clang/clang++ version 6.0 from Xcode 6.2
64-bit gfortran GNU Fortran (GCC) 4.9.2
(wren/quail) Intel icc/icpc/ifort version 15.0.3
Mac OS X Yosemite 10.10.5 Apple clang/clang++ version 6.1 from Xcode 7.0
64-bit gfortran GNU Fortran (GCC) 4.9.2
(osx1010dev/osx1010test) Intel icc/icpc/ifort version 15.0.3
Mac OS X El Capitan 10.11.4 Apple clang/clang++ version 7.3.0 from Xcode 7.3
64-bit gfortran GNU Fortran (GCC) 5.2.0
(osx1011dev/ox1011test) Intel icc/icpc/ifort version 16.0.2
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
SunOS 5.11 32-bit n y/y n y y y
SunOS 5.11 64-bit n y/y n y y y
Windows 7 y y/y n y y y
Windows 7 x64 y y/y n y y y
Windows 7 Cygwin n y/y n y y n
Windows 8.1 n y/y n y y y
Windows 8.1 x64 n y/y n y y y
Mac OS X Mountain Lion 10.8.5 64-bit n y/y n y y y
Mac OS X Mavericks 10.9.5 64-bit n y/y n y y y
Mac OS X Yosemeti 10.10.5 64-bit n y/y n y y y
AIX 6.1 32- and 64-bit n y/n n y y y
CentOS 6.7 Linux 2.6.32 x86_64 GNU y y/y y y y y
CentOS 6.7 Linux 2.6.32 x86_64 Intel n y/y n y y y
CentOS 6.7 Linux 2.6.32 x86_64 PGI n y/y n y y y
CentOS 7.1 Linux 3.10.0 x86_64 GNU y y/y y y y y
CentOS 7.1 Linux 3.10.0 x86_64 Intel n y/y n y y y
Linux 2.6.32-431.11.2.el6.ppc64 n y/n n y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
SunOS 5.11 32-bit y y y y
SunOS 5.11 64-bit y y y y
Windows 7 y y y y
Windows 7 x64 y y y y
Windows 7 Cygwin n n n y
Windows 8.1 y y y y
Windows 8.1 x64 y y y y
Mac OS X Mountain Lion 10.8.5 64-bit y n y y
Mac OS X Mavericks 10.9.5 64-bit y n y y
Mac OS X Yosemeti 10.10.5 64-bit y n y y
AIX 6.1 32- and 64-bit y n n y
CentOS 6.7 Linux 2.6.32 x86_64 GNU y y y y
CentOS 6.7 Linux 2.6.32 x86_64 Intel y y y y
CentOS 6.7 Linux 2.6.32 x86_64 PGI y y y y
CentOS 7.1 Linux 3.10.0 x86_64 GNU y y y y
CentOS 7.1 Linux 3.10.0 x86_64 Intel y y y y
Linux 2.6.32-431.11.2.el6.ppc64 y y y y
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
Linux 2.6.18-431.11.2.el6 g95 (GCC 4.0.3 (g95 0.94!)
#1 SMP x86_64 GNU/Linux
(platypus)
Windows 7 Visual Studio 2008 (cmake)
Windows 7 x64 Visual Studio 2008 (cmake)
Windows 10 Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Windows 10 x64 Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Debian7.5.0 3.2.0-4-amd64 #1 SMP Debian 3.2.51-1 x86_64 GNU/Linux
gcc (Debian 4.7.2-5) 4.7.2
GNU Fortran (Debian 4.7.2-5) 4.7.2
(cmake and autotools)
Fedora20 3.15.3-200.fc20.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.8.3 20140624 (Red Hat 4.8.3-1)
GNU Fortran (GCC) 4.8.3 20140624 (Red Hat 4.8.3-1)
(cmake and autotools)
SUSE 13.1 3.11.10-17-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.8.1
GNU Fortran (SUSE Linux) 4.8.1
(cmake and autotools)
Ubuntu 14.04 3.13.0-35-generic #62-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.9.1-0ubuntu1) 4.9.1
GNU Fortran (Ubuntu/Linaro 4.9.1-0ubuntu1) 4.9.1
(cmake and autotools)
hopper.nersc.gov PrgEnv-gnu/5.2.40
gcc (GCC) 4.9.2 20141030 (Cray Inc.)
GNU Fortran (GCC) 4.9.2 20141030 (Cray Inc.)
g++ (GCC) 4.9.2 20141030 (Cray Inc.)
Known Problems
==============
* On windows platforms in debug configurations, the VFD flush1 tests will fail
with the split and multi VFD drivers. These tests will display a modal debug
dialog which must be answered or wait for the test timeout to expire.
(ADB - 2014/06/23 - HDFFV-8851)
* CLANG compiler with the options -fcatch-undefined-behavior and -ftrapv
catches some undefined behavior in the alignment algorithm of the macro DETECT_I
in H5detect.c (Issue 8147). Since the algorithm is trying to detect the alignment
of integers, ideally the flag -fcatch-undefined-behavior shouldn't to be used for
H5detect.c. In the future, we can separate flags for H5detect.c from the rest of
the library. (SLU - 2013/10/16)
* Make provided by Solaris fails in "make check". Solaris users should use
gmake to build and install the HDF5 software. (AKC - 2013/10/08 - HDFFV-8534)
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. the issue. HL and C++ shared libraries should now be
working as intended, however.
(MAM - 2011/04/20)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.16%%%%
HDF5 version 1.8.16 released on 2015-11-10
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.15 and
HDF5-1.8.16, and contains information on the platforms tested and
known problems in HDF5-1.8.16.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.16 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.16 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.16 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.16 (current
release) versus Release 1.8.15":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.15
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration and Build
-------------
- The thread-safety + high-level library combination has been marked
as "unsupported" in the Autotools
The global lock used by the thread-safety feature has never been
raised to the high-level library level, making it possible that
the library state could change if a context switch were to occur in
a high-level library call. Because of this, the combination of
thread-safety and high-level library is officially unsupported by
The HDF Group.
In the past, although this combination has never been supported, this
was not enforced by the build systems. These changes will cause an
Autotools configure step to fail if --enable-threadsafe and
--enable-hl are combined unless additional options are specified.
Since the high-level library is built by default, this means that
these extra configuration options will need to be used any time
--enable-threadsafe is selected.
To build with --enable-threadsafe, either:
1) Use --disable-hl to disable the high-level library (recommended)
2) Use --enable-unsupported to build the high-level library with
the thread-safety feature.
(DER - 2015/09/10 HDFFV-8719)
- Using CMake now builds both static and shared libraries.
The CMake files have been updated to build both static and shared
libraries, with tools only built statically. The packaging of the
libraries and tools will include cmake-config files that allows
projects to choose either shared or static (default) libraries
to be found with the find_package command using the COMPONENTS
keyword and a list of components. The imported libraries will
include any interface specific settings and dependent libraries.
The default setting for BUILD_SHARED_LIBS has changed from OFF
to ON, which builds both static and shared libraries. The static
libraries are always built because of tools requirements.
(ADB - 2015/08/24 HDFFV-5881)
- Inline functions now correctly annotated with Autotools builds.
The method used to detect the appropriate inline markup scheme was
nonfunctional in Autotools builds. The Autotools have been modified
to correctly detect the compiler's inline markup symbol and apply it
to the source. Note that only a very small number of internal
functions are marked inline so this was not a very big change or
likely to significantly affect performance.
As a part of this change, the H5_inline symbol no longer appears in
H5pubconf.h.
(DER - 2015/08/13 HDFFV-9119, HDFFV-9421)
- Removed obsolete/unmaintained files from config/
Several files were removed from the config directory. These files
represent old operating systems, were no longer necessary, and/or
were no longer maintained. configure.ac was updated to reflect the
removed files.
Removed:
craynv
dec-flags
hpux11.23
ia64-linux-gnu
nec-superux14.1
sv1-cray
x86_64-redstorm-linux-gnu
powerpc-ibm-aix5.x
As a part of this work, a few lines that deal with locating the
sys/fpu.h header on SGI machines and some OSF/1 configure lines
were also removed. The Solaris config was also renamed to not have
a version number since the version number was ignored by configure
as it applies to all Solaris versions.
(DER - 2015/09/04 HDFFV-9116)
- Removed the FP_TO_INTEGER_OVERFLOW_WORKS macro/defines from the library
This was for working around bugs in the Cray X1 compiler, which is no
longer supported.
(DER - 2015/09/09 HDFFV-9191)
- Removed the H5_SW_ULONG_TO_FP_BOTTOM_BIT_WORKS and
H5_FP_TO_ULLONG_BOTTOM_BIT_WORKS symbols and associated code.
H5_SW_ULONG_TO_FP_BOTTOM_BIT_WORKS was a work-around on old 64-bit
SGI and Solaris systems.
H5_FP_TO_ULLONG_BOTTOM_BIT_WORKS was a work-around for old PGI
compilers on Linux.
Neither of these were used in any current library code and only appeared
in the dt_arith test.
(DER - 2015/09/09 HDFFV-9187)
- Removed CONVERT_DENORMAL_FLOAT symbol and associated code from the
library.
This was only set in configure files for Cray and NEC computers. These
config files no longer exist so there is no effect on currently
supported platforms.
(DER - 2015/09/09 HDFFV-9188)
- Removed _BSD_SOURCE and _DEFAULT_SOURCE from configure.ac
These are old BSD-compatibility symbols that are no longer needed by
the library.
(DER - 2015/09/10 HDFFV-9079)
- Removed HW_FP_TO_LLONG_NOT_WORKS symbol and associated code from the
library.
This was part of a work-around for the VS.NET 2003 compiler, which is
no longer supported.
(DER - 2015/09/10 HDFFV-9189)
- Removed the BAD_LOG2_CODE_GENERATED symbol and associated code from the
library.
This was an IRIX work-around.
(DER - 2015/09/11 HDFFV-9195)
- Decoupled shared object version numbers for wrapper libraries from the
shared object version number for the HDF5 library. These will be
maintained on an individual basis according to the interface changes
specific to these wrapper libraries.
For HDF5 1.8.16 the shared object version numbers were changed from
10.0.1 to 10.1.0 for the HDF5 library due to added APIs. For the C++
wrapper library they were changed from 10.0.1 to 11.0.0 due to changes
in existing APIs. For all other wrapper libraries the versions were
changed from 10.0.1 to 10.0.2 because while the APIs had no changes
there have been changes in code that did not result in changes to their
interfaces.
(LRK - 2015/10/28)
Library
-------
- H5F_ACC_DEBUG flag for H5Fopen/create: functionality removed
The symbol was used to emit some extra debugging information
for HDF Group developers in the multi VFD. The underlying
functionality has been removed due to disuse. The symbol
remains defined since it was visible in H5Fpublic.h but it
has been set to zero and has no effect anywhere in the library.
(DER - 2015-05-02, HDFFV-1074)
- New public API call: H5is_library_threadsafe()
This API call indicates if the library was built with thread-
safety enabled.
(DER - 2015-09-01, HDFFV-9496)
Parallel Library
----------------
- None
Tools
-----
- None
High-Level APIs
---------------
- None
Fortran API
-----------
- None
C++ API
-------
- Class H5::ObjCreatPropList is added for the object creation property
list class.
Class H5::ObjCreatPropList is derived from H5::PropList and is a
baseclass of H5::DSetCreatPropList. Additional property list classes
will be derived from H5::ObjCreatPropList when they are added to the
library in future releases.
(BMR, 2015/10/13, Part of HDFFV-9169)
- New Wrappers for C Functions H5P[s/g]et_attr_phase_change and
H5P[s/g]et_attr_creation_order.
Wrappers were added to class H5::ObjCreatPropList for the C Functions
H5Pset_attr_phase_change: H5::ObjCreatPropList::setAttrPhaseChange
H5Pget_attr_phase_change: H5::ObjCreatPropList::getAttrPhaseChange
H5Pset_attr_creation_order: H5::ObjCreatPropList::setAttrCrtOrder
H5Pget_attr_creation_order: H5::ObjCreatPropList::getAttrCrtOrder
(BMR, 2015/10/13, Part of HDFFV-9167 and HDFFV-9169)
Support for New Platforms, Languages, and Compilers
===================================================
- Added VS2015 with Intel Fortran 16 to supported Windows 7 platforms
Bug Fixes since HDF5-1.8.15
===========================
Configuration
-------------
- CMake test for long long printf format improved
The CMake configuration test for determining the printf format string
for printing a long long integer was fixed. The test would crash
when executed with VS2015.
(ADB - 2015-10-21 HDFFV-9488)
Library
-------
- VS2015 removed global variable timezone
The usage of the global variable timezone was modified for VS2015 by
adding an alias to Windows builds.
(ADB - 2015-10-23 HDFFV-9550)
- Fix potential error in H5Iclear_type
If the ID type's close callback could close another ID of the same type,
H5Iclear_type could occasionally run into problems due to the inner
workings of the skip list package. This could potentially cause an
error on library shutdown without calling H5Iclear_type directly. This
issue has been fixed.
(NAF - 2015-08-12)
- Fix uninitialized memory in dataspace selection code
When creating a dataspace with H5Screate and setting the extent with
H5Sextent_copy, the selection offset was not initialized, potentially
causing invalid I/O. There may be other cases where this happened.
Modified the library to always initialize the offset.
(NAF - 2015-09-08)
- Truncate file in H5Fflush() if EOA != EOF to avoid file
corruption in certain scenarios
In the following scenario, the resulting HDF5 file would be
incorrectly corrupted because the truncate operation
was at some point wrongly moved out of the flush operation:
- Create a new file with a single dataset.
- Write parts of the dataset (make sure that some values at
the end of the dataset are not initialized).
- Flush the file.
- Crash the program.
- Try to open the file with h5dump or h5debug, but the
resulting file is corrupted.
(MSC - 2015-06-15 HDFFV-9418)
Parallel Library
----------------
-
(XYZ - YYYY/MM/DD HDFFV-####)
Performance
-------------
- None
Tools
-----
- VS2015 changed the default format for printing of exponents
VS2015 default format for exponents changed with the elimination
of the leading '0'. CMake now tests for the VS2015 compiler and
adjusts which reference files are used by tests.
(ADB - 2015-10-23 HDFFV-9550)
- Fixed h5repack with user-defined filters
h5repack would throw a buffer overrun exception on Windows when
parsing a user-defined filter ID of 5 digits. A local variable in
the parse routine was not of sufficient size.
(ADB - 2015/09/01 HDFFV-9515)
Fortran API
------------
- None
C++ API
------
- Removed memory leaks
The static global constant objects were changed to constant references
referencing dynamically allocated objects. This ensures that the clean-up
process in the C++ library occurs before the termination of the C library
and prevents memory leaks because the previous global constants were not
properly deleted before the C library termination.
(BMR, 2015/10/13, HDFFV-9529)
- Fixed the problem about identifiers being closed prematurely.
The C++ library needs to increment the ID's reference counter when it is
duplicated in the form of C++ objects, but not when the ID is obtained
from a C function. With this approach, both problems, prematurely
closing ID's and memory leaks due to ID's not being closed, should be
eliminated.
(BMR, 2015/10/15, HDFFV-7947)
High-Level APIs:
------
- None
Fortran High-Level APIs:
------------------------
- None
Testing
-------
- None
Supported Platforms
===================
The following platforms are supported and have been tested for this release.
They are built with the configure process unless specified otherwise.
AIX 6.1 xlc/xlc_r 10.1.0.5
(NASA G-ADA) xlC/xlC_r 10.1.0.5
xlf90/xlf90_r 12.1.0.6
Linux 2.6.32-573.3.1.el6 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(platypus) Version 4.4.7 20120313
Version 4.8.4, Version 5.2.0
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 15.7-0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 15.0.3.187 Build 20150407
MPICH 3.1.4 compiled with GCC 4.9.3
Linux 2.6.32-504.8.1.el6.ppc64 gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-11)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-11)
(ostrich) GNU Fortran (GCC) 4.4.7 20120313 (Red Hat 4.4.7-11)
IBM XL C/C++ V13.1
IBM XL Fortran V15.1
Linux 3.10.0-229.14.1.el7 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(kituo/moohan) Version 4.8.3 20140911 (Red Hat 4.8.3-9)
Version 5.2.0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 15.0.3.187 Build 20150407
MPICH 3.1.4 compiled with GCC 4.9.3
SunOS 5.11 32- and 64-bit Sun C 5.12 SunOS_sparc
(emu) Sun Fortran 95 8.6 SunOS_sparc
Sun C++ 5.12 SunOS_sparc
Windows 7 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Visual Studio 2015 w/ Intel Fortran 16 (cmake)
Cygwin(CYGWIN_NT-6.1 2.2.1(0.289/5/3) gcc(4.9.3) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Visual Studio 2015 w/ Intel Fortran 16 (cmake)
Windows 8.1 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Windows 8.1 x64 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Mac OS X Mt. Lion 10.8.5 Apple clang/clang++ version 5.1 from Xcode 5.1
64-bit gfortran GNU Fortran (GCC) 4.8.2
(swallow/kite) Intel icc/icpc/ifort version 15.0.3
Mac OS X Mavericks 10.9.5 Apple clang/clang++ version 6.0 from Xcode 6.2.0
64-bit gfortran GNU Fortran (GCC) 4.9.2
(wren/quail) Intel icc/icpc/ifort version 15.0.3
Mac OS X Yosemite 10.10.5 Apple clang/clang++ version 6.0 from Xcode 7.0.0
64-bit gfortran GNU Fortran (GCC) 4.9.2
(osx1010dev/osx1010test) Intel icc/icpc/ifort version 15.0.3
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
SunOS 5.11 32-bit n y/y n y y y
SunOS 5.11 64-bit n y/y n y y y
Windows 7 y y/y n y y y
Windows 7 x64 y y/y n y y y
Windows 7 Cygwin n y/y n y y n
Windows 8.1 n y/y n y y y
Windows 8.1 x64 n y/y n y y y
Mac OS X Mountain Lion 10.8.5 64-bit n y/y n y y y
Mac OS X Mavericks 10.9.5 64-bit n y/y n y y y
Mac OS X Yosemeti 10.10.5 64-bit n y/y n y y y
AIX 6.1 32- and 64-bit n y/n n y y y
CentOS 6.7 Linux 2.6.32 x86_64 GNU y y/y y y y y
CentOS 6.7 Linux 2.6.32 x86_64 Intel n y/y n y y y
CentOS 6.7 Linux 2.6.32 x86_64 PGI n y/y n y y y
CentOS 7.1 Linux 3.10.0 x86_64 GNU y y/y y y y y
CentOS 7.1 Linux 3.10.0 x86_64 Intel n y/y n y y y
Linux 2.6.32-431.11.2.el6.ppc64 n y/n n y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
SunOS 5.11 32-bit y y y y
SunOS 5.11 64-bit y y y y
Windows 7 y y y y
Windows 7 x64 y y y y
Windows 7 Cygwin n n n y
Windows 8.1 y y y y
Windows 8.1 x64 y y y y
Mac OS X Mountain Lion 10.8.5 64-bit y n y y
Mac OS X Mavericks 10.9.5 64-bit y n y y
Mac OS X Yosemeti 10.10.5 64-bit y n y y
AIX 6.1 32- and 64-bit y n n y
CentOS 6.7 Linux 2.6.32 x86_64 GNU y y y y
CentOS 6.7 Linux 2.6.32 x86_64 Intel y y y y
CentOS 6.7 Linux 2.6.32 x86_64 PGI y y y y
CentOS 7.1 Linux 3.10.0 x86_64 GNU y y y y
CentOS 7.1 Linux 3.10.0 x86_64 Intel y y y y
Linux 2.6.32-431.11.2.el6.ppc64 y y y y
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
Linux 2.6.18-431.11.2.el6 g95 (GCC 4.0.3 (g95 0.94!)
#1 SMP x86_64 GNU/Linux
(platypus)
Mac OS X El Capitan 10.11 Apple clang/clang++ version 7.0.0 from Xcode 7.0.1
64-bit gfortran GNU Fortran (GCC) 5.2.0
(VM)
Windows 7 Visual Studio 2008 (cmake)
Windows 7 x64 Visual Studio 2008 (cmake)
Windows 10 Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Windows 10 x64 Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Debian7.5.0 3.2.0-4-amd64 #1 SMP Debian 3.2.51-1 x86_64 GNU/Linux
gcc (Debian 4.7.2-5) 4.7.2
GNU Fortran (Debian 4.7.2-5) 4.7.2
(cmake and autotools)
Fedora20 3.15.3-200.fc20.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.8.3 20140624 (Red Hat 4.8.3-1)
GNU Fortran (GCC) 4.8.3 20140624 (Red Hat 4.8.3-1)
(cmake and autotools)
SUSE 13.1 3.11.10-17-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.8.1
GNU Fortran (SUSE Linux) 4.8.1
(cmake and autotools)
Ubuntu 14.04 3.13.0-35-generic #62-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.9.1-0ubuntu1) 4.9.1
GNU Fortran (Ubuntu/Linaro 4.9.1-0ubuntu1) 4.9.1
(cmake and autotools)
hopper.nersc.gov PrgEnv-gnu/5.2.40
gcc (GCC) 4.9.2 20141030 (Cray Inc.)
GNU Fortran (GCC) 4.9.2 20141030 (Cray Inc.)
g++ (GCC) 4.9.2 20141030 (Cray Inc.)
Known Problems
==============
* On Windows platforms in debug configurations, the VFD flush1 tests will fail
with the split and multi VFD drivers. These tests will display a modal debug
dialog which must be answered or wait for the test timeout to expire.
(ADB - 2014/06/23 - HDFFV-8851)
* CLANG compiler with the options -fcatch-undefined-behavior and -ftrapv
catches some undefined behavior in the alignment algorithm of the macro DETECT_I
in H5detect.c (Issue 8147). Since the algorithm is trying to detect the alignment
of integers, ideally the flag -fcatch-undefined-behavior shouldn't to be used for
H5detect.c. In the future, we can separate flags for H5detect.c from the rest of
the library. (SLU - 2013/10/16)
* Make provided by Solaris fails in "make check". Solaris users should use
gmake to build and install the HDF5 software. (AKC - 2013/10/08 - HDFFV-8534)
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the Fortran and HL/Fortran
tests fail. HL and C++ shared libraries should now be working as intended,
however.
(MAM - 2011/04/20)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* All of the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian systems. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.15%%%%
HDF5 version 1.8.15 released on 2015-05-04
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.14 and
HDF5-1.8.15, and contains information on the platforms tested and
known problems in HDF5-1.8.15.
Links to the HDF5 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for HDF5 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document at this location:
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.14
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration
-------------
- CMake
Improvements made to the CMake build system.
The default options were changed to align with the Autotools configure
defaults. CMake configure files now support components when packaged
with CPack. Windows CPack supports WiX packaging, and will look for
WiX and NSIS in the standard locations.
The CMake minimum has been changed to 3.1.
(ADB - 2015/04/01 HDFFV-8074, 8968, 9006)
- cmakehdf5 for Cmake building.
Added configure options to support the building of Fortran or CXX API,
to enable/disable testings. Use "cmakehdf5 --help" for details.
(AKC - 2014/12/09 HDFFV-8932)
- Building Shared and Parallel Made Explicit
When --enable-parallel is specified, configure used to disable
shared by default.
Removed the restriction for building shared when parallel is
enabled. --disable-shared has to be used explicitly if user
wishes to disable shared libraries.
(MSC - 2015/02/19 HDFFV-9069)
- Inferring Parallel Compilers
configure was trying to infer if a compiler is a parallel
compiler with MPI support and enable parallel even if the user
did not explicitly enable parallel. This should not happen.
Disabled inferring parallel compilers to enable parallel HDF5
build. --enable-parallel has to be used explicitly to build
parallel HDF5 regardless of the compiler type being used.
(MSC - 2015/02/19 HDFFV-9068)
- Large File Support Configuration Option
Removed the option to enable or disable large file support. It will
always be enabled.
(MSC - 2015/02/19 HDFFV-9097)
- Removed Configuration Feature
When configure detected that the CodeWarrior compiler was being used it
would define a symbol that caused a test in test/tfile.c to be skipped
due to a broken CodeWarrior open() command.
Since this only masks the problem instead of fixing it and we don't
support CodeWarrior anyway, this functionality was removed.
(DER - 2015/02/21, HDFFV-9080)
- VMS Build/Test Files Have Been Removed
HDF5 no longer supports VMS, and the files were getting out of date.
Since we have no access to a VMS machine, there is no way for us to
maintain them.
A Subversion tag was created at:
https://svn.hdfgroup.uiuc.edu/tags/vms_last_support_1_8
immediately before removing the files.
(DER - 2015-02-26, HDFFV-9147)
- Removal of --with-default-vfd configure Option
In theory, this option was intended to allow setting a default
VFD that would be used by the library. In practice, the feature
only accepted the POSIX (SEC2) VFD (already the default) and
the stdio VFD (a demo VFD not intended for production use). The
inability to pass key VFD parameters at configure time limits the
full implementation of this feature, so it was retired.
(DER - 2015-02-26, HDFFV-9081)
- Direct VFD configure Behavior
The configure options for Linux now allow the Direct VFD to build
without passing additional compiler options/defines like _GNU_SOURCE.
Passing --enable-direct-vfd is now all that is needed to enable
the feature.
The Direct VFD is now disabled by default since it is intended for
specialized audiences. It was previously enabled by default, but the
configure script did not set correct POSIX levels, etc. making this
a moot point.
Note that the Direct VFD can only be configured on Linux when
the O_DIRECT flag to open()/create() and posix_memalign() function
are available. This is unchanged from previous behavior.
(DER - 2015-02-26, HDFFV-9057, 7567, 9088, 7566)
- _POSIX_C_SOURCE, _GNU_SOURCE, and _BSD_SOURCE No Longer Exported
to h5cc and Other Compiler Wrappers
The _POSIX_C_SOURCE, _GNU_SOURCE, and _BSD_SOURCE definitions are
not required for using API functions and may conflict with user
code requirements.
(DER - 2015-03-08, HDFFV-9152)
- Removed the --enable-filters Option from configure
This option allowed the user to disable selected internal filters,
presumably to make the library smaller. It has been removed since
it saved little space (the internal filters are small with respect
to the overall library size) and was not generally extendible to
the library at large due to the large number of #ifdefs that would
be required.
Note that this features applied to internal filters such as shuffle
and n-bit and not external filters like gzip or Szip. Those are still
enabled or disabled via their own configure options.
(DER - 2015-03-08, HDFFV-9086)
- Removed Obsolete Time Functionality from configure and the C Library
The library contained some residual functionality from obsolete
time zone handling code. This has been removed, and the configure
checks for the time functions have been cleaned up.
* Lumped all the time functionality together in configure.ac.
This was previously more spread out due to Solaris issues
with the ordering of certain checks.
* Removed processing that handles __tm_gmtoff members of struct
tm. (libc-4)
* Removed BSDgettimeofday(). (IRIX 5.3)
* Removed timezone struct handling in gettimeofday() (considered
harmful).
Note that the HDF5 Library stores timestamps in a platform-independent
manner, so old files can still be read. This only affects converting
system time to HDF5 timestamps.
The library currently uses the tm_gmtoff member of the tm struct
(preferred, if available) or the timezone global variable to
construct HDF5 timestamps.
(DER - 2015-03-09, HDFFV-9083 and 9085)
- Added -D_DEFAULT_SOURCE to CPPFLAGS on Linux Systems
This is the replacement for -D_BSD_SOURCE in versions of glibc since 2.19.
Since both are defined, it should work for all versions of glibc. Defining
both suppresses the warning about defining _BSD_SOURCE.
(NAF - 2015-04-02, HDFFV-9079)
Library
-------
- Added Memory Allocation Functions that Use the Library's Allocator
HDF5 filters may need to allocate or resize the buffer that is passed
to them from the library. If the filter has been compiled separately
from the library, it and the library may use different memory
allocation libraries for the (re)allocation and free calls. This can
cause heap corruption and crashes. This is particularly a problem on
Windows since each C run-time library is implemented as a separate
shared library, but can also show up on POSIX systems when debug or
high-performance allocation libraries are in use.
Two new functions (H5allocate_memory() and H5resize_memory()) were
added to the HDF5 C library. These functions have the same semantics as
malloc/calloc and realloc, respectively. Their primary purpose is to
allow filter authors to allocate or resize memory using the same
memory allocation library as the HDF5 library. Filter authors are
highly encouraged to use these new functions in place of malloc,
calloc, and realloc. They should also use the H5free_memory() call when
freeing memory.
Note that the filters provided with the library (zlib, szip, etc.) do
not experience the problems that these new functions are intended to
fix. This work only applies to third-party filters that are compiled
separately from the library.
(DER - 2015-04-01, HDFFV-9100)
- H5Pset_istore_k and H5Pset_sym_k
These two functions didn't check the value of the input parameter "ik".
When 2*ik exceeded 2 bytes of storage, data was lost in the file;
for example, some chunks would be overwritten.
Added validation of "ik" to not exceed the max v1 btree entries (2 bytes)
to these two routines.
(VC - 2015-03-24, HDFFV-9173)
- Added Functions to Control the Value of H5PL_no_plugin_g without
Using an Environment Variable
Sometimes it is necessary for an application to disable the use of
dynamically loaded plugin libraries without requiring the library to
be built with plugin support disabled or to set an environment
variable to disable plugin support globally.
Two new functions (H5PLset_loading_state() and H5PLget_loading_state())
were added to the HDF5 C Library. These functions require a parameter
that indicates which type of dynamically loaded plugin is enabled or
disabled.
(ADB - 2015-03-17, HDFFV-8520)
Parallel Library
----------------
- MPI_Finalize and HDF5 Library Shutdown
Calling HDF5 routines after MPI_Finalize has been closed should
not be done, since those routines might call MPI functions that
would not be possible to do after finalizing the MPI library.
Attached an attribute destroy callback to MPI_COMM_SELF that
shuts down the HDF5 library when MPI_COMM_SELF is destroyed,
in other words, on MPI_Finalize. This should fix several issues
that users see when they forget to close HDF5 objects before
calling MPI_Finalize().
(MSC - 2015/02/25, HDFFV-883)
Tools
-----
- None
High-Level APIs
---------------
- None
Fortran API
-----------
- Added Global Variables
These new global variables are equivalent to the C definitions
without the '_F':
H5G_UDLINK_F
H5G_SAME_LOC_F
H5O_TYPE_UNKNOWN_F
H5O_TYPE_GROUP_F
H5O_TYPE_DATASET_F
H5O_NAMED_DATATYPE_F
H5O_TYPE_NTYPES_F
(MSB - 2015/02/03, HDFFV-9040)
C++ API
-------
- New Wrappers for C Functions H5P[s/g]et_libver_bounds
Wrappers were added to class H5::FileAccPropList for the
C Functions H5Pget_libver_bounds and H5Pset_libver_bounds.
(BMR, 2015/04/06, Part of HDFFV-9167)
- New Wrappers to Get the Object Header's Version
The following wrappers are added to class H5::CommonFG
Returns the object header version of an object in a file or group,
given the object's name.
unsigned childObjVersion(const char* objname) const;
unsigned childObjVersion(const H5std_string& objname) const;
(BMR, 2015/04/06)
- New DataType Constructor
Added a DataType constructor that takes a PredType object, and this
constructor will invoke H5Tcopy to generate another datatype id
from a predefined datatype.
(BMR, 2015/04/06)
Support for New Platforms, Languages, and Compilers
===================================================
- Support for Linux 3.10.0-123.20.1.el7 added (LK - 2015/04/01)
- Support for Mac OS X Yosemite 10.10 added (AKC - 2015/03/04, HDFFV-9007)
- Support for AIX 6.1 added and AIX 5.3 is retired. (AKC - 2015/01/09)
Bug Fixes since HDF5-1.8.14
===========================
Configuration
-------------
- Make uninstall generated "test: argument expected".
The error is due to $EXAMPLETOPDIR is used without setting a value first.
Fixed by assign it with the proper value.
(AKC - 2015/04/29, HDFFV-9298)
- Windows Installer Incorrect Display of PATH Environment Variable
In the Windows installer, the dialog box where the user can elect to
add the product's bin path to the %PATH% environment variable displayed
an incorrect path. This path was missing the C:\Program Files part
and used the POSIX file separator '/' before the bin (<path>/bin,
instead of <path>\bin).
The dialog box text was changed to simply say that the product's bin
path would be added instead of explicitly displaying the path.
This is in line with most installers. The reason for not fixing the
displayed path instead is that it is difficult to pass the correct
path from CPack to the NSIS installer for display.
Note that this was never a code issue - it was just a display
problem. The installer always did the right thing when updating the
environment variable.
(DER - 2014/11/14, HDFFV-9016)
Library
-------
- Incorrect Usage of List in CMake COMPILE_DEFINITIONS set_property
The CMake command set_property with COMPILE_DEFINITIONS property
needs a quoted semi-colon separated list of values. CMake will
transform the list to a series of -D{value} for the compile.
(ADB - 2014/12/09, HDFV-9041)
- Fixed Compile Errors on Windows w/ Visual Studio and CMake When
UNICODE is Defined
The HDF5 Library could not be built on Windows with Visual Studio when
UNICODE was defined. This was due to the incorrect use of the TEXT()
macro and some Win32 API functions that take TCHAR parameters. The faulty
code was a part of the filter plugin functionality. This was a
compile-time error that only affected users who build HDF5 from source
and define UNICODE, usually when HDF5 is being built as a part of a
larger product. There were no run-time effects.
These errors caused no problems when UNICODE was not defined. HDF5 is
normally not built with UNICODE defined and the binaries were
unaffected.
The fix was to remove the TEXT() macro and explicitly use the
'A' form of the Win32 API calls, which expect char strings instead of
wchar_t strings.
Note that HDF5 currently does not support Unicode file paths on Windows.
(DER - 2015/02/22, HDFFV-8927)
- Addition of Error Tracing Functionality to Several C API Calls
A bug in a text processing script caused API calls that return a
pointer to not receive error tracing macros/functionality.
The bug has been corrected and error tracing functionality has been
added to the affected API calls. These functions will now correctly
print trace information when library errors are encountered.
(DER - 2015/02/26, HDFFV-9141)
- H5Rdereference Now Checks for HADDR_UNDEF or Uninitialized References
When passed HADDR_UNDEF or uninitialized references, the previous
behavior of H5Rdereference was to continue to process the reference
as a valid address.
H5Rdereference was changed to return immediately (with an error
message) if the references are HADDR_UNDEF or uninitialized.
(MSB - 2015/3/10, HDFFV-7959)
- Fixed Bugs in H5Sextent_copy
H5Sextent_copy would not free the previous extent, resulting in a memory
leak. Also, H5Sextent_copy would not update the number of elements
selected if the selection was "all", causing various problems. These
issues have been fixed.
(NAF - 2015/04/02)
Parallel Library
----------------
- Fixed a Potential Memory Error
Fixed a potential memory error when performing parallel I/O on a
dataset with a single chunk, and at least one process has nothing
to do.
(NAF - 2015/02/16)
- Parallel Test Problem Fixed
Fixed problem with parallel tests where they failed beyond a
certain number of ranks. All tests should work for any arbitrary
number of ranks.
(MSC - 2014/11/06, HDFFV-1027,8962,8963)
- MPE Support
Enabling MPE was causing HDF5 build to fail. Support for it was
dropped at some point in time.
Fixed problem with enabling MPE. Users should use the community
maintained MPE on github (http://git.mpich.org/mpe.git/).
(MSC - 2015/02/20, HDFFV-9135)
Performance
-------------
- None
Tools
-----
- h5repack crashed on enumerated 8-bit type.
Previous version 1.8.14 introduced an error that caused the reading
of enumerated 8-bit type nested in compound type to fail.
Fixed library code responsible for reading the particular type.
(AKC - 2015.03/31, HDFFV-8667)
- h52gif crashed non-8bit images.
h52gif crashed if instructed to convert images other than 8bit images.
h52gif could handle only 8bit images. Added code to detect non-8bit
images and flag them as failure. Update tool document page to reflect
the limit.
(AKC - 2015/03/31, HDFFV-8957)
- perform/benchpar.c retired.
benchpar.c has not been built for a long time and its original purpose
is not needed any more.
(AKC - 2014/12/19, HDFFV-8156)
- Source perform/ directory moved to tools/perform.
The perform directory is moved to tools/perform for easier maintenance.
(AKC - 2014/12/17, HDFFV-9046)
Fortran API
------------
- Fortran Fails with --enable-fortran2003 and Intel 15.x Compilers
Added BIND(C) to the offending APIs.
The Fortran Library (--enable-fortran2003) now works using Intel 15.x
without the need for any additional compilers flags.
(MSB - 2015/1/26, HDFFV-9049)
- h5tenum_insert_f Does Not Work with Default 8 Byte Integers
(xlf compiler)
In the Fortran 90 API, 'value' is no longer cast into the C int type.
Therefore, if h5tenum_insert_f is passed an 8 byte integer (via -i8)
then 'value' is written as the same type as the default Fortran
integer type (which can be 8 bytes).
A new Fortran 2003 API was added which is more in line with the C
API and users are strongly encouraged to use the Fortran 2003 API
instead of the Fortran 90 API.
SUBROUTINE h5tenum_insert_f(type_id, name, value, hdferr)
INTEGER(HID_T) , INTENT(IN) :: type_id
CHARACTER(LEN=*), INTENT(IN) :: name
TYPE(C_PTR) , INTENT(IN) :: value
INTEGER, INTENT(OUT) :: hdferr
(MSB - 2015/2/19, HDFFV-8908)
- Some Fortran APIs Never Returned the Error State
Some Fortran APIs never returned the error state: they
would always return a positive number. The APIs include
the following:
h5fget_file_image_f
h5lget_name_by_idx_f
h5oget_comment_by_name_f
They were corrected to return a negative number as described in
the Reference Manual if an error occurred.
(MSB - 2015/3/19, HDF5-239)
- Fixed h5pget_class_f
h5pget_class_f never correlated the class identifier to the property
list class name as indicated in the HDF5 Reference Manual; it instead
returned a property list class identifier as an INTEGER. The INTEGER
needed to be of type INTEGER(HID_T) to be correct.
The h5pget_class_f API was changed to return an INTEGER(HID_T)
property list class identifier instead of an INTEGER. This mimics the
intended behavior of the C API.
(MSB - 2015/3/16, HDFFV5-9162)
C++ API
------
- Combined Two H5File::getObjCount Overloaded Methods
The following two methods
ssize_t getObjCount(unsigned types) const;
ssize_t getObjCount() const;
were combined into one:
ssize_t getObjCount(unsigned types = H5F_OBJ_ALL) const;
(BMR - 2015/04/06)
- Many Warnings Were Removed
Many warnings such as conversion, unused variables, missing base
class initialization, and initializing base classes in wrong order
were removed.
(BMR, 2015/04/06)
- Functionality Deprecation
The following two constructors of classes AbstractDs, IdComponent,
H5Location, and H5Object are no longer appropriate after the data member
"id" had been moved from IdComponent to the sub-classes in previous
releases.
<Classname>(const hid_t h5_id);
<Classname>(const <Classname>& original);
The copy constructors were no-op and removed in 1.8.15. The other
constructors will be removed from 1.10 release, and then from 1.8.17
if their removal does not cause any problems.
(BMR, 2015-04-06)
High-Level APIs:
------
- Suppress Warnings from Flex/Bison-generated Code
Warning suppression #pragmas, etc. have been added to H5LTparse.c and
H5LTanalyze.c. We have no control over this code since it's created by
a generator.
(DER - 2015/03/08 - HDFFV-9149)
- Changed hdf5_hl.h to Include the HDF5 Main Library "hdf5.h"
User's no longer need to include both hdf5_hl.h and hdf5.h
(MSB - 2015/2/14, HDFFV-8685)
- H5PTcreate_fl Does Not Convert to Memory Datatype
H5PTcreate_fl now converts to the table's native memory datatype
to fix the problem of handling BE and LE packet tables.
(MSB - 2015/2/26 - HDFFV-9042)
- Fix for H5LT Attribute Functions
H5LT attribute functions fail to create attributes whose name
is a substring of an existing attribute.
H5LT attribute functions can now create attributes whose name
is a substring of an existing attribute.
(MSB - 2015/2/24, HDFFV-9132)
Fortran High-Level APIs:
------------------------
- Internal Library Fix for Missing Argument Declaration
In Interface block for h5tbmake_table_c, "max_char_size_field_names"
is listed as an input, but in the argument definitions it is
"INTEGER :: max_char_size". This caused no known problems with the
Fortran HL API.
Fixed missing argument definition.
(MSB - 2015/2/18, HDFFV-8559)
Testing
-------
- None
Supported Platforms
===================
The following platforms are supported and have been tested for this release.
They are built with the configure process unless specified otherwise.
AIX 6.1 xlc/xlc_r 10.1.0.5
(NASA G-ADA) xlC/xlC_r 10.1.0.5
xlf90/xlf90_r 12.1.0.6
Linux 2.6.18-308.13.1.el5PAE GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP i686 i686 i386 compilers for 32-bit applications;
(jam) Version 4.1.2 20080704 (Red Hat 4.1.2-55)
Version 4.8.4, 4.9.2
PGI C, Fortran, C++ Compilers for 32-bit
applications;
Version 14.10-0
Intel(R) C, C++, Fortran Compiler for 32-bit
applications;
Version 15.0.1.133 (Build 20141023)
Linux 2.6.18-371.6.1.el5 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers for 64-bit applications;
(koala) Version 4.1.2 20080704 (Red Hat 4.1.2-55)
Version 4.8.4, 4.9.2
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64;
Version 15.0.1.133 Build 20141023
Linux 2.6.32-431.11.2.el6 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(platypus) Version 4.4.7 20120313
Version 4.8.2, Version 4.9.2
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 14.10-0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 15.0.1.133 Build 20141023
Linux 3.10.0-123.20.1.el7 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(moohan) Version 4.8.2 20140120 (Red Hat 4.8.2-16)
Intel(R) C Intel(R) 64 Compiler XE for
applications running on Intel(R) 64,
Version 15.0.1.133 Build 20141023
Linux 2.6.32-431.29.2.el6.ppc64 gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-4)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-4)
(ostrich) GNU Fortran (GCC) 4.4.7 20120313 (Red Hat 4.4.7-4)
IBM XL C/C++ V13.1
IBM XL Fortran V15.1
SunOS 5.11 32- and 64-bit Sun C 5.12 SunOS_sparc
(emu) Sun Fortran 95 8.6 SunOS_sparc
Sun C++ 5.12 SunOS_sparc
Windows 7 Visual Studio 2008 (cmake)
Visual Studio 2010 w/ Intel Fortran 14 (cmake)
Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.34(0.285/5/3) gcc(4.9.2) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2008 (cmake)
Visual Studio 2010 w/ Intel Fortran 14 (cmake)
Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Windows 8.1 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Windows 8.1 x64 Visual Studio 2012 w/ Intel Fortran 15 (cmake)
Visual Studio 2013 w/ Intel Fortran 15 (cmake)
Mac OS X Mt. Lion 10.8.5 Apple clang/clang++ version 5.1 from Xcode 5.1
64-bit gfortran GNU Fortran (GCC) 4.8.2
(swallow/kite) Intel icc/icpc/ifort version 14.0.2
Mac OS X Mavericks 10.9.5 Apple clang/clang++ version 6.0 from Xcode 6.1.1
64-bit gfortran GNU Fortran (GCC) 4.8.2
(wren/quail) Intel icc/icpc/ifort version 14.0.2
Mac OS X Yosemite 10.10.2 Apple clang/clang++ version 6.0 from Xcode 6.1.1
64-bit gfortran GNU Fortran (GCC) 4.9.2
(osx1010dev/osx1010test) Intel icc/icpc/ifort version 15.0.1
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
SunOS 5.11 32-bit n y/y n y y y
SunOS 5.11 64-bit n y/y n y y y
Windows 7 y y/y n y y y
Windows 7 x64 y y/y n y y y
Windows 7 Cygwin n y/y n y y n
Windows 8.1 n y/y n y y y
Windows 8.1 x64 n y/y n y y y
Mac OS X Mountain Lion 10.8.5 64-bit n y/y n y y y
Mac OS X Mavericks 10.9.5 64-bit n y/y n y y y
Mac OS X Yosemeti 10.10.2 64-bit n y/y n y y y
AIX 6.1 32- and 64-bit n y/n n y y y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y/y y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel n y/y n y y y
CentOS 5.9 Linux 2.6.18-308 i686 PGI n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 GNU n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 GNU y y/y y y y y
CentOS 6.4 Linux 2.6.32 x86_64 Intel n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 PGI n y/y n y y y
CentOS 7.0 Linux 3.10.0 x86_64 GNU y y/y y y y y
CentOS 7.0 Linux 3.10.0 x86_64 Intel n y/y n y y y
Linux 2.6.32-431.11.2.el6.ppc64 n y/n n y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
SunOS 5.11 32-bit y y y y
SunOS 5.11 64-bit y y y y
Windows 7 y y y y
Windows 7 x64 y y y y
Windows 7 Cygwin n n n y
Windows 8.1 y y y y
Windows 8.1 x64 y y y y
Mac OS X Mountain Lion 10.8.5 64-bit y n y y
Mac OS X Mavericks 10.9.5 64-bit y n y y
Mac OS X Yosemeti 10.10.2 64-bit y n y y
AIX 6.1 32- and 64-bit y n n y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel y y y n
CentOS 5.9 Linux 2.6.18-308 i686 PGI y y y n
CentOS 5.9 Linux 2.6.18 x86_64 GNU y y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel y y y n
CentOS 6.4 Linux 2.6.32 x86_64 GNU y y y n
CentOS 6.4 Linux 2.6.32 x86_64 Intel y y y n
CentOS 6.4 Linux 2.6.32 x86_64 PGI y y y n
CentOS 7.0 Linux 3.10.0 x86_64 GNU y y y n
CentOS 7.0 Linux 3.10.0 x86_64 Intel y y y n
Linux 2.6.32-431.11.2.el6.ppc64 y y y n
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
Linux 2.6.18-308.13.1.el5PAE MPICH mpich 3.1.3 compiled with
#1 SMP i686 i686 i386 gcc 4.9.2 and gfortran 4.9.2
(jam) g95 (GCC 4.0.3 (g95 0.94!)
Linux 2.6.18-431.11.2.el6 MPICH mpich 3.1.3 compiled with
#1 SMP x86_64 GNU/Linux gcc 4.9.2 and gfortran 4.9.2
(platypus) g95 (GCC 4.0.3 (g95 0.94!)
FreeBSD 8.2-STABLE i386 gcc 4.5.4 [FreeBSD] 20110526
(loyalty) gcc 4.6.1 20110527
g++ 4.6.1 20110527
gfortran 4.6.1 20110527
FreeBSD 8.2-STABLE amd64 gcc 4.5.4 [FreeBSD] 20110526
(freedom) gcc 4.6.1 20110527
g++ 4.6.1 20110527
gfortran 4.6.1 20110527
Debian7.5.0 3.2.0-4-amd64 #1 SMP Debian 3.2.51-1 x86_64 GNU/Linux
gcc (Debian 4.7.2-5) 4.7.2
GNU Fortran (Debian 4.7.2-5) 4.7.2
(cmake and autotools)
Fedora20 3.15.3-200.fc20.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.8.3 20140624 (Red Hat 4.8.3-1)
GNU Fortran (GCC) 4.8.3 20140624 (Red Hat 4.8.3-1)
(cmake and autotools)
SUSE 13.1 3.11.10-17-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.8.1
GNU Fortran (SUSE Linux) 4.8.1
(cmake and autotools)
Ubuntu 14.04 3.13.0-35-generic #62-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.9.1-0ubuntu1) 4.9.1
GNU Fortran (Ubuntu/Linaro 4.9.1-0ubuntu1) 4.9.1
(cmake and autotools)
hopper.nersc.gov PrgEnv-gnu/5.2.40
gcc (GCC) 4.9.2 20141030 (Cray Inc.)
GNU Fortran (GCC) 4.9.2 20141030 (Cray Inc.)
g++ (GCC) 4.9.2 20141030 (Cray Inc.)
Known Problems
==============
* On Windows platforms in debug configurations, the VFD flush1 tests will fail
with the split and multi VFD drivers. These tests will display a modal debug
dialog which must be answered or wait for the test timeout to expire.
(ADB - 2014/06/23 - HDFFV-8851)
* CLANG compiler with the options -fcatch-undefined-behavior and -ftrapv
catches some undefined behavior in the alignment algorithm of the macro DETECT_I
in H5detect.c (Issue 8147). Since the algorithm is trying to detect the alignment
of integers, ideally the flag -fcatch-undefined-behavior shouldn't to be used for
H5detect.c. In the future, we can separate flags for H5detect.c from the rest of
the library. (SLU - 2013/10/16)
* Make provided by Solaris fails in "make check". Solaris users should use
gmake to build and install the HDF5 software. (AKC - 2013/10/08 - HDFFV-8534)
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. the issue. HL and C++ shared libraries should now be
working as intended, however.
(MAM - 2011/04/20)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.14%%%%
HDF5 version 1.8.14 released on 2014-11-12
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.13 and
HDF5-1.8.14, and contains information on the platforms tested and
known problems in HDF5-1.8.14.
All new and modified APIs are listed in the "HDF5 Software Changes
from Release to Release" document along with details about previous
releases at:
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
Links to the HDF5 1.8.14 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.14 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.14 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.13
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration
-------------
- bin/cmakehdf5 configures, builds and installs C, C++, Fortran and High
level API's. (It used to build the C API only).
(AKC 2014/10/17 HDFFV-8932).
Library
-------
- None
Parallel Library
----------------
- Chunk Fill Writes Changed to Collective
Slow performance in chunk fill writes. Chunk fills
in the past were written independently by rank 0 one block
at a time.
Optimized the chunk fill write algorithm so that all
chunk fill values will be written collectively in a single MPI-IO
call. This should show a great performance improvement when
creating chunked datasets in parallel when the chunk dimensions
are fairly small.
(MSC - 2014/08/22, HDFFV-8878)
Tools
-----
- None
High-level APIs
---------------
- None
Fortran API
-----------
- None
C++ API
-------
- Initialization of Object IDs
The data member "id" in classes that represent HDF5 objects were
initialized to 0, which caused problem for some users.
Replaced 0 with H5I_INVALID_HID to initialize these "id"s. For the
PropList class, H5P_DEFAULT is used instead of H5I_INVALID_HID.
(BMR - 2014/09/30, HDFFV-4259)
Support for New Platforms, Languages, and Compilers
===================================================
- None
Bug Fixes since HDF5-1.8.13
===========================
Configuration
-------------
- CMake and SVN URLs
The SVN URLs will be different when the HDF Group domain name changes.
Removed the SVN URL references in the cacheinit.cmake and release_docs files.
(ADB - 2014/10/27, HDFFV-8953)
- CMake Packaging
A Fortran module was not generated if the compiler was not F2003
compliant.
Removed the module name from the package list of Fortran modules because
that module was never generated. This was only an issue for Fortran
compliers that are not F2003 compatible.
(ADB - 2014/10/16, HDFFV-8932)
- Shared Library Interface Version Number (soname)
In order to increase the maintainability of HDF5, an architectural
change was made which required the renaming of several public symbols in
H5Ppublic.h.
The shared libary interface version number ("soname") has been increased
on account of these changes. For a full list of the changed symbols, see
the interface compatibility report, which is available as a link off of
the 'HDF5 Software Changes from Release to Release' document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
(AKC - 2014/10/03, HDFFV-8937)
- Configure Settings for Mac OSX Need Defaults for PROD_XXX, DEBUG_XXX,
and PROFILE_XXX
The configure setting files for Mac OSX (config/apple) did not
have the default settings of PROD_XXX, DEBUG_XXX, PROFILE_XXX.
Added the default settings. Mac platforms now builds library with
"-O3" optimization when the default clang compiler is used.
(AKC - 2014/10/01, HDFFV-8933)
- CMake ConfigureChecks
Two include files were missing from two C tests.
Propagated the configure test changes to H5_LDOUBLE_TO_INTEGER_WORKS_TEST
and H5_ULLONG_TO_LDOUBLE_PRECISION_TEST to ConfigureChecks.cmake (added
stdlib.h and string.h in the HDFTests.c file).
(ADB - 2014/09/02 HDFFV-8845)
- CMake Parallel Test Missing
The source file was removed in the previous release but the parallel
test t_posix_compliant was not.
Removed the t_posix_compliant parallel test from the library.
(ADB - 2014/8/14 HDFFV-8880)
- Autotools Reconfigure. Bison. Flex.
The Bison and Flex files were out of date.
Bison was upgraded to 2.7, and Flex was upgraded to 2.5.37. The
bin/reconfigure script now will execute Bison and Flex and update
the hl/src files.
(ADB - 2014/06/16 HDFFV-8709)
- Autotools Reconfigure. m4.
The m4 macro processor was out of date.
Reconfigured Autotools with m4 upgraded to 1.4.17.
(ADB - 2014/06/12 HDFFV-8743)
- Autotools: Modified configure to add an entry at the beginning of AM_LDFLAGS
for the hdf5 install directory. Without this entry the relink commands
invoked by "make install" to create libraries dependent on libhdf5.so added
a dependency on the first libhdf5.so found in any directory in AM_LDFLAGS
regardless of its version. (LRK - 2014/10/17 HDFFV-8944)
- Changed Autotools Build Behavior. Fortran High-level Library.
The Fortran high-level (HL) library did not compile if the default
size of a REAL is DOUBLE PRECISION; the build would fail during
compilation.
Configure now checks to see if REAL is DOUBLE PRECISION, Fortran is
enabled, and HL library is enabled. If this is true, then configure
will stop with an error message.
(MSB - 2014/8/11, HDFFV-8883/HDFFV-889)
Library
-------
- Fixed Identifier Management Code
Opening an object returns an identifier; closing the object should
free up the identifier. A problem was found where the identifiers
were not being freed up correctly.
Fixed the problem so that identifiers that have been used can be
used again after their object has been closed.
(QAK - 2014/10/16, HDFFV-8930)
- Removal of DllMain() from Static Windows Builds
A DllMain() function was added in HDF5 1.8.13 in order to handle
win32 thread cleanup. The preprocessor #ifdefs around the DllMain
function allowed it to be compiled when the static library is built,
which is incorrect behavior that can cause linkage problems in
clients.
The fix was to change the preprocessor #ifdefs to exclude compiling
DllMain() in static builds. Our DllMain function is now only
compiled when the shared, thread-safe library is built on Windows.
(DER - 2014/06/13, HDFFV-8837)
- Enforce Constraint on page_size Parameter in H5Pset_core_write_tracking()
The reference manual states that the page_size parameter cannot be
zero.
This change checks the page_size parameter to see it is zero and
returns an error code if it is.
(DER - 2014/08/11, HDFFV-8891)
- H5Ldelete_by_idx() fails on non-existent group name.
(MSC - 2014/07/31, HDFFV-8888)
- H5Ldelete_by_idx() Seg Fault on Non-existent Group Name
If a non-existent group name was used by H5Ldelete_by_idx(), a
segmentation fault would result.
Bug was fixed.
(MSC - 2014/07/31, HDFFV-8888)
- Bug in Test When Building Parallel HDF5 on PVFS2
There was a bug in a test when building Parallel HDF5 on PVFS2.
The build now uses MPI_File_get_size() instead of stat().
(MSC - 2014/07/14, HDFFV-8856)
- MPI-IO Driver Tried to Allocate Space for Zero-length Dataset
MPI-IO driver tried to allocate space for zero-length dataset
and asserts.
Fixed driver and added a regression test.
(MSC - 2014/07/03, HDFFV-8761)
Parallel Library
----------------
- None
Performance
-------------
- None
Tools
-----
- None
Fortran API
-------
- SIZEOF Replaced by C_SIZEOF and STORAGE_SIZE.
The intrinsic function SIZEOF is non-standard and should be replaced with a
standard intrinsic function.
If the F2008 intrinsic C_SIZEOF and STORAGE_SIZE are available, then they will
be used instead of the non-standard SIZEOF intrinsic, even when the SIZEOF
function is available.
(MSB - 2014/6/16, HDFFV-8653)
- Non-functional API: h5pget_fill_value_f
The Fortran wrapper h5pget_fill_value_f was calling the wrong C API.
The correct C API, H5Pget_fill_value, is now called by the Fortran
wrapper.
(MSB - 2014/9/25, HDFFV-8879)
- Interoperability with C HDF5: H5Literate and h5literate_f
h5literate_f assumes the return value for the callback function to
be of type int (or int_f in C). However, in the C wrapper the return
value of H5Literate is type herr_t, and this could cause
interoperability issues.
The callback function should be declared INTEGER(C_INT) for
portability. The tests were updated accordingly.
(MSB - 2014/9/26, HDFFV-8909)
- Interoperability with C HDF5: Constant INTEGER Parameters with the
H5FD Interface
Wrong type cast of constant Fortran INTEGER parameters was used.
The following parameter constant types were changed from INTEGER to
INTEGER(HID_T) to match the C types: H5FD_CORE, H5FD_FAMILY, H5FD_LOG,
H5FD_MPIO, H5FD_MULTI, H5FD_SEC2, and H5FD_STDIO.
Other internal 'int' types where changed to 'hid_t'; these are
transparent to the user.
(MSB - 2014/7/18, HDFFV-8748)
C++ API
------
- Memory Leaks
There were several potential memory leaks in the library due to
dynamically allocated strings not being freed when failure occurs.
Applied user's patches to remove these potential memory leaks.
(BMR - 2014/09/30, HDFFV-8928)
- Disallow H5F_ACC_CREAT
H5F_ACC_CREAT was included in the C++ API but the C library does not
allow it at this time.
Removed this flag from the functions in H5File class.
(BMR - 2014/09/29, HDFFV-8852)
- Missing Flags in Documentation: H5F_ACC_RDONLY and H5F_ACC_RDWR
The H5F_ACC_RDONLY and H5F_ACC_RDWR flags were missing from the
documentation of the H5File constructors.
These two flags are now included in the documentation for opening
files.
(BMR - 2014/09/29, HDFFV-8852)
High-level APIs:
------
- Seg Faults in H5TBread_field_name and H5TBread_field_name_f
When H5TBread_field_name or H5TBread_field_name_f were used to read a
field and if the name of the field was wrong, a segmentation fault
would result.
Both C and Fortran APIs were fixed so they no longer seg fault if
the name of the field is wrong, and both APIs return a negative
value if the name of the field is wrong.
(MSB - 2014/09/29, HDFFV-8912)
- Possible Buffer Overflow in High-level (HL) APIs
Multiple HL APIs (H5DSis_scale is one example) had issues:
(1) The datatype from the file was re-used as the memory datatype,
and
(2) No effort was made to ensure that strings were actually
null-terminated.
All of the HL routines now check for NULL pointers, for null-terminated
strings, and to see if string buffers are short enough not to overflow
the buffer. The minimum length of the buffers is now used in strncmp
to avoid overflow.
(MSB - 2014/9/29, HDFFV-8670)
- Behavior Change of H5LTdtype_to_text
If a user buffer was passed in to H5LTdtype_to_text along with the
length, then the function would not truncate at the end of the
buffer, but would exceed the end of the user buffer.
H5LTdtype_to_text was changed to truncate the string if the user
buffer is too small.
(MSB - 2014/9/29, HDFFV-8855)
Fortran High-level APIs:
------
- See entry for HDFFV-8912 above.
Testing
-------
- A subtest in parallel h5diff (ph5diff) testing was bypassed for the
local Linux 32 machine due to unknown issue in the previous version of
Mpich. The failure no long exists in the current Mpich. Therefore the
bypass is removed. (AKC - 2014/11/03 HDFFV-8954)
- Fixed incorrect exit code values (was -1) in testframe which is commonly
used by several test programs. (AKC - 2014/07/22 HDFFV-8881)
- Fixed Incorrect Exit Code Values in Testframe
The testframe which is commonly used by several test programs
had some incorrect exit code values. Fixed the incorrect exit code
values. (AKC - 2014/07/22, HDFFV-8881)
Supported Platforms
===================
The following platforms are supported and have been tested for this release.
They are built with the configure process unless specified otherwise.
AIX 5.3 xlc 10.1.0.5
(NASA G-ADA) xlC 10.1.0.5
xlf90 12.1.0.6
Linux 2.6.18-308.13.1.el5PAE GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP i686 i686 i386 compilers for 32-bit applications;
(jam) Version 4.1.2 20080704 (Red Hat 4.1.2-54)
Version 4.8.2
PGI C, Fortran, C++ Compilers for 32-bit
applications;
Version 13.7-0
Intel(R) C, C++, Fortran Compiler for 32-bit
applications;
Version 14.0.2 (Build 20140120)
Linux 2.6.18-371.6.1.el5 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers for 64-bit applications;
(koala) Version 4.1.2 20080704 (Red Hat 4.1.2-54)
Version 4.8.2
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64;
Version 14.0.2 (Build 20140120)
Linux 2.6.32-431.11.2.el6 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(platypus) Version 4.4.7 20120313
Version 4.8.2
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 13.7-0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 14.0.2 (Build 20140120)
Linux 2.6.32-431.29.2.el6.ppc64 gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-4)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-4)
(ostrich) GNU Fortran (GCC) 4.4.7 20120313 (Red Hat 4.4.7-4)
IBM XL C/C++ V13.1
IBM XL Fortran V15.1
SunOS 5.11 32- and 64-bit Sun C 5.12 SunOS_sparc
(emu) Sun Fortran 95 8.6 SunOS_sparc
Sun C++ 5.12 SunOS_sparc
Windows 7 Visual Studio 2008 (cmake)
Visual Studio 2010 w/ Intel Fortran 14 (cmake)
Visual Studio 2012 w/ Intel Fortran 14 (cmake)
Visual Studio 2013 w/ Intel Fortran 14 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.32(0.274/5/3) gcc(4.8.3) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2008 (cmake)
Visual Studio 2010 w/ Intel Fortran 14 (cmake)
Visual Studio 2012 w/ Intel Fortran 14 (cmake)
Visual Studio 2013 w/ Intel Fortran 14 (cmake)
Windows 8.1 Visual Studio 2012 w/ Intel Fortran 14 (cmake)
Visual Studio 2013 w/ Intel Fortran 14 (cmake)
Windows 8.1 x64 Visual Studio 2012 w/ Intel Fortran 14 (cmake)
Visual Studio 2013 w/ Intel Fortran 14 (cmake)
Mac OS X Lion 10.7.5 Apple clang/clang++ version 3.0 from Xcode 4.6.1
64-bit gfortran GNU Fortran (GCC) 4.8.2
(duck) Intel icc/icpc/ifort version 13.0.3
Mac OS X Mt. Lion 10.8.5 Apple clang/clang++ version 5.1 from Xcode 5.1
64-bit gfortran GNU Fortran (GCC) 4.8.2
(swallow/kite) Intel icc/icpc/ifort version 14.0.2
Mac OS X Mavericks 10.9.5 Apple clang/clang++ version 6.0 from Xcode 6.0.1
64-bit gfortran GNU Fortran (GCC) 4.8.2
(wren) Intel icc/icpc/ifort version 14.0.2
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
Solaris2.11 32-bit n y/y n y y y
Solaris2.11 64-bit n y/y n y y y
Windows 7 y y/y n y y y
Windows 7 x64 y y/y n y y y
Windows 7 Cygwin n y/y n y y n
Windows 8.1 n y/y n y y y
Windows 8.1 x64 n y/y n y y y
Mac OS X Lion 10.7.5 64-bit n y/y n y y y
Mac OS X Mountain Lion 10.8.5 64-bit n y/y n y y y
Mac OS X Mavericks 10.9.5 64-bit n y/y n y y ?
AIX 5.3 32- and 64-bit n y/n n y y y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y/y y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel n y/y n y y y
CentOS 5.9 Linux 2.6.18-308 i686 PGI n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 GNU n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 GNU y y/y y y y y
CentOS 6.4 Linux 2.6.32 x86_64 Intel n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 PGI n y/y n y y y
Linux 2.6.32-431.11.2.el6.ppc64 n y/n n y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.11 32-bit y y y y
Solaris2.11 64-bit y y y y
Windows 7 y y y y
Windows 7 x64 y y y y
Windows 7 Cygwin n n n y
Windows 8.1 y y y y
Windows 8.1 x64 y y y y
Mac OS X Lion 10.7.5 64-bit y n y y
Mac OS X Mountain Lion 10.8.5 64-bit y n y y
Mac OS X Mavericks 10.9.5 64-bit y n y y
AIX 5.3 32- and 64-bit y n n y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel y y y n
CentOS 5.9 Linux 2.6.18-308 i686 PGI y y y n
CentOS 5.9 Linux 2.6.18 x86_64 GNU y y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel y y y n
CentOS 6.4 Linux 2.6.32 x86_64 GNU y y y n
CentOS 6.4 Linux 2.6.32 x86_64 Intel y y y n
CentOS 6.4 Linux 2.6.32 x86_64 PGI y y y n
Linux 2.6.32-431.11.2.el6.ppc64 y y y n
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
Linux 2.6.18-308.13.1.el5PAE MPICH mpich 3.1.2 compiled with
#1 SMP i686 i686 i386 gcc 4.9.1 and gfortran 4.9.1
(jam) g95 (GCC 4.0.3 (g95 0.94!)
Linux 2.6.18-431.11.2.el6 MPICH mpich 3.1.2 compiled with
#1 SMP x86_64 GNU/Linux gcc 4.9.1 and gfortran 4.9.1
(platypus) g95 (GCC 4.0.3 (g95 0.94!)
FreeBSD 8.2-STABLE i386 gcc 4.5.4 [FreeBSD] 20110526
(loyalty) gcc 4.6.1 20110527
g++ 4.6.1 20110527
gfortran 4.6.1 20110527
FreeBSD 8.2-STABLE amd64 gcc 4.5.4 [FreeBSD] 20110526
(freedom) gcc 4.6.1 20110527
g++ 4.6.1 20110527
gfortran 4.6.1 20110527
Debian7.5.0 3.2.0-4-amd64 #1 SMP Debian 3.2.51-1 x86_64 GNU/Linux
gcc (Debian 4.7.2-5) 4.7.2
GNU Fortran (Debian 4.7.2-5) 4.7.2
(cmake and autotools)
Fedora20 3.15.3-200.fc20.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.8.3 20140624 (Red Hat 4.8.3-1)
GNU Fortran (GCC) 4.8.3 20140624 (Red Hat 4.8.3-1)
(cmake and autotools)
SUSE 13.1 3.11.10-17-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.8.1
GNU Fortran (SUSE Linux) 4.8.1
(cmake and autotools)
Ubuntu 14.04 3.13.0-35-generic #62-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.9.1-0ubuntu1) 4.9.1
GNU Fortran (Ubuntu/Linaro 4.9.1-0ubuntu1) 4.9.1
(cmake and autotools)
Cray Linux Environment (CLE) PrgEnv-pgi/4.2.34
hopper.nersc.gov pgcc 13.6-0 64-bit target on x86-64 Linux -tp istanbul
pgf90 13.6-0 64-bit target on x86-64 Linux -tp istanbul
pgCC 13.6-0 64-bit target on x86-64 Linux -tp istanbul
Known Problems
==============
* On cygwin platforms the feature to load dynamic filter libraries only looks
for libraries with the a so extension. Support for cygwin cygxxx.dll libraries
is planned for the next release.
(ADB - 2014/11/04 - HDFFV-8736)
* On windows platforms in debug configurations, the VFD flush1 tests will fail
with the split and multi VFD drivers. These tests will display a modal debug
dialog which must be answered or wait for the test timeout to expire.
The flush1 and flush2 tests will be skipped under debug for this release.
(ADB - 2014/06/23 - HDFFV-8851)
* CLANG compiler with the options -fcatch-undefined-behavior and -ftrapv
catches some undefined behavior in the alignment algorithm of the macro DETECT_I
in H5detect.c (Issue 8147). Since the algorithm is trying to detect the alignment
of integers, ideally the flag -fcatch-undefined-behavior shouldn't to be used for
H5detect.c. In the future, we can separate flags for H5detect.c from the rest of
the library. (SLU - 2013/10/16)
* Make provided by Solaris fails in "make check". Solaris users should use
gmake to build and install the HDF5 software. (AKC - 2013/10/08 - HDFFV-8534)
* The h5dump and h5diff utilities occasionally produce different output
between Linux and Windows systems. This is caused by lower-level library
routines that fail to write fill values to the user's buffer when reading
unallocated chunks from datasets that have a fill value set to
H5D_FILL_VALUE_DEFAULT. Due to platform differences the return of
spurious data values has only been encountered on Windows 32-bit systems.
(Issue HDFFV-8247; JP - 2013/03/27)
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however.
(MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2.
(AKC - 2011/03/10)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO.
(CMC - 2009/04/28)
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.13%%%%
HDF5 version 1.8.13 released on 2014-05-05
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.12 and
HDF5-1.8.13, and contains information on the platforms tested and
known problems in HDF5-1.8.13.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.13 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.13 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.13 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.13 (current
release) versus Release 1.8.12":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.12
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration
-------------
- Autotools: Automake updated to 1.14.1 (ADB - 2014/04/08)
- CMake: Moved minimum CMake version to 2.8.11 which enables better library
include processing. (ADB - 2014/03/26)
- When configuring a thread-safe HDF5 Library it is no longer necessary
to specify --enable-threadsafe with --with-pthreads if the Pthreads
library is in a standard location. (DER - 2014/04/11 HDFFV-8693)
Library
-------
- Added an H5free_memory API call. This should be used to free memory
that has been allocated by HDF5 API calls. H5Tget_member_name and
H5Pget_class_name are two examples. The main motivation for this call
is Windows, where it is common for application code and the HDF5 Library
to be using different C run-time libraries (CRT). Using the new call
ensures that the same CRT handles both the allocation and free. This
new function can also be useful in any case where the library uses a
different memory manager than the application, such as when a debug
memory manager is in use or when the HDF5 Library is wrapped for use
in a managed language like Python or Java. Fixes HDFFV-7710, 8519,
and 8851. (DER - 2014/04/11)
- The Core VFD (aka Memory VFD) can now be configured to track dirty
regions in the file and only write out the changed regions on
flush/close. Additionally, a "page aggregation" size can be set that
will aggregate small writes into larger writes. For example, setting
a 1 MiB page aggregation size will logically partition the the
in-memory file into 1 MiB pages that will be written out in their
entirety if even a single byte is dirtied. The feature is controlled
via the new H5Pset/get_core_write_tracking() API call. A new
"core_paged" target has been added to the check-vfd target in
test/Makefile.am that exercises the feature over all HDF5 VFD-aware
tests. (DER - 2014/04/12)
Parallel Library
----------------
- Removed MPI-POSIX VFD, as it wasn't helping anyone and was just
generating support questions. Application developers performing
parallel I/O should always use the MPI-IO VFD.
(QAK - 2014/03/28 HDFFV-8659)
- Improved parallel I/O support to allow collective I/O on point
selections. (QAK - 2014/03/15)
Tools
-----
- None
High-Level APIs
---------------
- None
Fortran API
-----------
- Wrappers h5pset_file_image_f and h5pget_file_image_f were added to the
library. (MSB - 2014/1/2014)
C++ API
-------
- The following new features are added:
+ Wrappers to class H5Object to get an object's name (HDFFV-8548).
ssize_t getObjName(char *obj_name, size_t buf_size = 0)
ssize_t getObjName(H5std_string& obj_name, size_t len = 0)
H5std_string getObjName()
+ Wrappers to class H5CommonFG to get a child object's type from a
group or file (HDFFV-8367).
H5O_type_t childObjType(const H5std_string& objname)
H5O_type_t childObjType(const char* objname)
H5O_type_t childObjType(hsize_t index,
H5_index_t index_type=H5_INDEX_NAME,
H5_iter_order_t order=H5_ITER_INC, const char* objname=".")
+ Wrappers to class DSetMemXferPropList for setting/getting a transform
property list (HDFFV-7907).
DSetMemXferPropList(const char* expression);
void setDataTransform(const char* expression)
void setDataTransform(const H5std_string& expression)
ssize_t getDataTransform(char* exp, size_t buf_size=0)
H5std_string getDataTransform()
+ Wrapper to CompType for setting size to compound datatype (HDFFV-8642).
void setSize(size_t size)
+ Overloaded functions to provide prototypes that declare constant
arguments const (HDFFV-3384). These include:
DataSet::fillMemBuf
DataSet::getVlenBufSize
DataSpace::extentCopy
DataType::commit
FileAccPropList::setSplit
H5File::getVFDHandle
+ Additional overload to class H5Location to get a comment as a char*
ssize_t getComment(const char* name, size_t buf_size, char* comment)
+ Additional overloads to class Attribute to get an attribute's name for
convenience:
ssize_t getName(char* attr_name, size_t buf_size=0)
ssize_t getName(H5std_string& attr_name, size_t buf_size=0)
(BMR, 2014/04/15)
+ A static wrapper to Exception for printing the error stack without an
instance of Exception
static void printErrorStack(FILE* stream = stderr,
hid_t err_stack = H5E_DEFAULT);
(BMR, 2014/04/25)
Support for New Platforms, Languages, and Compilers
===================================================
Mac OS X 10.6 Snow Leopard is not supported by Apple any more. In view of
the added support of Mac OS X 10.9, Mac OS X 10.6 is retired from HDF5
supported platforms. (AKC - 2014/03/14 HDFFV-8704)
Mac OS X 10.9 Mavericks is supported. (AKC - 2014/03/04 HDFFV-8694)
Bug Fixes since HDF5-1.8.12
===========================
Configuration
-------------
- CMake: When CMake commands are executed individually on the command line
and the external filters are being built, the CMAKE_BUILD_TYPE define
must be set to the same value as the configuration
(-DCMAKE_BUILD_TYPE:STRING=Release if using -C Release). This is needed
by the the szip and zlib filter build commands. (ADB - HDFFV-8695)
- CMake: Removed use of the XLATE_UTILITY program.
(ADB - 2014/03/28 HDFFV-8640)
- CMake: Added missing quotes in setting the CMAKE_EXE_LINKER_FLAGS for the
MPI option. (ADB - 2014/02/27 HDFFV-8674)
- CMake: Configuration of the HDF5 C++ or Fortran libraries with the
thread-safety feature.
C++ and/or Fortran + thread-safe is enforced as a non-supported
configuration. This matches the autotools. (DER - 2014/04/11)
- CMake: Configuration of static HDF5 C library with the thread-safety
feature.
Static + thread-safe + Win32 threads is not a supported configuration
due to the inability to automatically clean up thread-local storage.
This is expected to be fixed in a future release. In the meantime, a
work-around that uses internal functionality may allow the combination
to be used without resource leaks. Contact the help desk for more
information. (DER - 2014/04/11)
- Autotools: Several changes were done to configure and installcheck.
An export of LD_LIBRARY_PATH=<szip library location> was
removed from configure; make installcheck was revised to run
scripts installed in share/hdf5_examples to use the installed h5cc, etc.
to compile and run example source files also installed there.
Make installcheck will now fail when a shared szip or other external lib
file cannot be found in the same manner that executables compiled and
linked with h5cc will fail to run when those lib files cannot be found
after install. Make installcheck should pass after setting
LD_LIBRARY_PATH to the szip location. (LRK - 2014/04/16)
Library
-------
- A Gnu Make directive (.NOTPARALLEL) is added to fortran/test/Makefile.
AIX native make does not support this directive and would fail if
parallel make (e.g. make -j4) is used to build the library. AIX users
either do not use parallel make or install Gnu Make to build the library.
(AKC 2014/04/08 HDFFV-8738)
- H5R.c: H5Rget_name gave an assertion failure if the "name" parameter
was NULL.
Fixed H5Rget_name to return the size of the buffer needed to read a
name of the referenced object in this case. The size doesn't include
the NULL terminator. H5Rget_name returns negative on failure.
(MSB - 2014/01/22 HDFFV-8620)
- H5Z.c: H5Zfilter_avail didn't check if a filter was available as a
dynamically loaded filter. The error manifested itself in the h5repack
tool when removing user-defined dynamically loaded filter.
Added a code to find the filter among the dynamically loaded filters
after the function fails to find it among the registered filters.
(ADB - 2014/03/03 HDFFV-8629)
- Memory leak: a memory leak was observed in conjunction to the
H5TS_errstk_key_g thread-local variable allocated in the H5E_get_stack
function in H5E.c.
The shared HDF5 thread-safe library now no longer leaks thread-local
storage resources on Windows with Win32 threads. Currently, there is
no solution for this problem when HDF5 is statically built. We
disabled the build of the static HDF5 thread-safe library with
Win32 threads. (DER - 2014/04/11 HDFFV-8518)
- H5Dio.c: Improved handling of NULL pointers to H5Dread/H5Dwrite
calls. Credit to Jason Newton (nevion@gmail.com) for the original patch.
H5Dwrite/read failed when a NULL pointer was passed for a data buffer
and 0 elements were selected. Fixed. (QAK - 2014/04/16 HDFFV-8705)
- Deprecated API (1_6 API): Improved handling of closing the library and
re-accessing it with a deprecated routine.
When a program used a deprecated API (for example, H5Gcreate1),
closed the library, and reopened it again to access a group, dataset,
datatype, dataspace, attribute, or property list, HDF5 failed to
provide an identifier for the object. Fixed.
(NAF, QAK - 2014/04/16 HDFFV-8232)
Parallel Library
----------------
- Fixed a missing H5F_Provisional module in HDF5mpio.f90
(MSB - 2014/2/7 HDFFV-8651)
Performance
-------------
- None
Tools
-----
- The h5diff tool would report that a datafile compared with an exact
copy of the same datafile had differences. This was due to the issue
below of reading un-written chunks. This problem is also fixed.
(AKC - 2014/05/01 HDFFV-8637)
- The h5dump and h5diff utilities occasionally produced different output
between Linux and Windows systems. This has been fixed.
This happened to datasets that used chunked storage, with default fill
values, and some of the chunks had not been written.
When the dataset was read, the library failed to write the default fill
values to parts of the use buffer that were associated with the unwritten
chunks. (JP - 2014/05/01 HDFFV-8247)
- The compress option is retired from bin/release.
(AKC - 2014/04/25 HDFFV-8755)
- bin/release has a new option "zip" that produces a release zip file for
the Windows platform. (AKC - 2014/04/24 HDFFV-8433)
- h5diff: Several failures relating to handling of strings attributes
are fixed.
The tool crashed or gave an error message when one of the strings had
fixed size type and another variable-length size type. h5diff now flags such
strings as "not comparable". We plan to enhance the tool to handle
strings of the different types in the future releases.
(AKC - 2014/04/18 HDFFV-8625, 8639, 8745)
- h5repack: h5repack would not remove user-defined filters.
Fixed by modifying h5repack to check if the filter is registered or
can be dynamically loaded. (ADB - 2014/03/03 HDFFV-8629)
F90 API
-------
- H5D_CHUNK_CACHE_NSLOTS_DFLT_F and H5D_CHUNK_CACHE_NBYTES_DFLT_F were
changed from the default KIND for INTEGER to INTEGER of KIND size_t.
(MSB - 2014/3/31 HDFFV-8689)
C++ API
------
- Added throw() to all exception destructors. Credit to Jason Newton
(nevion@gmail.com) for the patch. (BMR - 2014/4/15 HDFFV-8623)
- Changed the default value for H5Location::getComment from 256 to 0
to conform to C function and because it makes more sense.
(BMR - 2014/4/15)
High-Level APIs:
------
- None
Fortran High-Level APIs:
------
- None
Testing
-------
- testhdf5 now exits with EXIT_SUCCESS(0) if no errors, else
EXIT_FAILURE(1). (AKC - 2014/01/27 HDFFV-8572)
- The big test now pays attention to the HDF5_DRIVER environment variable.
Previously, it would run all tests with the family, stdio, and sec2
virtual file drivers (VFDs) for each VFD in the check-vfd make target,
regardless of the variable setting. It now checks the variable and
either runs the appropriate VFD-specific tests or skips as needed.
This saves much testing time. Fixes HDFFV-8554. (DER - 2014/04/11)
Supported Platforms
===================
The following platforms are supported and have been tested for this release.
They are built with the configure process unless specified otherwise.
AIX 5.3 xlc 10.1.0.5
(NASA G-ADA) xlC 10.1.0.5
xlf90 12.1.0.6
gmake v3.82
Linux 2.6.18-308.13.1.el5PAE GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP i686 i686 i386 compilers for 32-bit applications;
(jam) Version 4.1.2 20080704 (Red Hat 4.1.2-54)
Version 4.8.2
PGI C, Fortran, C++ Compilers for 32-bit
applications;
Version 13.7-0
Intel(R) C, C++, Fortran Compiler for 32-bit
applications;
Version 14.0.2 (Build 20140120)
Linux 2.6.18-371.6.1.el5 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers for 64-bit applications;
(koala) Version 4.1.2 20080704 (Red Hat 4.1.2-54)
Version 4.8.2
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 13.7-0
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64;
Version 14.0.2 (Build 20140120)
Linux 2.6.32-431.11.2.el6 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(platypus) Version 4.4.7 20120313
Version 4.8.2
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 13.7-0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 14.0.2 (Build 20140120)
Linux 2.6.32-431.11.2.el6.ppc64 gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
(ostrich) GNU Fortran (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
IBM XL C/C++ V11.1
IBM XL Fortran V13.1
SunOS 5.11 32- and 64-bit Sun C 5.12 SunOS_sparc
(emu) Sun Fortran 95 8.6 SunOS_sparc
Sun C++ 5.12 SunOS_sparc
Windows 7 Visual Studio 2008 w/ Intel Fortran 14 (cmake)
Visual Studio 2010 w/ Intel Fortran 14 (cmake)
Visual Studio 2012 w/ Intel Fortran 14 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.28(0.271/5/3) gcc(4.8.2) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 14 (cmake)
Visual Studio 2010 w/ Intel Fortran 14 (cmake)
Visual Studio 2012 w/ Intel Fortran 14 (cmake)
Windows 8.1 Visual Studio 2012 w/ Intel Fortran 14 (cmake)
Windows 8.1 x64 Visual Studio 2012 w/ Intel Fortran 14 (cmake)
Mac OS X Lion 10.7.3 Apple clang/clang++ version 3.0 from Xcode 4.6.1
64-bit gfortran GNU Fortran (GCC) 4.6.2
(duck) Intel icc/icpc/ifort version 13.0.3
Mac OS X Mt. Lion 10.8.5 Apple clang/clang++ version 5.0 from Xcode 5.0.2
64-bit gfortran GNU Fortran (GCC) 4.8.2
(swallow/kite) Intel icc/icpc/ifort version 14.0.2
Mac OS X Mavericks 10.9.2 Apple clang/clang++ version 5.1 from Xcode 5.1
64-bit gfortran GNU Fortran (GCC) 4.8.2
(wren/quail) Intel icc/icpc/ifort version 14.0.2
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
Solaris2.11 32-bit n y/y n y y y
Solaris2.11 64-bit n y/y n y y y
Windows 7 y y/y n y y y
Windows 7 x64 y y/y n y y y
Windows 7 Cygwin n y/y n y y y
Windows 8.1 n y/y n y y y
Windows 8.1 x64 n y/y n y y y
Mac OS X Lion 10.7.3 64-bit n y/y n y y y
Mac OS X Mountain Lion 10.8.1 64-bit n y/y n y y y
Mac OS X Mavericks 10.9.1 64-bit n y/y n y y y
AIX 5.3 32- and 64-bit n y/n n y y y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y/y y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel n y/y n y y y
CentOS 5.9 Linux 2.6.18-308 i686 PGI n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 GNU n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 GNU y y/y y y y y
CentOS 6.4 Linux 2.6.32 x86_64 Intel n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 PGI n y/y n y y y
Linux 2.6.32-431.11.2.el6.ppc64 n y/n n y y y
OpenVMS IA64 V8.4 n y/n n y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.11 32-bit y y y y
Solaris2.11 64-bit y y y y
Windows 7 y y y y
Windows 7 x64 y y y y
Windows 7 Cygwin n n n y
Windows 8.1 y y y y
Windows 8.1 x64 y y y y
Mac OS X Lion 10.7.3 64-bit y n y y
Mac OS X Mountain Lion 10.8.1 64-bit y n y y
Mac OS X Mavericks 10.9.1 64-bit y n y y
AIX 5.3 32- and 64-bit y n n y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel y y y n
CentOS 5.9 Linux 2.6.18-308 i686 PGI y y y n
CentOS 5.9 Linux 2.6.18 x86_64 GNU y y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel y y y n
CentOS 6.4 Linux 2.6.32 x86_64 GNU y y y n
CentOS 6.4 Linux 2.6.32 x86_64 Intel y y y n
CentOS 6.4 Linux 2.6.32 x86_64 PGI y y y n
Linux 2.6.32-431.11.2.el6.ppc64 y y y n
OpenVMS IA64 V8.4 n n n n
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
Linux 2.6.18-308.13.1.el5PAE MPICH mpich 3.1 compiled with
#1 SMP i686 i686 i386 gcc 4.8.2 and gfortran 4.8.2
(jam) g95 (GCC 4.0.3 (g95 0.94!)
Linux 2.6.18-431.11.2.el6 MPICH mpich 3.1 compiled with
#1 SMP x86_64 GNU/Linux gcc 4.8.2 and gfortran 4.8.2
(platypus) g95 (GCC 4.0.3 (g95 0.94!)
FreeBSD 8.2-STABLE i386 gcc 4.5.4 [FreeBSD] 20110526
(loyalty) gcc 4.6.1 20110527
g++ 4.6.1 20110527
gfortran 4.6.1 20110527
FreeBSD 8.2-STABLE amd64 gcc 4.5.4 [FreeBSD] 20110526
(freedom) gcc 4.6.1 20110527
g++ 4.6.1 20110527
gfortran 4.6.1 20110527
Debian7.1.0 3.2.0-4-amd64 #1 SMP Debian 3.2.51-1 x86_64 GNU/Linux
gcc (Debian 4.7.2-5) 4.7.2
GNU Fortran (Debian 4.7.2-5) 4.7.2
(cmake and autotools)
Fedora20 3.11.10-301.fc20.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.8.2 20131212 (Red Hat 4.8.2-7)
GNU Fortran (GCC) 4.8.2 20130603 (Red Hat 4.8.2-7)
(cmake and autotools)
SUSE 13.1 3.11.6-4-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.8.1
GNU Fortran (SUSE Linux) 4.8.1
(cmake and autotools)
Ubuntu 13.10 3.11.0-13-generic #20-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.8.1-10ubuntu8) 4.8.1
GNU Fortran (Ubuntu/Linaro 4.8.1-10ubuntu8) 4.8.1
(cmake and autotools)
Cray Linux Environment (CLE) PrgEnv-pgi/4.0.46
hopper.nersc.gov pgcc 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgf90 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgCC 12.5-0 64-bit target on x86-64 Linux -tp shanghai
Known Problems
==============
* When reading or writing a dataset (H5Dread/H5Dwrite) with a large selection
size (e.g., 2GB ~= 500 million of 4 bytes integers or floating point
numbers), some I/O systems may not be able to process it correctly.
We advise users to find out system limits before using large selections. If
I/O size limits exist, application should use HDF5 partial I/O capabilities
(e.g., H5Sselect_hyperslab(...)) to divide large requests into smaller sizes.
In this case we also advise users not to use chunk storage sizes larger that
2GB since the HDF5 library performs I/O on the entire chunk. We will work
on the HDF5 library to divide large data requests to smaller I/O requests.
(AKC 2014/05/02 HDFFV-8479)
* Due to an Intel compiler bug introduced in version 14.0.1, the HDF5 FORTRAN
wrappers do not work with configure option --enable-fortran2003.
However, the option --enable-fortran works with Intel 14.0.1. The compiler
bug was fixed in Intel version 14.0.2 and resolved the issue.
(MSB - 2014/4/15)
* Due to a PGI compiler bug introduced in versions before 13.3 and versions
after 14.2, the FORTRAN test 'Testing get file image' will fail.
(MSB - 2014/4/15)
* On CYGWIN, when building the library dynamically, testing will fail on
dynamically loaded filters. The test process will build dynamic filter
libraries with the *.dll.a extension, and the HDF5 Library will be looking
for *.so libraries. Entered as issue HDFFV-8736. (ADB - 2014/04/14)
* A Gnu Make directive (.NOTPARALLEL) is added to fortran/test/Makefile.
AIX native make does not support this directive and would fail if
parallel make (e.g. make -j4) is used to build the library. AIX users
either do not use parallel make or install Gnu Make to build the library.
(AKC 2014/04/08 HDFFV-8738)
* CLANG compiler with the options -fcatch-undefined-behavior and -ftrapv
catches some undefined behavior in the alignment algorithm of the macro
DETECT_I in H5detect.c. Since the algorithm is trying to detect the alignment
of integers, ideally the flag -fcatch-undefined-behavior should not to be
used for H5detect.c. In the future, we can separate flags for H5detect.c
from the rest of the library. (SLU - 2013/10/16 HDFFV-8147)
* Make provided by Solaris fails in "make check". Solaris users should use
gmake to build and install the HDF5 software. (AKC - 2013/10/08 - HDFFV-8534)
* On OpenVMS, two soft conversion functions (H5T__conv_i_f and H5T__conv_f_i)
have bugs. They convert data between floating-point numbers and integers.
But the library's default is hard conversion. The user should avoid
explicitly enabling soft conversion between floating-point numbers and
integers. (Issue VMS-8; SLU - 2013/09/19)
* On OpenVMS, ZLIB 1.2.8 library doesn't work properly. ZLIB 1.2.5 works
fine. So please use ZLIB 1.2.5 to build HDF5 library. (Issue VMS-5;
SLU 2013/09/19)
* When building using the Cray compilers on Cray machines, HDF5
configure mistakenly thinks the compiler is an intel compiler and
sets the -std=c99 flag which breaks configure on Cray. To build HDF5
properly on a Cray machine, please consult with the instructions in
INSTALL_parallel for building on Hopper.
(MSC - 2013/04/26 - HDFFV-8429)
* The 5.9 C++ compiler on Sun failed to compile a C++ test ttypes.cpp. It
complains with this message:
"/home/hdf5/src/H5Vprivate.h", line 130: Error: __func__ is not defined.
The reason is that __func__ is a predefined identifier in C99 standard. The
HDF5 C library uses it in H5private.h. The test ttypes.cpp includes
H5private.h (H5Tpkg.h<-H5Fprivate.h<-H5Vprivate.h<-H5private.h). Sun's 5.9
C++ compiler doesn't support __func__, thus fails to compile the C++ test.
But Sun's 5.11 C++ compiler does. To check whether your Sun C++ compiler
knows this identifier, try to compile the following simple C++ program:
#include<stdio.h>
int main(void)
{
printf("%s\n", __func__);
return 0;
}
(SLU - 2012/11/5)
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* On hopper, the build failed when RUNSERIAL and RUNPARALLEL are set
to aprun -np X, because the H5lib_settings.c file was not generated
properly. Not setting those environment variables works, because
configure was able to automatically detect that it's a Cray system
and used the proper launch commands when necessary.
(MSC - 2012/04/18)
* The data conversion test dt_arith.c fails in "long double" to integer
conversion on Ubuntu 11.10 (3.0.0.13 kernel) with GCC 4.6.1 if the library
is built with optimization -O3 or -O2. The older GCC (4.5) or newer kernal
(3.2.2 on Fedora) doesn't have the problem. Users should lower the
optimization level (-O1 or -O0) by defining CFLAGS in the command line of
"configure" like:
CFLAGS=-O1 ./configure
This will overwrite the library's default optimization level.
(SLU - 2012/02/07 - HDFFV-7829)
This issue is no longer present on Ubuntu 12.10 (3.5.0 kernel) with
gcc 4.7.2.
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* The C++ and FORTRAN bindings are not currently working on FreeBSD.
(QAK - 2011/04/26)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however.
(MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2.
(AKC - 2011/03/10)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO.
(CMC - 2009/04/28)
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.12%%%%
HDF5 version 1.8.12 released on 2013-11-04
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.11 and
HDF5-1.8.12, and contains information on the platforms tested and
known problems in HDF5-1.8.12.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.12 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.12 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.12 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.12 (current
release) versus Release 1.8.11":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.11
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration
-------------
- Added a configuration option to change the default plugin path.
The configure option is --with-default-plugindir=location.
The cmake option is -DH5_DEFAULT_PLUGINDIR:PATH=location.
HDFFV-8513. (ADB 2013/09/04)
- Renamed FFLAGS to FCFLAGS in configure. (ADB 2013/08/13)
- CMake can now package a compressed examples file, the default for
Windows binaries from HDF Group. (ADB - 2013/07/22)
Library
-------
- None
Parallel Library
----------------
- None
Tools
-----
- h5repack: Added the ability to use plugin filters to read and write
files. The option uses the filter number. HDFFV-8345
(ADB - 2013/09/04).
- h5dump: Added the option -N --any_path, which searches the file for
paths that match the search path. HDFFV-7989 (ADB - 2013/08/12).
- h5dump: Added the optional arg 0 to -A, which excludes attributes
from display. HDFFV-8134 (ADB - 2013/08/01).
High-Level APIs
---------------
- None
Fortran API
-----------
- None
C++ API
-------
- Added tutorial examples to C++/examples. They can be installed by
"make install-examples" and, in the installed directory, they can be
executed by running the script file run-c++-ex.sh. (BMR - 2013/09/28)
- A new class, H5::H5Location, is added to represent the location concept
in the C library. It is a base class to H5::H5File and H5::H5Ojbect,
whose member functions are moved into H5::H5Location. H5::H5File can
now inherent those functions. As a result, an H5::H5File object can have
an attribute. (BMR - 2013/09/27)
- Added wrappers for H5Rget_obj_type2 to retrieve the type of the object
that an object reference points to. (BMR - 2013/09/27)
H5O_type_t H5Location::getRefObjType(void *ref, H5R_type_t ref_type)
- Added wrappers for H5Aexist to check whether an attribute exists given
a name. (BMR - 2013/09/27)
bool H5::H5Location::attrExists(const char* name)
bool H5::H5Location::attrExists(const H5std_string& name)
- Added a number of overloaded functions for convenience. (BMR - 2013/09/27)
Support for New Platforms, Languages, and Compilers
===================================================
- None
Bug Fixes since HDF5-1.8.11
===========================
Configuration
-------------
- Modified H5detect.c to scan floating point types for padding bits before
analyzing the type further. This should fix problems with gcc 4.8.
(NAF - 2013/09/19 - HDFFV-8523/HDFFV-8500)
- HDF5 rpaths are no longer encoded in the library files when configured
with --disable-sharedlib-rpath. (LRK-2013-09-23 - HDFFV-8276)
Library
-------
- Added const qualifier to source buffer parameters in H5Dgather and
H5D_scatter_func_t (H5Dscatter callback). (NAF - 2013/7/09)
- CMake now creates *.so.{lt_version} files with the same version as
configure. (ADB - 2013/06/05 HDFFV-8437)
Parallel Library
----------------
- None
Performance
-------------
- None
Tools
-----
- h5dump: Added the option -N --any_path, which searches the file for
paths that match the search path. HDFFV-7989 (ADB - 2013/08/12).
- h5dump: Added the optional arg 0 to -A, which excludes attributes
from display. HDFFV-8134 (ADB - 2013/08/01).
- h5dump correctly exports subsetted data to a file, using the --output
option. (ADB - 2013/06/07 HDFFV-8447)
- h5cc and other compile scripts now default to linking shared libraries
when HDF5 is configured with the --disable-static option.
(LRK - 2013-09-23 - HDFFV-8141)
F90 API
-------
- None
C++ API
------
- None
High-Level APIs:
------
- None
Fortran High-Level APIs:
------
- None
Testing
-------
- test/big sometimes failed with the message of "file selection+offset not
within extent". This has been fixed. (AKC - 2013/09/28 HDFFV-8271).
- tools/h5diff/testh5diff.sh is run in every "make check", even after it
has passed in the previous run. It should not run again if there are no
code changes. Fixed. (AKC - 2013/07/19 HDFFV-8392)
Supported Platforms
===================
The following platforms are supported and have been tested for this release.
They are built with the configure process unless specified otherwise.
AIX 5.3 xlc 10.1.0.5
(NASA G-ADA) xlC 10.1.0.5
xlf90 12.1.0.6
Linux 2.6.18-308.13.1.el5PAE GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP i686 i686 i386 compilers for 32-bit applications;
(jam) Version 4.1.2 20080704 (Red Hat 4.1.2-54)
Version 4.8.1
PGI C, Fortran, C++ Compilers for 32-bit
applications;
Version 13.7-0
Intel(R) C, C++, Fortran Compiler for 32-bit
applications;
Version 13.1.3 (Build 20130607)
Linux 2.6.18-308.16.1.el5 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers for 64-bit applications;
(koala) Version 4.1.2 20080704 (Red Hat 4.1.2-54)
Version 4.8.1
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 13.7-0
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64;
Version 13.1.3 (Build 20130607)
Linux 2.6.32-358.18.1.el6 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(platypus) Version 4.4.7 20120313
Version 4.8.1
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 13.7-0
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 13.1.3 (Build 20130607)
Linux 2.6.32-358.18.1.el6.ppc64 gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
(ostrich) GNU Fortran (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
IBM XL C/C++ V11.1
IBM XL Fortran V13.1
SunOS 5.11 32- and 64-bit Sun C 5.12 SunOS_sparc
(emu) Sun Fortran 95 8.6 SunOS_sparc
Sun C++ 5.12 SunOS_sparc
Windows 7 Visual Studio 2008 w/ Intel Fortran 11 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Visual Studio 2012 w/ Intel Fortran 13 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.15(0.260/5/3) gcc(4.5.3) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 11 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Visual Studio 2012 w/ Intel Fortran 13 (cmake)
Windows 8.1 Visual Studio 2012 w/ Intel Fortran 13 (cmake)
Windows 8.1 x64 Visual Studio 2012 w/ Intel Fortran 13 (cmake)
Mac OS X Snow Leopard 10.6.8 gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 3.2.6
Darwin Kernel Version 10.8.0 g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 3.2.6
64-bit gfortran GNU Fortran (GCC) 4.6.2
(fred) Intel C (icc), Fortran (ifort), C++ (icpc)
12.1 Build 20120928
Mac OS X Lion 10.7.3 Apple clang/clang++ version 3.0 from Xcode 4.6.1
64-bit gfortran GNU Fortran (GCC) 4.6.2
(duck) Intel icc/icpc/ifort version 13.0.3
Mac OS X Mountain Lion 10.8.1 Apple clang/clang++ version 4.2 from Xcode 4.6.1
64-bit gfortran GNU Fortran (GCC) 4.6.2
(wren) Intel icc/icpc/ifort version 13.0.3
OpenVMS IA64 V8.4 HP C V7.3-018
HP Fortran V8.2-104939-50H96
HP C++ V7.4-004
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
Solaris2.11 32-bit n y/y n y y y
Solaris2.11 64-bit n y/y n y y y
Windows 7 y y/y n y y y
Windows 7 x64 y y/y n y y y
Windows 7 Cygwin n y/n n y y y
Windows 8.1 n y/y n y y y
Windows 8.1 x64 n y/y n y y y
Mac OS X Snow Leopard 10.6.8 64-bit n y/y n y y y
Mac OS X Lion 10.7.3 64-bit n y/y n y y y
Mac OS X Mountain Lion 10.8.1 64-bit n y/y n y y y
AIX 5.3 32- and 64-bit n y/n n y y y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y/y y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel n y/y n y y y
CentOS 5.9 Linux 2.6.18-308 i686 PGI n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 GNU y y/y y y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 PGI n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 GNU n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 Intel n y/y n y y y
Linux 2.6.32-358.2.1.el6.ppc64 n y/n n y y y
OpenVMS IA64 V8.4 n y/n n y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.11 32-bit y y y y
Solaris2.11 64-bit y y y y
Windows 7 y y y y
Windows 7 x64 y y y y
Windows 7 Cygwin n n n y
Windows 8.1 y y y y
Windows 8.1 x64 y y y y
Mac OS X Snow Leopard 10.6.8 64-bit y n y n
Mac OS X Lion 10.7.3 64-bit y n y y
Mac OS X Mountain Lion 10.8.1 64-bit y n y y
AIX 5.3 32- and 64-bit y n n y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel y y y n
CentOS 5.9 Linux 2.6.18-308 i686 PGI y y y n
CentOS 5.9 Linux 2.6.18 x86_64 GNU y y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel y y y n
CentOS 5.9 Linux 2.6.18 x86_64 PGI y y y n
CentOS 6.4 Linux 2.6.32 x86_64 GNU y y y n
CentOS 6.4 Linux 2.6.32 x86_64 Intel y y y n
Linux 2.6.32-358.2.1.el6.ppc64 y y y n
OpenVMS IA64 V8.4 n n n n
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
Linux 2.6.18-308.13.1.el5PAE MPICH mpich2-1.4.1p1 compiled with
#1 SMP i686 i686 i386 gcc 4.1.2 and gfortran 4.1.2
(jam) g95 (GCC 4.0.3 (g95 0.94!)
Linux 2.6.18-308.16.1.el5 MPICH mpich2-1.4.1p1 compiled with
#1 SMP x86_64 GNU/Linux gcc 4.1.2 and gfortran 4.1.2
(koala) g95 (GCC 4.0.3 (g95 0.94!)
FreeBSD 8.2-STABLE i386 gcc 4.5.4 [FreeBSD] 20110526
(loyalty) gcc 4.6.1 20110527
g++ 4.6.1 20110527
gfortran 4.6.1 20110527
FreeBSD 8.2-STABLE amd64 gcc 4.5.4 [FreeBSD] 20110526
(freedom) gcc 4.6.1 20110527
g++ 4.6.1 20110527
gfortran 4.6.1 20110527
Debian7.1.0 3.2.0-4-amd64 #1 SMP Debian 3.2.46-1 x86_64 GNU/Linux
gcc (Debian 4.7.2-5) 4.7.2
GNU Fortran (Debian 4.7.2-5) 4.7.2
(cmake and autotools)
Fedora19 3.11.1-200.fc19.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.8.1 20130603 (Red Hat 4.8.1-1)
GNU Fortran (GCC) 4.8.1 20130603 (Red Hat 4.8.1-1)
(cmake and autotools)
SUSE 12.3 3.7.10-1.16-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.7.2
GNU Fortran (SUSE Linux) 4.7.2
(cmake and autotools)
Ubuntu 13.04 3.8.0-30-generic #44-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.7.3-1ubuntu1) 4.7.3
GNU Fortran (Ubuntu/Linaro 4.7.3-1ubuntu1) 4.7.3
(cmake and autotools)
Cray Linux Environment (CLE) PrgEnv-pgi/4.0.46
hopper.nersc.gov pgcc 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgf90 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgCC 12.5-0 64-bit target on x86-64 Linux -tp shanghai
Known Problems
==============
* Several HDF5 command-line tools and tests leave behind generated files
that are not cleaned up with "make clean" or "make distclean" when software
is built in place. The issue will be addressed in the 1.8.13 release. We
recommend to use build directory to compile and test HDF5 as described
in the INSTALL file, section 4.2.
* Source directory names with spaces in them will cause failures in configure
or make on Mac (HDFFV-8152), Linux, and probably all other platforms. If a
configure command with a space is run from a build directory, it will exit
with an error message: "checking whether build environment is sane...
configure: error: unsafe srcdir value: '/scr/lrknox/hdf5 v1.8.12'". If
configure is run inside or below the directory with the space in the name,
libtool will get the directory path from the system, put the part of the
path before the space in the libdir variable in .../src/libhdf5.la, and
then fail to find the nonexistent directory. This is a known libtool issue
and the suggested workaround is to rename the directory without spaces.
(LRK - 2013/10/22)
* CLANG compiler with the options -fcatch-undefined-behavior and -ftrapv
catches some undefined behavior in the alignment algorithm of the macro
DETECT_I in H5detect.c (HDFFV-8147). This issue will be addressed in the
next release. (SLU - 2013/10/16)
* Running make check for the tools can fail in the tools tests if make was not
run prior. The tests for the tools use other tools in the tests, therefore
all the tools should be built before testing the tools. (ADB - 2013/10/09)
* Make provided by Solaris fails in "make check". Solaris users should use
gmake to build and install HDF5 software. (AKC - 2013/10/08 - HDFFV-8534)
* On OpenVMS, two soft conversion functions (H5T__conv_i_f and H5T__conv_f_i)
have bugs. They convert data between floating-point numbers and integers.
But the library's default is hard conversion. The user should avoid
explicitly enabling soft conversion between floating-point numbers and
integers. (Issue VMS-8; SLU - 2013/09/19)
* On OpenVMS, ZLIB 1.2.8 library doesn't work properly. ZLIB 1.2.5 works
fine. So please use ZLIB 1.2.5 to build HDF5 library. (Issue VMS-5;
SLU 2013/09/19)
* When building using the Cray compilers on Cray machines, HDF5
configure mistakenly thinks the compiler is an intel compiler and
sets the -std=c99 flag which breaks configure on Cray. To build HDF5
properly on a Cray machine, please consult with the instructions in
INSTALL_parallel for building on Hopper.
(MSC - 2013/04/26 - HDFFV-8429)
* The h5dump and h5diff utilities occasionally produce different output
between Linux and Windows systems. This is caused by lower-level library
routines that fail to write fill values to the user's buffer when reading
unallocated chunks from datasets that have a fill value set to
H5D_FILL_VALUE_DEFAULT. Due to platform differences the return of
spurious data values has only been encountered on Windows 32-bit systems.
(Issue HDFFV-8247; JP - 2013/03/27)
* The 5.9 C++ compiler on Sun failed to compile a C++ test ttypes.cpp. It
complains with this message:
"/home/hdf5/src/H5Vprivate.h", line 130: Error: __func__ is not defined.
The reason is that __func__ is a predefined identifier in C99 standard. The
HDF5 C library uses it in H5private.h. The test ttypes.cpp includes
H5private.h (H5Tpkg.h<-H5Fprivate.h<-H5Vprivate.h<-H5private.h). Sun's 5.9
C++ compiler doesn't support __func__, thus fails to compile the C++ test.
But Sun's 5.11 C++ compiler does. To check whether your Sun C++ compiler
knows this identifier, try to compile the following simple C++ program:
#include<stdio.h>
int main(void)
{
printf("%s\n", __func__);
return 0;
}
(SLU - 2012/11/5)
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* On hopper, the build failed when RUNSERIAL and RUNPARALLEL are set
to aprun -np X, because the H5lib_settings.c file was not generated
properly. Not setting those environment variables works, because
configure was able to automatically detect that it is a Cray system
and used the proper launch commands when necessary.
(MSC - 2012/04/18)
* The data conversion test dt_arith.c fails in "long double" to integer
conversion on Ubuntu 11.10 (3.0.0.13 kernel) with GCC 4.6.1 if the library
is built with optimization -O3 or -O2. The older GCC (4.5) or newer kernel
(3.2.2 on Fedora) do not have the problem. Users should lower the
optimization level (-O1 or -O0) by defining CFLAGS in the command line of
"configure" like:
CFLAGS=-O1 ./configure
This will overwrite the library's default optimization level.
(SLU - 2012/02/07 - HDFFV-7829)
This issue is no longer present on Ubuntu 12.10 (3.5.0 kernel) with
gcc 4.7.2.
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* The C++ and FORTRAN bindings are not currently working on FreeBSD.
(QAK - 2011/04/26)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however.
(MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2.
(AKC - 2011/03/10)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO.
(CMC - 2009/04/28)
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.11%%%%
HDF5 version 1.8.11 released on 2013-05-08
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.10 and
HDF5-1.8.11-*, and contains information on the platforms tested and
known problems in HDF5-1.8.11-*.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.11 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.11 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.11 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.11 (current
release) versus Release 1.8.10":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.10
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration
-------------
- Libtool version number is changed to 8.0.0 because there are API
changes. See below for details. (AKC - 2013/05/07 HDFFV-8435)
- Mac OS X 10.7 (Lion) and 10.8 (Mountain Lion) uses clang/clang++ as the
default C and C++ compilers. (AKC - 2013/04/19 HDFFV-8245)
- CMake minimum is now 2.8.10. (ADB 2013/1/14)
- A new tool, cmakehdf5, which is a build command script similar to
buildhdf5 is added and is available in the bin directory.
(AKC - 2013/01/16 HDFFV-8336)
Library
-------
- The library can load filter libraries dynamically during runtime. Users
can set the search path through environment variable HDF5_PLUGIN_PATH
and call H5Pset_filter to enable a dynamic filter. (SLU - 2013/04/08)
- Added new API functions H5Dscatter and H5Dgather to scatter data to and
and gather data from a selection within a memory buffer.
(NAF - 2013/02/05)
- The library now supports the data conversion from enumeration to numeric
(integer and floating-point number) datatypes. See Issue HDFFV-8221.
(SLU - 2012/10/23)
Parallel Library
----------------
- None
Tools
-----
- h5dump: added new option -O or -ddl to output the ddl text to a file. This
is a complement to the -o or --output option, which redirects the data to
a file. HDFFV-8229 (ADB - 2013/2/25)
High-Level APIs
---------------
- A new API function, H5DOwrite_chunk. This function writes a data chunk
directly into a file, bypassing hyperslab selection, data conversion,
and the filter pipeline. The user must be careful with the function and
clearly understand the I/O process of the library. (SLU - 2013/2/11)
Fortran API
-----------
- New API functions added (MSB - 2013/3/23):
h5odecr_refcount_f, h5oexists_by_name_f, h5oget_comment_f,
h5oget_comment_by_name_f, h5oincr_refcount_f, h5oopen_by_idx_f,
h5oset_comment_f, h5oset_comment_by_name_f, h5oset_comment_by_name_f
F2003: h5oget_info_f, h5oget_info_by_idx_f, h5ovisit_by_name_f
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- SunOS 5.11 (emu) 32-bit and 64-bit with Sun C/C++ 5.12 compiler and
Sun Fortran 95 8.6 compiler.
- Visual Studio 2012 w/ Intel Fortran 13 on Windows 7
- g95 released new version recently and is tested in this release.
Bug Fixes since HDF5-1.8.10
==========================
Configuration
-------------
- Fixed Thread-safe configure failure for the AIX platform.
(AKC - 2013/04/19 HDFFV-8390)
- Configure will check the result of header searches before searching for
the library.
Fixes HDFFV-8257 (ADB 2013/03/04)
- HDF does not support building SHARED Fortran libraries on OSX. Added
CMake code to check for this condition.
Fixes HDFFV-8227 (ADB 2013/03/04)
- CMake builds on Windows will no longer use legacy naming for libraries.
The "dll" tag will no longer be added to the name of *.lib and *.dll.
The option HDF_LEGACY_NAMING is now OFF by default.
Fixes HDFFV-8292 (ADB 2013/01/30)
Library
-------
- The library now behaves correctly when performing large I/O operations
on Mac OS-X. Previously, single I/O operations > 2 GB would fail
since the Darwin read/write calls cannot handle the number of bytes
that their parameter types imply.
Fixes HDFFV-7975 and HDFFV-8240 (DER 2013/01/07)
- Fixed a bug in the core VFD that causes failures when opening files
> 2 GB.
Fixes HDFFV-8124 and HDFFV-8158 (DER 2013/01/07)
- Fixed a bug where unintialized memory was read during variable-length
type conversion. This caused segfaults in netCDF. Fixes HDFFV-8159.
(DER 2013/03/30)
- Removed the H5Pset/get_dxpl_multi functions from the library. The
intended functionality for them was never fully implemented, and they
have always been fundamentally broken. NOTE: This does not affect
setting the multi VFD or any expected VFD functionality. Multi VFD
usage remains unchanged.
Fixes HDFFV-8296. (DER 2013/03/30)
Parallel Library
----------------
- None
Performance
-------------
- None
Tools
-----
- h5redeploy is changed to do this by default:
Installation directories:
prefix architecture-independent files.
exec_prefix architecture-dependent files, default is <prefix>.
libdir libraries, default is <exec_prefix>/lib.
includedir header files, default is <prefix/include>.
This allows users to just change the first line of prefix=<...> and the
effect will change libdir and includedir too. (AKC 2013/04/05 HDFFV-8358)
- h5repack: Fixed failure to convert the layout of a small chunked
dataset (size < 1K) to contiguous layout. HDFFV-8214 (JKM 2013/03/26)
- h5dump: Fixed displaying compression ratio for unknown or user-defined
filters. HDFFV-8344 (XCAO 2013/03/19)
- h5dump: Changed UNKNOWN_FILTER to USER_DEFINED_FILTER for user defined
filter. HDFFV-8346 (XCAO 2013/03/19)
- h5diff: Fixed to return the correct exit code 1 when the program
detects a unique extra attribute. Prior to this fix, h5diff returned
exit code 0 indicating the two files are identical.
HDFFV-7643 (JKM 2013/02/15)
- h5dump: Fixed writing nulls to a binary file when exporting a dataset
with compound string datatype. HDFFV-8169 (ADB 2013/1/31)
- The following h5stat test case failed in BG/P machines (and potentially
other machines that display extra output if an MPI task returns with a
non-zero code.)
Testing h5stat notexist.h5
The test script was fixed to ignore the extra output. HDFFV-8233
(AKC - 2012/11/30)
- h5diff: Improved speed when comparing HDF5 files with lots of
attributes. Much slower performance was identified with release versions
from 1.8.7 to 1.8.10 compared to 1.8.6. (JKM 2012/10/19)
F90 API
-------
- The integer type of the 'offset' argument in h5pset_external_f and
h5pget_external_f was changed to INTEGER(KIND=OFF_T) to support 8-byte
integers, matching the C type definition of off_t. (MSB - 2013/3/23)
- h5fc updated to recognize .f95, .f03 and .f08 file extensions.
C++ API
------
- The C++ wrappers DSetMemXferPropList::setMulti/getMulti were removed
because the C functions H5Pset/get_dxpl_multi functions are removed
from the library. Fixes HDFFV-8296 by DER. (BMR 2013/03/30)
- An exception thrown by an internal function was not propagating to the
test program during stack unwinding, so it couldn't be caught by the
test, and the program terminated "without an active exception." It
seemed that the problem happened when c_str() was used to generate
an equivalent const char* from a std::string and the resulting string
was passed to the internal function. As a work-around, we added a
try/catch around the the call to the internal function and when the
exception is caught there, it is re-thrown. Fixes HDFFV-8067.
(BMR 2013/03/30)
High-Level APIs:
------
- Fixed a problem with H5DSget_scale_name including the NULL terminator
in the size calculation returned by the function. The API was changed
to NOT include the NULL terminator in the size of name returned
(MSB- 2013/2/10)
Fortran High-Level APIs:
------
- None
Testing
-------
- In some Mac systems, testlibinfo.sh failed with this error:
Check file ../src/.libs/libhdf5.7.dylib
strings: object: ../src/.libs/libhdf5.7.dylib malformed object \
(unknown load command 15)
The strings command of Mac systems inspects library files, and older
versions of strings may not know newer library formats, resulting
in errors. Fixed by sending the library file as stdin to the strings
command to avoid this problem. (AKC - 2013/03/08 HDFFV-8305)
- Fixed a typo in the ERROR macro in test/testhdf5.h. It segmentation
faulted when used before. (AKC - 2013/02/12 HDFFV-8267)
Supported Platforms
===================
The following platforms are supported and have been tested for this release.
They are built with the configure process unless specified otherwise.
AIX 5.3 xlc 10.1.0.5
(NASA G-ADA) xlC 10.1.0.5
xlf90 12.1.0.6
Linux 2.6.18-308.13.1.el5PAE GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP i686 i686 i386 compilers for 32-bit applications;
(jam) Version 4.1.2 20080704 (Red Hat 4.1.2-54)
Version 4.6.3
PGI C, Fortran, C++ Compilers for 32-bit
applications;
Version 11.9-0
Version 12.5-0
Intel(R) C, C++, Fortran Compiler for 32-bit
applications;
Version 12.1 (Build 20110811)
Version 12.1 (Build 20120212)
Linux 2.6.18-308.16.1.el5 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers for 64-bit applications;
(koala) Version 4.1.2 20080704 (Red Hat 4.1.2-52)
Version 4.6.3
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 11.9-0
Version 12.5-0
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64;
Version 12.1 (Build 20110811)
Version 12.1 (Build 20120212)
Linux 2.6.32-358.2.1.el6 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers:
(platypus) Version 4.4.7 20120313
Intel(R) C (icc), C++ (icpc), Fortran (icc)
compilers:
Version 12.1 20120212
Linux 2.6.32-358.2.1.el6.ppc64 gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
(ostrich) GNU Fortran (GCC) 4.4.7 20120313 (Red Hat 4.4.7-3)
IBM XL C/C++ V11.1
IBM XL Fortran V13.1
SunOS 5.11 32- and 64-bit Sun C 5.12 SunOS_sparc
(emu) Sun Fortran 95 8.6 SunOS_sparc
Sun C++ 5.12 SunOS_sparc
Windows 7 Visual Studio 2008 w/ Intel Fortran 11 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Visual Studio 2012 w/ Intel Fortran 13 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.15(0.260/5/3) gcc(4.5.3) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 11 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Visual Studio 2012 w/ Intel Fortran 13 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.15(0.260/5/3) gcc(4.5.3) compiler and gfortran)
(cmake and autotools)
Mac OS X Snow Leopard 10.6.8 gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 3.2.6
Darwin Kernel Version 10.8.0 g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 3.2.6
64-bit gfortran GNU Fortran (GCC) 4.6.2
(fred) Intel C (icc), Fortran (ifort), C++ (icpc)
12.1 Build 20120928
Mac OS X Lion 10.7.3 Apple clang/clang++ version 3.0 from Xcode 4.6.1
64-bit gfortran GNU Fortran (GCC) 4.6.2
(duck) Intel icc/icpc/ifort version 13.0
Mac OS X Mountain Lion 10.8.1 Apple clang/clang++ version 4.2 from Xcode 4.6.1
64-bit gfortran GNU Fortran (GCC) 4.6.2
(wren) Intel icc/icpc/ifort version 13.0.1.119
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
Solaris2.11 32-bit n y/y n y y y
Solaris2.11 64-bit n y/n n y y y
Windows 7 y y/y n y y y
Windows 7 x64 y y/y n y y y
Windows 7 Cygwin n y/n n y y y
Windows 7 x64 Cygwin n y/n n y y y
Mac OS X Snow Leopard 10.6.8 64-bit n y/y n y y y
Mac OS X Lion 10.7.3 64-bit n y/y n y y y
Mac OS X Mountain Lion 10.8.1 64-bit n y/y n y y y
AIX 5.3 32- and 64-bit n y/n n y y y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y/y y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel n y/y n y y y
CentOS 5.9 Linux 2.6.18-308 i686 PGI n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 GNU y y/y y y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel n y/y n y y y
CentOS 5.9 Linux 2.6.18 x86_64 PGI n y/y n y y y
CentOS 6.4 Linux 2.6.32 x86_64 GNU n y/n n y y y
CentOS 6.4 Linux 2.6.32 x86_64 Intel n y/y n y y y
Linux 2.6.32-358.2.1.el6.ppc64 n y/n n y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.11 32-bit y y y y
Solaris2.11 64-bit y y y y
Windows 7 y y y y
Windows 7 x64 y y y y
Windows 7 Cygwin n n n y
Windows 7 x64 Cygwin n n n y
Mac OS X Snow Leopard 10.6.8 64-bit y n y n
Mac OS X Lion 10.7.3 64-bit y n y y
Mac OS X Mountain Lion 10.8.1 64-bit y n y y
AIX 5.3 32- and 64-bit y n n y
CentOS 5.9 Linux 2.6.18-308 i686 GNU y y y y
CentOS 5.9 Linux 2.6.18-308 i686 Intel y y y n
CentOS 5.9 Linux 2.6.18-308 i686 PGI y y y n
CentOS 5.9 Linux 2.6.18 x86_64 GNU y y y y
CentOS 5.9 Linux 2.6.18 x86_64 Intel y y y n
CentOS 5.9 Linux 2.6.18 x86_64 PGI y y y n
CentOS 6.4 Linux 2.6.32 x86_64 GNU y y y n
CentOS 6.4 Linux 2.6.32 x86_64 Intel y y y n
Linux 2.6.32-358.2.1.el6.ppc64 y y y n
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
Linux 2.6.18-308.13.1.el5PAE MPICH mpich2-1.4.1p1 compiled with
#1 SMP i686 i686 i386 gcc 4.1.2 and gfortran 4.1.2
(jam) g95 (GCC 4.0.3 (g95 0.94!)
Linux 2.6.18-308.16.1.el5 MPICH mpich2-1.4.1p1 compiled with
#1 SMP x86_64 GNU/Linux gcc 4.1.2 and gfortran 4.1.2
(koala) g95 (GCC 4.0.3 (g95 0.94!)
FreeBSD 8.2-STABLE i386 gcc 4.2.1 [FreeBSD] 20070719
(loyalty) gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
FreeBSD 8.2-STABLE amd64 gcc 4.2.1 [FreeBSD] 20070719
(freedom) gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
Debian6.0.7 2.6.32-5-amd64 #1 SMP x86_64 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
(cmake and autotools)
Fedora18 3.7.9-205.fc18.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.7.2 20121109 (Red Hat 4.7.2-8)
GNU Fortran (GCC) 4.7.2 20120507 (Red Hat 4.7.2-8)
(cmake and autotools)
SUSE 12.3 3.7.10-1.1-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.7.2
GNU Fortran (SUSE Linux) 4.7.2
(cmake and autotools)
Ubuntu 12.10 3.5.0-25-generic #39-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.7.2-2ubuntu1) 4.7.2
GNU Fortran (Ubuntu/Linaro 4.7.2-2ubuntu1) 4.7.2
(cmake and autotools)
Cray Linux Environment (CLE) PrgEnv-pgi/4.0.46
hopper.nersc.gov pgcc 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgf90 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgCC 12.5-0 64-bit target on x86-64 Linux -tp shanghai
Known Problems
==============
* When building using the Cray compilers on Cray machines, HDF5
configure mistakenly thinks the compiler is an intel compiler and
sets the -std=c99 flag which breaks configure on Cray. To build HDF5
properly on a Cray machine, please consult with the instructions in
INSTALL_parallel for building on Hopper.
(MSC - 2013/04/26 - HDFFV-8429)
* The h5dump and h5diff utilities occasionally produce different output
between Linux and Windows systems. This is caused by lower-level library
routines that fail to write fill values to the user's buffer when reading
unallocated chunks from datasets that have a fill value set to
H5D_FILL_VALUE_DEFAULT. Due to platform differences the return of
spurious data values has only been encountered on Windows 32-bit systems.
(Issue HDFFV-8247; JP - 2013/03/27)
* The 5.9 C++ compiler on Sun failed to compile a C++ test ttypes.cpp. It
complains with this message:
"/home/hdf5/src/H5Vprivate.h", line 130: Error: __func__ is not defined.
The reason is that __func__ is a predefined identifier in C99 standard. The
HDF5 C library uses it in H5private.h. The test ttypes.cpp includes
H5private.h (H5Tpkg.h<-H5Fprivate.h<-H5Vprivate.h<-H5private.h). Sun's 5.9
C++ compiler doesn't support __func__, thus fails to compile the C++ test.
But Sun's 5.11 C++ compiler does. To check whether your Sun C++ compiler
knows this identifier, try to compile the following simple C++ program:
#include<stdio.h>
int main(void)
{
printf("%s\n", __func__);
return 0;
}
(SLU - 2012/11/5)
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* On hopper, the build failed when RUNSERIAL and RUNPARALLEL are set
to aprun -np X, because the H5lib_settings.c file was not generated
properly. Not setting those environment variables works, because
configure was able to automatically detect that it's a Cray system
and used the proper launch commands when necessary.
(MSC - 2012/04/18)
* The data conversion test dt_arith.c fails in "long double" to integer
conversion on Ubuntu 11.10 (3.0.0.13 kernel) with GCC 4.6.1 if the library
is built with optimization -O3 or -O2. The older GCC (4.5) or newer kernal
(3.2.2 on Fedora) doesn't have the problem. Users should lower the
optimization level (-O1 or -O0) by defining CFLAGS in the command line of
"configure" like:
CFLAGS=-O1 ./configure
This will overwrite the library's default optimization level.
(SLU - 2012/02/07 - HDFFV-7829)
This issue is no longer present on Ubuntu 12.10 (3.5.0 kernel) with
gcc 4.7.2.
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* The C++ and FORTRAN bindings are not currently working on FreeBSD.
(QAK - 2011/04/26)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however.
(MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2.
(AKC - 2011/03/10)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* The library's test dt_arith.c showed a compiler's rounding problem on
Cygwin when converting from unsigned long long to long double. The
library's own conversion works fine. We defined a macro for Cygwin to
skip this test until we can solve the problem.
(SLU - 2010/05/05 - HDFFV-1264)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO.
(CMC - 2009/04/28)
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.10-patch1%%%%
HDF5 version 1.8.10-patch1 released on 2013-01-22
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.9 and
HDF5 1.8.10, and contains information on the platforms tested and
known problems in HDF5-1.8.10.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.10 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.10 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.10 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.10 (current
release) versus Release 1.8.9":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.10
- Bug Fixes since HDF5-1.8.9
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration
-------------
- A new tool, cmakehdf5, which is a build command script similar to
buildhdf5 is added and is available in the bin directory.
(AKC - 2013/01/21)
Library
-------
- Updated to latest autotools and changed all hard *.sh scripts to
configure managed *.sh.in files. Removed overloading of autotools
TESTS variable by examples and tests. Renamed configure.in to
configure.ac. (ADB - 2012/08/23 - HDFFV-8129)
- The data sieve buffer size was set for all the datasets in the file. It
could waste memory if any dataset size is smaller than the sieve buffer
size. Now the library picks the smaller one between the dataset size
and the sieve buffer size from the file access property. See Issue 7934.
(SLU - 2012/4/11)
Parallel Library
----------------
- Added the H5Pget_mpio_no_collective_cause() function that retrieves
reasons why the collective I/O was broken during read/write IO access.
(JKM - 2012/08/30 HDFFV-8143)
- Added H5Pget_mpio_actual_io_mode_f (MSB - 2012/09/27)
Tools
-----
- h5import: Changed to allow the use of h5dump output as input files to
h5import. h5dump must include the "-p" option to print the properties;
configuration file is captured output of h5dump. The restrictions are
that only one dataset with a simple datatype (integer, floating-point,
or string) can be processed. Integers and floating-point imports from
h5dump must use the "binary" option for the data file. The string version
uses the h5dump "-y --width=1" options to disable the indexing printouts,
print single columns, and obviously NOT use the "binary" option.
(ADB - 2012/07/19 HDFFV-721)
High-Level APIs
---------------
- None
Fortran API
-----------
- Fixed a typo in return value of the nh5dread_f_c function (was 1
instead of 0 on success); fixed the return value to make it consistent
with other Fortran functions; cleaned debug statements from the code.
(EIP - 2012/06/23)
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- None
Bug Fixes since HDF5-1.8.10
===========================
Library
-------
- The library now behaves correctly when performing large I/O operations on
Mac OS-X. Previously, single I/O operations > 2 GB would fail since the
Darwin read/write calls cannot handle the number of bytes that their
parameter types imply.
Fixes HDFFV-7975 and HDFFV-8240 (DER - 07 JAN 2013)
- Fixed a bug in the core VFD that cause failures when opening files > 2 GB.
Fixes HDFFV-8124 and HDFFV-8158 (DER - 07 JAN 2013)
Tools
-----
- The following h5stat test case failed in BG/P machines (and potentially
other machines that display extra output if an MPI task returns with a
non-zero code.)
Testing h5stat notexist.h5
The test script was fixed to ignore the extra output.
HDFFV-8233 (AKC - 2012/12/17)
- h5diff: Fixed slowness when comparing HDF5 files with many attributes.
Much slower performance was identified with later release version
(from 1.8.7 to 1.8.10) compared to 1.8.6. The issue was introduced
from fixing an attribute related bug for 1.8.7 release in the past.
HDFFV-8145 (JKM 2012/12/13)
Testing
-------
- None
Bug Fixes since HDF5-1.8.9
==========================
Configuration
-------------
- Fixed configure --enable-production to not use -O optimization for Lion
and Mountain Lion systems when gcc (i686-apple-darwin11-llvm-gcc-4.2
(GCC) 4.2.1) is used. Somehow the -O optimization will cause some of
the hard conversion code in test/dt_arith.c to fail. HDFFV-8017.
(AKC - 2012/10/10)
- Fixed AIX Fortran compiler flags to use appropriate settings for
debugging, profiling, and optimization situations. HDFFV-8069.
(AKC 2012/09/27)
Library
-------
- Fixed a memory leak exposed when inserting/removing a property
from a property list several times. HDFFV-8022. (MSC 2012/05/18)
- The file_image test will fail in the "initial file image and callbacks in
the core VFD" sub-test if the source directory is read-only as the test
fails to create its test files in the build directory. This has been
fixed. HDFFV-8009 (AKC - 2012/07/06)
Parallel Library
----------------
- The MPI-POSIX VFD was updated to include the POSIX and Windows
correctness features added that had already been added to the other VFDs.
HDFFV-8058/7845. (DER 2012/09/17)
Performance
-------------
- Removed program perform/benchpar from the enable-build-all list. The
program will be retired or moved to another location. HDFFV-8156
(AKC 2012/10/01)
- Retired program perform/mpi-perf. Its purpose has been incorporated
into h5perf. (AKC 2012/09/21)
Tools
-----
- h5repack: "h5repack -f NONE file1.h5 out.h5" command failed if
source file contains chunked dataset and a chunk dim is bigger than
the dataset dim. Another issue is that the command changed max dims
if chunk dim is smaller than the dataset dim. These issue occurred
when dataset size is smaller than 64k (compact size limit) Fixed both.
HDFFV-8012 (JKM 2012/09/24)
- h5diff: Fixed the counter in verbose mode (-v, -r) so that it will no
longer add together the differences between datasets and the differences
between attributes of those datasets. This change makes the output of
verbose mode consistent for datasets, groups, and committed datatypes.
HDFFV-5919 (JKM 2012/09/10)
- h5diff: Fixed the incorrect result when comparing attribute data
values and the data type has the same class but different sizes.
HDFFV-7942 (JKM 2012/08/15)
- h5dump: Replaced single element fwrite with block writes.
HDFFV-1208 (ADB 2012/08/13)
- h5diff: Fixed test failure for "make check" due to failure of
copying test files when performed in HDF5 source tree. Also applied
to other tools. HDFFV-8107 (JKM 2012/08/01)
- ph5diff: Fixed intermittent hang issue on a certain operation in
parallel mode. It was detected by daily test for comparing
non-comparable objects, but it could have occurred in other
operations depending on machine condition. HDFFV-8003 (JKM 2012/08/01)
- h5diff: Fixed the function COPY_TESTFILES_TO_TESTDIR() of testh5diff.sh
to better report when there is an error in the file copying.
HDFFV-8105 (AKC 2012/07/22)
- h5dump: Fixed the sort by name display to maintain correct parent/child
relationships between ascending/descending order.
HDFFV-8095 (ADB 2012/07/12)
- h5dump: Fixed the display by creation order when using option -n
(print contents).
HDFFV-5942 (ADB 2012/07/09)
- h5dump: Changed to allow H5T_CSET_UTF8 to be displayed in h5dump output.
Used technique similar to what was done in h5ls (matches library
options).
HDFFV-7999 (ADB 2012/05/23)
- h5diff: Fixed the tool so that it will not check and display the status
of dangling links without setting the --follow-symlinks option. This
also improved performance when comparing lots of external links without
the --follow-symlinks option.
HDFFV-7998 (JKM 2012/04/26)
F90 API
-------
- Fixed a typo in return value of the nh5dread_f_c function (was 1
instead of 0 on success); fixed the return value to make it consistent
with other Fortran functions; cleaned debug statements from the code.
(EIP - 2012/06/23)
- Fixed a problem writing/reading control characters to a dataset; writing
a string containing alerts, backspace, carriage_return, form_feed,
horizontal_tab, vertical_tab, or new_line is now tested and working.
(MSB - 2012/09/01)
- Corrected the integer type of H5S_UNLIMITED_F to HSIZE_T (MSB - 2012/09/01)
- Corrected the number of continuation lines in the src files
to be less than 32 lines for F95 compliance. (MSB - 2012/10/01)
C++ API
------
- None
High-Level APIs:
------
- Fixed problem with H5TBdelete_record destroying all data following the
deletion of a row. (MSB- 2012/7/26)
- Fixed H5LTget_attribute_string not closing an object identifier when an
error occurs. (MSB- 2012/7/21)
- Corrected the return type of H5TBAget_fill from herr_t to htri_t to
reflect that a return value of 1 indicates that a fill value is
present, 0 indicates a fill value is not present, and <0 indicates an
error.
Fortran High-Level APIs:
------
- None
Supported Platforms
===================
AIX 5.3 xlc 10.1.0.5
(NASA G-ADA) xlC 10.1.0.5
xlf90 12.1.0.6
Linux 2.6.18-308.13.1.el5PAE GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP i686 i686 i386 compilers for 32-bit applications;
(jam) Version 4.1.2 20080704 (Red Hat 4.1.2-52)
Version 4.6.3
PGI C, Fortran, C++ Compilers for 32-bit
applications;
Version 11.9-0
Intel(R) C, C++, Fortran Compiler for 32-bit
applications;
Version 12.1
MPICH mpich2-1.4.1p1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.18-308.16.1.el5 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers for 32-bit applications;
(koala) Version 4.1.2 20080704 (Red Hat 4.1.2-52)
Version 4.6.3
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 11.9-0
Version 12.5-0
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64;
Version 12.1 (Build 20110811)
Version 12.1 (Build 20120212)
MPICH mpich2-1.4.1p1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.32-220.7.1.el6.ppc64 gcc (GCC) 4.4.6 20110731 (Red Hat 4.4.6-3)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.6 20110731
(ostrich) GNU Fortran (GCC) 4.4.6 20110731 (Red Hat 4.4.6-3)
Linux 2.6.32-220.23.1.1chaos Intel C, C++, Fortran Compilers
ch5.x86_64 GNU/Linux Version 12.1.5.339
(LLNL Aztec)
IBM Blue Gene/P XL C for Blue Gene/P, bgxlc V9.0
(LLNL uDawn) XL C++ for Blue Gene/P, bgxlC V9.0
XL Fortran for Blue Gene/P, bgxlf90 V11.1
SunOS 5.10 32- and 64-bit Sun C 5.9 Sun OS_sparc Patch 124867-16
(linew) Sun Fortran 95 8.3 Sun OS_sparc Patch 127000-13
Sun C++ 5.9 Sun OS_sparc Patch 124863-26
Sun C 5.11 SunOS_sparc
Sun Fortran 95 8.5 SunOS_sparc
Sun C++ 5.11 SunOS_sparc
Windows XP Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Windows XP x64 Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Windows 7 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.15(0.260/5/3) gcc(4.5.3) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.15(0.260/5/3) gcc(4.5.3) compiler and gfortran)
(cmake and autotools)
Mac OS X Snow Leopard 10.6.8 gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 3.2.6
Darwin Kernel Version 10.8.0 g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 3.2.6
(fred) gfortran GNU Fortran (GCC) 4.6.2
Intel C (icc), Fortran (ifort), C++ (icpc)
12.1.0.038 Build 20110811
Mac OS X Snow Leopard 10.6.8 gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 3.2.6
Darwin Kernel Version 10.8.0 g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 3.2.6
Intel 32-bit gfortran GNU Fortran (GCC) 4.6.1
(tejeda) Intel C (icc), Fortran (ifort), C++ (icpc)
12.1.0.038 Build 20110811
Mac OS X Lion 10.7.3 gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 4.2.1
32- and 64-bit g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 4.2.1
(duck) gfortran GNU Fortran (GCC) 4.6.2
Mac OS X Mountain Lion 10.8.1 cc Apple clang version 4.0 from Xcode 4.5.1
(owl) c++ Apple clang version 4.0 from Xcode 4.5.1
gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 4.5.1
g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 4.5.1
gfortran GNU Fortran (GCC) 4.6.2
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
Solaris2.10 32-bit n y/y n y y y
Solaris2.10 64-bit n y/n n y y y
Windows 7 y y/n n y y y
Windows 7 x64 y y/n n y y y
Mac OS X Snow Leopard 10.6.8 32-bit n y/y n y y n
Mac OS X Snow Leopard 10.6.8 64-bit n y/y n y y y
Mac OS X Lion 10.7.3 32-bit n y/y n y y n
Mac OS X Lion 10.7.3 64-bit n y/y n y y y
Mac OS X Mountain Lion 10.8.1 64-bit n y/n n y y n
AIX 5.3 32- and 64-bit y y/n y y y y
CentOS 5.5 Linux 2.6.18-308 i686 GNU y y/y y y y y
CentOS 5.5 Linux 2.6.18-308 i686 Intel n y/y n y y y
CentOS 5.5 Linux 2.6.18-308 i686 PGI n y/y n y y y
CentOS 5.5 Linux 2.6.18 x86_64 GNU y y/y y y y y
CentOS 5.5 Linux 2.6.18 x86_64 Intel n y/y n y y y
CentOS 5.5 Linux 2.6.18 x86_64 PGI n y/y n y y y
Linux 2.6.32-220.7.1.el6.ppc64 n y/n n y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit n n n n
Windows 7 y y y y
Windows 7 x64 y y y y
Mac OS X Snow Leopard 10.6.8 32-bit y n y n
Mac OS X Snow Leopard 10.6.8 64-bit y n y n
Mac OS X Lion 10.7.3 32-bit y n y y
Mac OS X Lion 10.7.3 64-bit y n y y
Mac OS X Mountain Lion 10.8.1 64-bit y n y y
AIX 5.3 32- and 64-bit n n n y
CentOS 5.5 Linux 2.6.18-308 i686 GNU y y y y
CentOS 5.5 Linux 2.6.18-308 i686 Intel y y y n
CentOS 5.5 Linux 2.6.18-308 i686 PGI y y y n
CentOS 5.5 Linux 2.6.18 x86_64 GNU y y y y
CentOS 5.5 Linux 2.6.18 x86_64 Intel y y y n
CentOS 5.5 Linux 2.6.18 x86_64 PGI y y y n
Linux 2.6.32-220.7.1.el6.ppc64 y y y n
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
FreeBSD 8.2-STABLE i386 gcc 4.2.1 [FreeBSD] 20070719
(loyalty) gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
FreeBSD 8.2-STABLE amd64 gcc 4.2.1 [FreeBSD] 20070719
(freedom) gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
Debian6.0.3 2.6.32-5-686 #1 SMP i686 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
(cmake and autotools)
Debian6.0.3 2.6.32-5-amd64 #1 SMP x86_64 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
(cmake and autotools)
Fedora17 3.5.2-1.fc17.i6866 #1 SMP i686 i686 i386 GNU/Linux
gcc (GCC) 4.7.0 20120507 (Red Hat 4.7.0-5)
GNU Fortran (GCC) 4.7.0 20120507 (Red Hat 4.7.0-5)
(cmake and autotools)
Fedora17 3.5.2-1.fc17.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.7.0 20120507 (Red Hat 4.7.0-5)
GNU Fortran (GCC) 4.7.0 20120507 (Red Hat 4.7.0-5)
(cmake and autotools)
SUSE 12.2 3.4.6-2.10-desktop #1 SMP PREEMPT i686 i686 i386 GNU/Linux
gcc (SUSE Linux) 4.7.1
GNU Fortran (SUSE Linux) 4.7.1
(cmake and autotools)
SUSE 12.2 3.4.6-2.10-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.7.1
GNU Fortran (SUSE Linux) 4.7.1
(cmake and autotools)
Ubuntu 12.04 3.2.0-29-generic #46-Ubuntu SMP i686 GNU/Linux
gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
GNU Fortran (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
(cmake and autotools)
Ubuntu 12.04 3.2.0-29-generic #46-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
GNU Fortran (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
(cmake and autotools)
(Use optimization level -O1)
Cray Linux Environment (CLE) PrgEnv-pgi/4.0.46
hopper.nersc.gov pgcc 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgf90 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgCC 12.5-0 64-bit target on x86-64 Linux -tp shanghai
Known Problems
==============
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* On hopper, the build failed when RUNSERIAL and RUNPARALLEL are set
to aprun -np X, because the H5lib_settings.c file was not generated
properly. Not setting those environment variables works, because
configure was able to automatically detect that it's a Cray system
and used the proper launch commands when necessary.
(MSC - 2012/04/18)
* The data conversion test dt_arith.c fails in "long double" to integer
conversion on Ubuntu 11.10 (3.0.0.13 kernal) with GCC 4.6.1 if the library
is built with optimization -O3 or -O2. The older GCC (4.5) or newer kernal
(3.2.2 on Fedora) doesn't have the problem. Users should lower the
optimization level (-O1 or -O0) by defining CFLAGS in the command line of
"configure" like:
CFLAGS=-O1 ./configure
This will overwrite the library's default optimization level.
(SLU - 2012/02/07 - HDFFV-7829)
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* The C++ and FORTRAN bindings are not currently working on FreeBSD.
(QAK - 2011/04/26)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however.
(MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2.
(AKC - 2011/03/10)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* The library's test dt_arith.c showed a compiler's rounding problem on
Cygwin when converting from unsigned long long to long double. The
library's own conversion works fine. We defined a macro for Cygwin to
skip this test until we can solve the problem.
(SLU - 2010/05/05 - HDFFV-1264)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO.
(CMC - 2009/04/28)
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.10%%%%
HDF5 version 1.8.10 released on 2012-10-26
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.9 and
HDF5 1.8.10, and contains information on the platforms tested and
known problems in HDF5-1.8.10.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.10 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.10 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.10 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.10 (current
release) versus Release 1.8.9":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.9
- Supported Platforms
- Supported Configuration Features Summary
- More Tested Platforms
- Known Problems
New Features
============
Configuration
-------------
- None
Library
-------
- Updated to latest autotools and changed all hard *.sh scripts to
configure managed *.sh.in files. Removed overloading of autotools
TESTS variable by examples and tests. Renamed configure.in to
configure.ac. (ADB - 2012/08/23 - HDFFV-8129)
- The data sieve buffer size was set for all the datasets in the file. It
could waste memory if any dataset size is smaller than the sieve buffer
size. Now the library picks the smaller one between the dataset size
and the sieve buffer size from the file access property. See Issue 7934.
(SLU - 2012/4/11)
Parallel Library
----------------
- Added the H5Pget_mpio_no_collective_cause() function that retrieves
reasons why the collective I/O was broken during read/write IO access.
(JKM - 2012/08/30 HDFFV-8143)
- Added H5Pget_mpio_actual_io_mode_f (MSB - 2012/09/27)
Tools
-----
- h5import: Changed to allow the use of h5dump output as input files to
h5import. h5dump must include the "-p" option to print the properties;
configuration file is captured output of h5dump. The restrictions are
that only one dataset with a simple datatype (integer, floating-point,
or string) can be processed. Integers and floating-point imports from
h5dump must use the "binary" option for the data file. The string version
uses the h5dump "-y --width=1" options to disable the indexing printouts,
print single columns, and obviously NOT use the "binary" option.
(ADB - 2012/07/19 HDFFV-721)
High-Level APIs
---------------
- None
Fortran API
-----------
- Fixed a typo in return value of the nh5dread_f_c function (was 1
instead of 0 on success); fixed the return value to make it consistent
with other Fortran functions; cleaned debug statements from the code.
(EIP - 2012/06/23)
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- None
Bug Fixes since HDF5-1.8.9
==========================
Configuration
-------------
- Fixed configure --enable-production to not use -O optimization for Lion
and Mountain Lion systems when gcc (i686-apple-darwin11-llvm-gcc-4.2
(GCC) 4.2.1) is used. Somehow the -O optimization will cause some of
the hard conversion code in test/dt_arith.c to fail. HDFFV-8017.
(AKC - 2012/10/10)
- Fixed AIX Fortran compiler flags to use appropriate settings for
debugging, profiling, and optimization situations. HDFFV-8069.
(AKC 2012/09/27)
Library
-------
- Fixed a memory leak exposed when inserting/removing a property
from a property list several times. HDFFV-8022. (MSC 2012/05/18)
- The file_image test will fail in the "initial file image and callbacks in
the core VFD" sub-test if the source directory is read-only as the test
fails to create its test files in the build directory. This has been
fixed. HDFFV-8009 (AKC - 2012/07/06)
Parallel Library
----------------
- The MPI-POSIX VFD was updated to include the POSIX and Windows
correctness features added that had already been added to the other VFDs.
HDFFV-8058/7845. (DER 2012/09/17)
Performance
-------------
- Removed program perform/benchpar from the enable-build-all list. The
program will be retired or moved to another location. HDFFV-8156
(AKC 2012/10/01)
- Retired program perform/mpi-perf. Its purpose has been incorporated
into h5perf. (AKC 2012/09/21)
Tools
-----
- h5repack: "h5repack -f NONE file1.h5 out.h5" command failed if
source file contains chunked dataset and a chunk dim is bigger than
the dataset dim. Another issue is that the command changed max dims
if chunk dim is smaller than the dataset dim. These issue occurred
when dataset size is smaller than 64k (compact size limit) Fixed both.
HDFFV-8012 (JKM 2012/09/24)
- h5diff: Fixed the counter in verbose mode (-v, -r) so that it will no
longer add together the differences between datasets and the differences
between attributes of those datasets. This change makes the output of
verbose mode consistent for datasets, groups, and committed datatypes.
HDFFV-5919 (JKM 2012/09/10)
- h5diff: Fixed the incorrect result when comparing attribute data
values and the data type has the same class but different sizes.
HDFFV-7942 (JKM 2012/08/15)
- h5dump: Replaced single element fwrite with block writes.
HDFFV-1208 (ADB 2012/08/13)
- h5diff: Fixed test failure for "make check" due to failure of
copying test files when performed in HDF5 source tree. Also applied
to other tools. HDFFV-8107 (JKM 2012/08/01)
- ph5diff: Fixed intermittent hang issue on a certain operation in
parallel mode. It was detected by daily test for comparing
non-comparable objects, but it could have occurred in other
operations depending on machine condition. HDFFV-8003 (JKM 2012/08/01)
- h5diff: Fixed the function COPY_TESTFILES_TO_TESTDIR() of testh5diff.sh
to better report when there is an error in the file copying.
HDFFV-8105 (AKC 2012/07/22)
- h5dump: Fixed the sort by name display to maintain correct parent/child
relationships between ascending/descending order.
HDFFV-8095 (ADB 2012/07/12)
- h5dump: Fixed the display by creation order when using option -n
(print contents).
HDFFV-5942 (ADB 2012/07/09)
- h5dump: Changed to allow H5T_CSET_UTF8 to be displayed in h5dump output.
Used technique similar to what was done in h5ls (matches library
options).
HDFFV-7999 (ADB 2012/05/23)
- h5diff: Fixed the tool so that it will not check and display the status
of dangling links without setting the --follow-symlinks option. This
also improved performance when comparing lots of external links without
the --follow-symlinks option.
HDFFV-7998 (JKM 2012/04/26)
F90 API
-------
- Fixed a typo in return value of the nh5dread_f_c function (was 1
instead of 0 on success); fixed the return value to make it consistent
with other Fortran functions; cleaned debug statements from the code.
(EIP - 2012/06/23)
- Fixed a problem writing/reading control characters to a dataset; writing
a string containing alerts, backspace, carriage_return, form_feed,
horizontal_tab, vertical_tab, or new_line is now tested and working.
(MSB - 2012/09/01)
- Corrected the integer type of H5S_UNLIMITED_F to HSIZE_T (MSB - 2012/09/01)
- Corrected the number of continuation lines in the src files
to be less than 32 lines for F95 compliance. (MSB - 2012/10/01)
C++ API
------
- None
High-Level APIs:
------
- Fixed problem with H5TBdelete_record destroying all data following the
deletion of a row. (MSB- 2012/7/26)
- Fixed H5LTget_attribute_string not closing an object identifier when an
error occurs. (MSB- 2012/7/21)
- Corrected the return type of H5TBAget_fill from herr_t to htri_t to
reflect that a return value of 1 indicates that a fill value is
present, 0 indicates a fill value is not present, and <0 indicates an
error.
Fortran High-Level APIs:
------
- None
Supported Platforms
===================
AIX 5.3 xlc 10.1.0.5
(NASA G-ADA) xlC 10.1.0.5
xlf90 12.1.0.6
Linux 2.6.18-308.13.1.el5PAE GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP i686 i686 i386 compilers for 32-bit applications;
(jam) Version 4.1.2 20080704 (Red Hat 4.1.2-52)
Version 4.6.3
PGI C, Fortran, C++ Compilers for 32-bit
applications;
Version 11.9-0
Intel(R) C, C++, Fortran Compiler for 32-bit
applications;
Version 12.1
MPICH mpich2-1.4.1p1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.18-308.16.1.el5 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers for 32-bit applications;
(koala) Version 4.1.2 20080704 (Red Hat 4.1.2-52)
Version 4.6.3
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 11.9-0
Version 12.5-0
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64;
Version 12.1 (Build 20110811)
Version 12.1 (Build 20120212)
MPICH mpich2-1.4.1p1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.32-220.7.1.el6.ppc64 gcc (GCC) 4.4.6 20110731 (Red Hat 4.4.6-3)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.6 20110731
(ostrich) GNU Fortran (GCC) 4.4.6 20110731 (Red Hat 4.4.6-3)
Linux 2.6.32-220.23.1.1chaos Intel C, C++, Fortran Compilers
ch5.x86_64 GNU/Linux Version 12.1.5.339
(LLNL Aztec)
IBM Blue Gene/P XL C for Blue Gene/P, bgxlc V9.0
(LLNL uDawn) XL C++ for Blue Gene/P, bgxlC V9.0
XL Fortran for Blue Gene/P, bgxlf90 V11.1
SunOS 5.10 32- and 64-bit Sun C 5.9 Sun OS_sparc Patch 124867-16
(linew) Sun Fortran 95 8.3 Sun OS_sparc Patch 127000-13
Sun C++ 5.9 Sun OS_sparc Patch 124863-26
Sun C 5.11 SunOS_sparc
Sun Fortran 95 8.5 SunOS_sparc
Sun C++ 5.11 SunOS_sparc
Windows XP Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Windows XP x64 Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Windows 7 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.15(0.260/5/3) gcc(4.5.3) compiler and gfortran)
(cmake and autotools)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Cygwin(CYGWIN_NT-6.1 1.7.15(0.260/5/3) gcc(4.5.3) compiler and gfortran)
(cmake and autotools)
Mac OS X Snow Leopard 10.6.8 gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 3.2.6
Darwin Kernel Version 10.8.0 g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 3.2.6
(fred) gfortran GNU Fortran (GCC) 4.6.2
Intel C (icc), Fortran (ifort), C++ (icpc)
12.1.0.038 Build 20110811
Mac OS X Snow Leopard 10.6.8 gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 3.2.6
Darwin Kernel Version 10.8.0 g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 3.2.6
Intel 32-bit gfortran GNU Fortran (GCC) 4.6.1
(tejeda) Intel C (icc), Fortran (ifort), C++ (icpc)
12.1.0.038 Build 20110811
Mac OS X Lion 10.7.3 gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 4.2.1
32- and 64-bit g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 4.2.1
(duck) gfortran GNU Fortran (GCC) 4.6.2
Mac OS X Mountain Lion 10.8.1 cc Apple clang version 4.0 from Xcode 4.5.1
(owl) c++ Apple clang version 4.0 from Xcode 4.5.1
gcc i686-apple-darwin11-llvm-gcc-4.2 (GCC) 4.2.1 from Xcode 4.5.1
g++ i686-apple-darwin11-llvm-g++-4.2 (GCC) 4.2.1 from Xcode 4.5.1
gfortran GNU Fortran (GCC) 4.6.2
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90/ F90 C++ zlib SZIP
parallel F2003 parallel
Solaris2.10 32-bit n y/y n y y y
Solaris2.10 64-bit n y/n n y y y
Windows 7 y y/n n y y y
Windows 7 x64 y y/n n y y y
Mac OS X Snow Leopard 10.6.8 32-bit n y/y n y y n
Mac OS X Snow Leopard 10.6.8 64-bit n y/y n y y y
Mac OS X Lion 10.7.3 32-bit n y/y n y y n
Mac OS X Lion 10.7.3 64-bit n y/y n y y y
Mac OS X Mountain Lion 10.8.1 64-bit n y/n n y y n
AIX 5.3 32- and 64-bit y y/n y y y y
CentOS 5.5 Linux 2.6.18-308 i686 GNU y y/y y y y y
CentOS 5.5 Linux 2.6.18-308 i686 Intel n y/y n y y y
CentOS 5.5 Linux 2.6.18-308 i686 PGI n y/y n y y y
CentOS 5.5 Linux 2.6.18 x86_64 GNU y y/y y y y y
CentOS 5.5 Linux 2.6.18 x86_64 Intel n y/y n y y y
CentOS 5.5 Linux 2.6.18 x86_64 PGI n y/y n y y y
Linux 2.6.32-220.7.1.el6.ppc64 n y/n n y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit n n n n
Windows 7 y y y y
Windows 7 x64 y y y y
Mac OS X Snow Leopard 10.6.8 32-bit y n y n
Mac OS X Snow Leopard 10.6.8 64-bit y n y n
Mac OS X Lion 10.7.3 32-bit y n y y
Mac OS X Lion 10.7.3 64-bit y n y y
Mac OS X Mountain Lion 10.8.1 64-bit y n y y
AIX 5.3 32- and 64-bit n n n y
CentOS 5.5 Linux 2.6.18-308 i686 GNU y y y y
CentOS 5.5 Linux 2.6.18-308 i686 Intel y y y n
CentOS 5.5 Linux 2.6.18-308 i686 PGI y y y n
CentOS 5.5 Linux 2.6.18 x86_64 GNU y y y y
CentOS 5.5 Linux 2.6.18 x86_64 Intel y y y n
CentOS 5.5 Linux 2.6.18 x86_64 PGI y y y n
Linux 2.6.32-220.7.1.el6.ppc64 y y y n
Compiler versions for each platform are listed in the preceding
"Supported Platforms" table.
More Tested Platforms
=====================
The following platforms are not supported but have been tested for this release.
FreeBSD 8.2-STABLE i386 gcc 4.2.1 [FreeBSD] 20070719
(loyalty) gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
FreeBSD 8.2-STABLE amd64 gcc 4.2.1 [FreeBSD] 20070719
(freedom) gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
Debian6.0.3 2.6.32-5-686 #1 SMP i686 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
(cmake and autotools)
Debian6.0.3 2.6.32-5-amd64 #1 SMP x86_64 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
(cmake and autotools)
Fedora17 3.5.2-1.fc17.i6866 #1 SMP i686 i686 i386 GNU/Linux
gcc (GCC) 4.7.0 20120507 (Red Hat 4.7.0-5)
GNU Fortran (GCC) 4.7.0 20120507 (Red Hat 4.7.0-5)
(cmake and autotools)
Fedora17 3.5.2-1.fc17.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.7.0 20120507 (Red Hat 4.7.0-5)
GNU Fortran (GCC) 4.7.0 20120507 (Red Hat 4.7.0-5)
(cmake and autotools)
SUSE 12.2 3.4.6-2.10-desktop #1 SMP PREEMPT i686 i686 i386 GNU/Linux
gcc (SUSE Linux) 4.7.1
GNU Fortran (SUSE Linux) 4.7.1
(cmake and autotools)
SUSE 12.2 3.4.6-2.10-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.7.1
GNU Fortran (SUSE Linux) 4.7.1
(cmake and autotools)
Ubuntu 12.04 3.2.0-29-generic #46-Ubuntu SMP i686 GNU/Linux
gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
GNU Fortran (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
(cmake and autotools)
Ubuntu 12.04 3.2.0-29-generic #46-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
GNU Fortran (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3
(cmake and autotools)
(Use optimization level -O1)
Cray Linux Environment (CLE) PrgEnv-pgi/4.0.46
hopper.nersc.gov pgcc 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgf90 12.5-0 64-bit target on x86-64 Linux -tp shanghai
pgCC 12.5-0 64-bit target on x86-64 Linux -tp shanghai
Known Problems
==============
* The following h5stat test case fails in BG/P machines (and potentially other
machines that display extra output if an MPI task returns with a non-zero
code.)
Testing h5stat notexist.h5
The test actually runs and passes as expected. It is the extra output from
the MPI process that causes the test script to fail. This will be fixed
in the next release. (AKC - 2012/10/25 - HDFFV-8233)
* The C++ and FORTRAN bindings are not currently working on FreeBSD with the
native release 8.2 compilers (4.2.1), but are working with gcc 4.6 from the
ports (and probably gcc releases after that).
(QAK - 2012/10/19)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* On hopper, the build failed when RUNSERIAL and RUNPARALLEL are set
to aprun -np X, because the H5lib_settings.c file was not generated
properly. Not setting those environment variables works, because
configure was able to automatically detect that it's a Cray system
and used the proper launch commands when necessary.
(MSC - 2012/04/18)
* The data conversion test dt_arith.c fails in "long double" to integer
conversion on Ubuntu 11.10 (3.0.0.13 kernal) with GCC 4.6.1 if the library
is built with optimization -O3 or -O2. The older GCC (4.5) or newer kernal
(3.2.2 on Fedora) doesn't have the problem. Users should lower the
optimization level (-O1 or -O0) by defining CFLAGS in the command line of
"configure" like:
CFLAGS=-O1 ./configure
This will overwrite the library's default optimization level.
(SLU - 2012/02/07 - HDFFV-7829)
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14 - HDFFV-8235)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* The C++ and FORTRAN bindings are not currently working on FreeBSD.
(QAK - 2011/04/26)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however.
(MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2.
(AKC - 2011/03/10)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* The library's test dt_arith.c showed a compiler's rounding problem on
Cygwin when converting from unsigned long long to long double. The
library's own conversion works fine. We defined a macro for Cygwin to
skip this test until we can solve the problem.
(SLU - 2010/05/05 - HDFFV-1264)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO.
(CMC - 2009/04/28)
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
%%%%1.8.9%%%%
HDF5 version 1.8.9 released on 2012-05-09
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.8 and
HDF5 1.8.9. It also contains information on the platforms tested and
known problems in HDF5-1.8.9.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.9 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.9 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.9 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.9 (current
release) versus Release 1.8.8":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.8
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- None
Library
-------
- Added new feature to merge committed datatypes when copying objects,
using new H5O_COPY_MERGE_COMMITTED_DTYPE_FLAG, modified by new API
routines: H5Padd_merge_committed_dtype_path(),
H5Pfree_merge_committed_dtype_paths(), H5Pset_mcdt_search_cb() and
H5Pget_mcdt_search_cb(). (QAK - 2012/03/30)
- Added new feature which allows working with files in memory in the
same ways files are worked with on disk. New API routines include
H5Pset_file_image, H5Pget_file_image, H5Pset_file_image_callbacks,
H5Pget_file_image_callbacks, H5Fget_file_image, and
H5LTopen_file_image. (QAK - 2012/04/17)
Parallel Library
----------------
- Corrected memory allocation error in MPI datatype construction code.
(QAK - 2012/04/23)
- Add two new routines to set/get the atomicity parameter in the
MPI library to perform atomic operations. Some file systems (for
example PVFS2) do not support atomic updates, so those routines
would not be supported. (MSC - 2012/03/27 - HDFFV-7961)
Tools
-----
- h5repack: Added ability to set the metadata block size of the output
file, with the '-M'/'--metadata_block_size' command line parameter.
(QAK - 2012/03/30)
- h5stat: Added ability to display a summary of the file space usage for a
file, with the '-S'/'--summary' command line parameter. (QAK - 2012/03/28)
- h5dump: Added capability for "-a" option to show attributes containing "/"
by using an escape character. For example, for a dataset "/dset"
containing attribute "speed(m/h)", use "h5dump -a "/dset/speed(\/h)"
to show the content of the attribute. (PC - 2012/03/12 - HDFFV-7523)
- h5dump: Added ability to apply command options across multiple files using a
wildcard in the filename. Unix example; "h5dump -H -d Dataset1 tarr*.h5".
Cross platform example; "h5dump -H -d Dataset1 tarray1.h5 tarray2.h5 tarray3.h5".
(ADB - 2012/03/12 - HDFFV-7876).
- h5dump: Added new option --no-compact-subset. This option will not
interpret the '[' character as starting the compact form of
subsetting. This is useful when the "h5dump error: unable to
open dataset "datset_name"" message is output because a dataset
name contains a '[' character. (ADB - 2012/03/05 - HDFFV-7689).
- h5repack: Improved performance for big chunked datasets (size > 128MB)
when used with the layout (-l) or compression (-f) options.
Before this change, repacking datasets with chunks with a large first
dimension would take extremely long. For example, repacking a dataset
with chunk dimensions of 1024x5x1 might take many hours to process
while changing a dataset with chunk dimensions set to 1x5x1024
might take under an hour. After this change, processing the dataset
with chunk dimensions of 1024x5x1 takes about 15 minutes, and processing
a dataset with chunk dimensions of 1x5x1024 takes about 14 minutes.
(JKM - 2012/03/01 - HDFFV-7862)
High-Level APIs
---------------
- New API: H5LTpath_valid (Fortran: h5ltpath_valid_f) checks
if a path is correct, determines if a link resolves to a valid
object, and checks that the link does not dangle. (MSB - 2012/03/15)
Fortran API
-----------
- Added for the C API the Fortran wrapper:
h5ocopy_f (MSB - 2012/03/22)
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- None
Bug Fixes since HDF5-1.8.8
==========================
Configuration
-------------
- Fixed Makefile issue in which "-Wl," was not properly specified
prior to -rpath when building parallel Fortran libraries with
an Intel compiler. (MAM - 2012/03/26)
- Makefiles generated by other packages using h5cc as the compiler
no longer error when 'make' is invoked more than once in order
to 'rebuild' after changes to source. (MAM - 2012/03/26)
- Added code to display the version information of XL Fortran and C++
in the summary of configure. (AKC - 2012/02/28 - HDFFV-7793)
- Updated all CMakeLists.txt files to indicate the minimum CMake version is
the current standard of 2.8.6 (ADB - 2011/12/05 - HDFFV-7854)
Library
-------
- Windows and STDIO correctness changes have been propagated from the SEC2
and old Windows drivers to the STDIO VFD. (DER - 2012/03/30 - HDFFV-7917)
- Fixed an error that would occur when copying an object with attribute
creation order tracked and indexed. (NAF - 2012/03/28 - HDFFV-7762)
- Fixed a bug in H5Ocopy(): When copying an opened object, call the
object's flush class action to ensure that cached data is flushed so
that H5Ocopy will get the correct data. (VC - 2012/03/27 - HDFFV-7853)
- The istore test will now skip the sparse 50x50x50 test when the VFD does
not support sparse files on that platform. The most important platforms
on which this will be skipped are Windows (NTFS sparse files are not
supported) and Mac OS-X (HFS sparse files are not supported). This
fixes CTest timeout issues on Windows. (DER - 2012/03/27 - HDFFV-7769)
- Windows and POSIX correctness changes have been propagated from the SEC2
VFD to the Core VFD. This mainly affects file operations on the
driver's backing store and fixes a problem on Windows where large files
could not be read. (DER - 2012/03/27 - HDFFV-7916 - HDFFV-7603)
- When an application tries to write or read many small data chunks and
runs out of memory, the library had a segmentation fault. The fix is to
return the error stack with proper information.
(SLU - 2012/03/23 - HDFFV-7785)
- H5Pset_data_transform had a segmentation fault in some cases like x*-100.
It works correctly now and handles other cases like 100-x or 2/x.
(SLU - 2012/03/15 - HDFFV-7922)
- Fixed rare corruption bugs that could occur when using the new object
header format. (NAF - 2012/03/15 - HDFFV-7879)
- Fixed an error that occurred when creating a contiguous dataset with a
zero-sized dataspace and space allocation time set to 'early'.
(QAK - 2012/03/12)
- Changed Windows thread creation to use _beginthread() instead of
CreateThread(). Threads created by the latter can be killed in
low-memory situations. (DER - 2012/02/10 - HDFFV-7780)
- Creating a dataset in a read-only file caused a segmentation fault when
the file is closed. It's fixed. The attempt to create a dataset will
fail with an error indicating the file is read-only.
(SLU - 2012/01/25 - HDFFV-7756)
- Fixed a segmentation fault that could occur when shrinking a dataset
with chunks larger than 1 MB. (NAF - 2011/11/30 - HDFFV-7833)
- Fixed a bug that could cause H5Oget_info to return the wrong address
after copying a committed (named) datatype. (NAF - 2011/11/14)
- The library allowed the conversion of strings between ASCII and UTF8
We have corrected it to report an error under this situation.
(SLU - 2011/11/8 - HDFFV-7582)
- Fixed a segmentation fault when the library tried to shrink the size
of a compound datatype through H5Tset_size immediately after the
datatype was created. (SLU - 2011/11/4 - HDFFV-7618)
Parallel Library
----------------
- None
Tools
-----
- h5unjam: Fixed a segmentation fault that occurred when h5unjam was used
with the -V (show version) option. (JKM - 2012/04/19 - HDFFV-8001)
- h5repack: Fixed a failure that occurred when repacking the chunk size
of a specified chunked dataset with unlimited max dims.
(JKM - 2012/04/11 - HDFFV-7993)
- h5diff: Fixed a failure when comparing groups. Before the fix, if an
object in a group was compared with an object in another group where
both had the same name but the object type was different, then h5diff
would fail. After the fix, h5diff detects such cases as non-comparable
and displays appropriate error messages.
(JKM - 2012/03/28 - HDFFV-7644)
- h5diff: If unique objects exist only in one file and if h5diff is set to
exclude the unique objects with the --exclude-path option, then h5diff
might miss excluding some objects. This was fixed to correctly exclude
objects. (JKM - 2012/03/20 - HDFFV-7837)
- h5diff: When two symbolic dangling links are compared with the
--follow-symlinks option, the result should be the same. This worked when
comparing two files, but didn't work when comparing two objects.
h5diff now works when comparing two objects.
(JKM - 2012/03/09 - HDFFV-7835)
- h5dump: Added the tools library error stack to properly catch error
information generated within the library. (ADB - 2012/03/12 - HDFFV-7958)
- h5dump: Changed the process where an open link used to fail. Now dangling
links no longer throw error messages. (ADB - 2012/03/12 - HDFFV-7839)
- h5dump: Refactored code to remove duplicated functions. Split XML
functions from DDL functions. Corrected indentation and formatting
errors. Also fixed subsetting counting overflow (HDFFV-5874). Verified
all tools call tools_init() in main. The USER_BLOCK data now correctly
displays within the SUPER_BLOCK info. NOTE: WHITESPACE IN THE OUTPUT
HAS CHANGED. (ADB - 2012/02/17 - HDFFV-7560)
- h5diff: Fixed to prevent from displaying error stack message when
comparing two dangling symbolic links with the follow-symlinks option.
(JKM - 2012/01/13 - HDFFV-7836)
- h5repack: Fixed a memory leak that occurred with the handling of
variable length strings in attributes.
(JKM - 2012/01/10 - HDFFV-7840)
- h5ls: Fixed a segmentation fault that occurred when accessing region
reference data in an attribute. (JKM - 2012/01/06 - HDFFV-7838)
F90 API
-------
- None
C++ API
------
- None
High-Level APIs:
------
- None
Fortran High-Level APIs:
------
- h5ltget_attribute_string_f: The h5ltget_attribute_string_f used to return
the C NULL character in the returned character buffer. The returned
charactor buffer now does not return the C NULL character; the buffer
is blank-padded if needed. (MSB - 2012/03/23)
Platforms Tested
================
The following platforms and compilers have been tested for this release.
AIX 5.3 xlc 10.1.0.5
(NASA G-ADA) xlC 10.1.0.5
xlf90 12.1.0.6
FreeBSD 8.2-STABLE i386 gcc 4.2.1 [FreeBSD] 20070719
(loyalty) g++ 4.2.1 [FreeBSD] 20070719
gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
FreeBSD 8.2-STABLE amd64 gcc 4.2.1 [FreeBSD] 20070719
(freedom) g++ 4.2.1 [FreeBSD] 20070719
gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
Linux 2.6.18-194.3.1.el5PAE GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP i686 i686 i386 compilers for 32-bit applications;
(jam) Version 4.1.2 20080704 (Red Hat 4.1.2-52)
Version 4.5.2
PGI C, Fortran, C++ Compilers for 32-bit
applications;
Version 11.8-0
Version 11.9-0
Intel(R) C, C++, Fortran Compiler for 32-bit
applications;
Version 12.0
Version 12.1
MPICH mpich2-1.3.1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.18-308.1.1.el5 GNU C (gcc), Fortran (gfortran), C++ (g++)
#1 SMP x86_64 GNU/Linux compilers for 32-bit applications;
(koala) Version 4.1.2 20080704 (Red Hat 4.1.2-52)
Version 4.5.2
PGI C, Fortran, C++ for 64-bit target on
x86-64;
Version 11.9-0 (64-bit)
Version 11.8-0 (32-bit)
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64;
Version 12.0
Version 12.1
MPICH mpich2-1.3.1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.32-220.7.1.el6.ppc64 gcc (GCC) 4.4.6 20110731 (Red Hat 4.4.6-3)
#1 SMP ppc64 GNU/Linux g++ (GCC) 4.4.6 20110731
(ostrich) GNU Fortran (GCC) 4.4.6 20110731 (Red Hat 4.4.6-3)
Linux 2.6.18-108chaos Intel C, C++, Fortran Compilers Version 11.1
#1 SMP x86_64 GNU/Linux
(LLNL Aztec)
IBM Blue Gene/P XL C for Blue Gene/P, bgxlc V9.0
(LLNL uDawn) XL C++ for Blue Gene/P, bgxlC V9.0
XL Fortran for Blue Gene/P, bgxlf0 V11.1
SunOS 5.10 32- and 64-bit Sun C 5.9 Sun OS_sparc Patch 124867-16
(linew) Sun Fortran 95 8.3 Sun OS_sparc Patch 127000-13
Sun C++ 5.9 Sun OS_sparc Patch 124863-26
Sun C 5.11 SunOS_sparc
Sun Fortran 95 8.5 SunOS_sparc
Sun C++ 5.11 SunOS_sparc
SGI Altix UV Intel(R) C, Fortran Compilers
SGI ProPack 7 Linux Version 11.1 20100806
2.6.32.24-0.2.1.2230.2.PTF- SGI MPT 2.02
default #1 SMP
(NCSA ember)
Dell NVIDIA Cluster Intel(R) C, Fortran Compilers
Red Hat Enterprise Linux 6 Version 12.0.4 20110427
2.6.32-131.4.1.el6.x86_64 mvapich2 1.7rc1-intel-12.0.4
(NCSA forge)
Windows XP Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Windows XP x64 Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Windows 7 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Cygwin(1.7.9 native gcc(4.5.3) compiler and gfortran)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 w/ Intel Fortran 12 (cmake)
Cygwin(1.7.9 native gcc(4.5.3) compiler and gfortran)
Mac OS X Snow Leopard 10.6.8 i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (gcc)
Darwin Kernel Version 10.8.0 i686-apple-darwin10-g++-4.2.1 (GCC) 4.2.1 (g++)
Intel 64-bit (Apple Inc. build 5666) (dot 3)
(fred) GNU Fortan (GCC) 4.6.1 (gfortran)
Intel C (icc), Fortran (ifort), C++ (icpc)
12.1.0.038 Build 20110811
Mac OS X Snow Leopard 10.6.8 i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (gcc)
Darwin Kernel Version 10.8.0 i686-apple-darwin10-g++-4.2.1 (GCC) 4.2.1 (g++)
Intel 32-bit (Apple Inc. build 5666) (dot 3)
(tejeda) GNU Fortran (GCC) 4.6.1 (gfortran)
Intel C (icc), Fortran (ifort), C++ (icpc)
12.1.0.038 Build 20110811
Mac OS X Lion 10.7.3 GCC 4.2.1 gcc
32- and 64-bit GNU Fortran (GCC) 4.6.1 gfortran
(duck) GCC 4.2.1. g++
Debian6.0.3 2.6.32-5-686 #1 SMP i686 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
Debian6.0.3 2.6.32-5-amd64 #1 SMP x86_64 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
Fedora16 3.2.9-2.fc16.i6866 #1 SMP i686 i686 i386 GNU/Linux
gcc (GCC) 4.6.2 20111027 (Red Hat 4.6.2-1)
GNU Fortran (GCC) 4.6.2 20111027 (Red Hat 4.6.2-1)
Fedora16 3.2.9-2.fc16.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.6.2 20111027 (Red Hat 4.6.2-1)
GNU Fortran (GCC) 4.6.2 20111027 (Red Hat 4.6.2-1)
SUSE 12.1 3.1.9-1.4-desktop #1 SMP PREEMPT i686 i686 i386 GNU/Linux
gcc (SUSE Linux) 4.6.2
GNU Fortran (SUSE Linux) 4.6.2
SUSE 12.1 3.1.9-1.4-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.6.2
GNU Fortran (SUSE Linux) 4.6.2
Ubuntu 11.10 3.0.0-16-generic #29-Ubuntu SMP i686 GNU/Linux
gcc (Ubuntu/Linaro 4.6.1-9ubuntu3) 4.6.1
GNU Fortran (Ubuntu/Linaro 4.6.4-9ubuntu3) 4.6.1
Ubuntu 11.10 3.0.0-16-generic #29-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.6.1-9ubuntu3) 4.6.1
GNU Fortran (Ubuntu/Linaro 4.6.1-9ubuntu3) 4.6.1
Cray Linux Environment (CLE) PrgEnv-pgi 2.2.74
hopper.nersc.gov pgcc 11.9-0 64-bit target on x86-64 Linux -tp k8e
pgf90 11.9-0 64-bit target on x86-64 Linux -tp k8e
pgCC 11.9-0 64-bit target on x86-64 Linux -tp k8e
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
Solaris2.10 32-bit n y n y y y
Solaris2.10 64-bit n y n y y y
Windows XP n y(4) n y y y
Windows XP x64 n y(4) n y y y
Windows Vista n y(4) n y y y
Windows Vista x64 n y(4) n y y y
Mac OS X Snow Leopard 10.6.8 32-bit n y n y y n
Mac OS X Snow Leopard 10.6.8 64-bit n y n y y y
Mac OS X Lion 10.7.3 32-bit n y n y y n
Mac OS X Lion 10.7.3 64-bit n y n y y y
AIX 5.3 32- and 64-bit y y y y y y
FreeBSD 8.2-STABLE 32&64 bit n x n x y y
CentOS 5.5 Linux 2.6.18-194 i686 GNU (1)W y y(2) y y y y
CentOS 5.5 Linux 2.6.18-194 i686 Intel W n y n y y y
CentOS 5.5 Linux 2.6.18-194 i686 PGI W n y n y y y
CentOS 5.5 Linux 2.6.18 x86_64 GNU (1) W y y(3) y y y y
CentOS 5.5 Linux 2.6.18 x86_64 Intel W n y n y y y
CentOS 5.5 Linux 2.6.18 x86_64 PGI W n y n y y y
Linux 2.6.32-220.7.1.el6.ppc64 n y n y y y
SGI ProPack 7 Linux 2.6.32.24 y y y y y y
Red Hat Enterprise Linux 6 y y y y y y
CLE hopper.nersc.gov y y(3) y y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit n n n n
Windows XP y y(4) y n
Windows XP x64 y y(4) y n
Windows Vista y y(4) y y
Windows Vista x64 y y(4) y y
Mac OS X Snow Leopard 10.6.8 32-bit y n y n
Mac OS X Snow Leopard 10.6.8 64-bit y n y n
Mac OS X Lion 10.7.3 32-bit y n y y
Mac OS X Lion 10.7.3 64-bit y n y y
AIX 5.3 32- and 64-bit n n n y
FreeBSD 8.2-STABLE 32&64 bit y x x y
CentOS 5.5 Linux 2.6.18-194 i686 GNU (1)W y y(2) y y
CentOS 5.5 Linux 2.6.18-194 i686 Intel W y y y n
CentOS 5.5 Linux 2.6.18-194 i686 PGI W y y y n
CentOS 5.5 Linux 2.6.18 x86_64 GNU (1) W y y y y
CentOS 5.5 Linux 2.6.18 x86_64 Intel W y y y n
CentOS 5.5 Linux 2.6.18 x86_64 PGI W y y y n
Linux 2.6.32-220.7.1.el6.ppc64 y y y n
SGI ProPack 7 Linux 2.6.32.24 y y y n
Red Hat Enterprise Linux 6 y y y n
CLE hopper.nersc.gov n n n n
(1) Fortran compiled with gfortran.
(2) With PGI and Absoft compilers.
(3) With PGI compiler for Fortran.
(4) Using Visual Studio 2008 w/ Intel Fortran 10.1 (Cygwin shared libraries are not supported)
(5) C and C++ shared libraries will not be built when Fortran is enabled.
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* The h5repacktst test fails on AIX 32-bit because the test uses more
memory than the default amount. The failure message typically looks like:
"time: 0551-010 The process was stopped abnormally. Try again."
This is an issue with the test only and does not represent a problem with
the library. To allow the test to pass, request more memory when testing
via appropriate command such as:
$ env LDR_CNRTL=MAXDATA=0x20000000@DSA make check
(AKC - 2012/05/09 - HDFFV-8016)
* The file_image test will fail in the "initial file image and callbacks in
the core VFD" sub-test if the source directory is read-only as the test
fails to create its test files in the build directory. This will be
resolved in a future release.
(AKC - 2012/05/05 - HDFFV-8009)
* The dt_arith test reports several errors involving "long double" on
Mac OS X 10.7 Lion when any level of optimization is enabled. The test does
not fail in debug mode. This will be addressed in a future release.
(SLU - 2012/05/08)
* The following h5dump test case fails in BG/P machines (and potentially other
machines that use a command script to launch executables):
h5dump --no-compact-subset -d "AHFINDERDIRECT::ah_centroid_t[0] it=0 tl=0"
tno-subset.h5
This is due to the embedded spaces in the dataset name being interpreted
by the command script launcher as meta-characters, thus passing three
arguments to h5dump's -d flag. The command passes if run by hand, just
not via the test script.
(AKC - 2012/05/03)
* The ph5diff (parallel h5diff) tool can intermittently hang in parallel mode
when comparing two HDF5 files that contain objects with the same names but
with different object types.
(JKM - 2012/04/27)
* On hopper, the build failed when RUNSERIAL and RUNPARALLEL are set
to aprun -np X, because the H5lib_settings.c file was not generated
properly. Not setting those environment variables works, because
configure was able to automatically detect that it's a Cray system
and used the proper launch commands when necessary.
(MSC - 2012/04/18)
* The data conversion test dt_arith.c fails in "long double" to integer
conversion on Ubuntu 11.10 (3.0.0.13 kernal) with GCC 4.6.1 if the library
is built with optimization -O3 or -O2. The older GCC (4.5) or newer kernal
(3.2.2 on Fedora) doesn't have the problem. Users should lower the
optimization level (-O1 or -O0) by defining CFLAGS in the command line of
"configure" like:
CFLAGS=-O1 ./configure
This will overwrite the library's default optimization level.
(SLU - 2012/02/07 - HDFFV-7829)
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release.
(DER - 2011/10/14)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in a future release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value.
(AKC - 2011/05/07)
* The C++ and FORTRAN bindings are not currently working on FreeBSD.
(QAK - 2011/04/26)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however.
(MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2.
(AKC - 2011/03/10)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* The library's test dt_arith.c showed a compiler's rounding problem on
Cygwin when converting from unsigned long long to long double. The
library's own conversion works fine. We defined a macro for Cygwin to
skip this test until we can solve the problem.
(SLU - 2010/05/05 - HDFFV-1264)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* MinGW has a missing libstdc++.dll.a library file and will not successfully link
C++ applications/tests. Do not use the enable-cxx configure option. Read all of
the INSTALL_MINGW.txt file for all restrictions.
(ADB - 2009/11/11)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* The PathScale MPI implementation, accessing a Panasas file system, would
cause H5Fcreate() with H5F_ACC_EXCL to fail even when the file does not
exist. This is due to the MPI_File_open() call failing if the mode has
the MPI_MODE_EXCL bit set.
(AKC - 2009/08/11 - HDFFV-988)
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO.
(CMC - 2009/04/28)
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh and
tools/h5copy/testh5copy.sh will fail some of its sub-tests. These sub-tests
are expected to fail and should exit with a non-zero code but the yod
command does not propagate the exit code of the executables. Yod always
returns 0 if it can launch the executable. The test suite shell expects
a non-zero for this particular test, therefore it concludes the test has
failed when it receives 0 from yod. Skip all the "failing" test for now
by changing them as following.
======== Original tools/h5ls/testh5ls.sh =========
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Change to ===============================
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
==================================================
======== Original tools/h5copy/testh5copy.sh =========
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
H5LSTEST $FILEOUT
======== Change to ===============================
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
echo SKIP H5LSTEST $FILEOUT
==================================================
(AKC - 2008/11/10)
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message, "yod allocation delayed for node recovery". This interferes with
test suites that do not expect to see this message. See the section of "Red
Storm" in file INSTALL_parallel for a way to deal with this problem.
(AKC - 2008/05/28)
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use",
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
(AKC - 2004/12/08)
%%%%1.8.8%%%%
HDF5 version 1.8.8 released on 2011-11-15
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.7 and
HDF5 1.8.8, and contains information on the platforms tested and
known problems in HDF5-1.8.8.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.8 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.8 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.8 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.8 (current
release) versus Release 1.8.7":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.7
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- Added the --enable-fortran2003 flag to enable Fortran2003 support
in the HDF5 Fortran library. The flag should be used along with the
--enable-fortran flag and takes affect only when the Fortran compiler
is Fortran2003 compliant. (EIP - 2011/11/14)
- Added checks for clock_gettime and mach/mach_time.h to both configure and
CMake. This will support the move from gettimeofday to clock_gettime's
monotonic timer in the profiling code in a future release.
(DER - 2011/10/12)
Library
-------
- The Windows VFD code has been removed with the exception of the functions
which set it (H5Pset_fapl_windows, for example). Setting the Windows
VFD now really sets the SEC2 VFD. The WINDOWS_MAX_BUF and
WINDOWS_USE_STDIO configuration options and #defines have also been
removed. NOTE: Since the Windows VFD was a clone of the SEC2 VFD, this
change should be transparent to users.
(DER - 2011/10/12 - HDFFV-7740, HDFFV-7744)
- H5Tcreate now supports the string type (fixed-length and variable-
length). (SLU - 2011/05/20)
Parallel Library
----------------
- Added new H5Pget_mpio_actual_chunk_opt_mode and
H5Pget_mpio_actual_io_mode API routines for querying whether/how
a collective I/O operation completed. (QAK - 2011/10/12)
Tools
-----
- None
High-Level APIs
---------------
- Added the following Fortran wrappers for the Dimension Scale APIs:
h5dsset_scale_f
h5dsattach_scale_f
h5dsdetach_scale_f
h5dsis_attached_f
h5dsis_scale_f
h5dsset_label_f
h5dsget_label_f
h5dsget_scale_name_f
h5dsget_num_scales_f
(EIP for SB - 2011/10/13 - HDFFV-3797)
Fortran API
-----------
- The HDF5 Fortran library was enhanced to support the Fortran 2003 standard.
The following features are available when the HDF5 library is configured
using the --enable-fortran and --enable-fortran2003 configure flags AND
if the Fortran compiler is Fortran 2003 compliant:
- Subroutines overloaded with the C_PTR derived type:
h5pget_f
h5pget_fill_value_f
h5pinsert_f
h5pregister_f
h5pset_f
h5pset_fill_value_f
h5rcreate_f
h5rderefrence_f
h5rget_name_f
h5rget_obj_type_f
- Subroutines overloaded with the C_PTR derived type
and simplified signatures:
h5aread_f
h5awrite_f
h5dread_f
h5dwrite_f
- New subroutines
h5dvlen_reclaim_f
h5literate_by_name_f
h5literate_f
h5ovisit_f
h5tconvert_f
h5pset_nbit_f
h5pset_scaleoffset_f
- Subroutines with additional optional parameters:
h5pcreate_class_f
(EIP - 2011/10/14)
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- None
Bug Fixes since HDF5-1.8.7
==========================
Configuration
-------------
- Changed the size of H5_SIZEOF_OFF_T to 4 bytes (was 8) in the VMS
h5pubconf.h based on the output of a test program. (DER - 2011/10/12)
- The Windows and VMS versions of H5pubconf.h were brought into sync with
the linux/posix version. (DER - 2011/10/12)
- Fixed a bug in the bin/trace Perl script where API functions
that take a variable number of arguments were not processed for
trace statement fixup. (DER - 2011/08/25)
- The --enable-h5dump-packed-bits configure option has been removed.
The h5dump code that this option conditionally enabled is now always
compiled into h5dump. Please refer to the h5dump reference manual for
usage of the packed bits feature. (MAM - 2011/06/23 - HDFFV-7592)
- Configure now uses the same flags and symbols in its tests that are
used to build the library. (DER - 2011/05/24)
Library
-------
- Corrected the error when copying attributes between files which are using
different versions of the file format. (QAK - 2011/10/20 - HDFFV-7718)
- Corrected the error when loading local heaps from the file, which could
cause the size of the local heap's data block to increase dramatically.
(QAK - 2011/10/14 - HDFFV-7767)
- An application does not need to do H5O_move_msgs_forward() when writing
attributes. Tests were checked into the performance suite.
(VC - 2011/10/13 - HDFFV-7640)
- Fixed a bug that occurred when using H5Ocopy on a committed datatype
containing an attribute using that committed datatype.
(NAF - 2011/10/13 - HDFFV-5854)
- Added generic VFD I/O types to the SEC2 and log VFDs to ensure correct
I/O sizes (and remove compiler warnings) between Windows and true POSIX
systems. (DER - 2011/10/12)
- Corrected some Windows behavior in the SEC2 and log VFDs. This mainly
involved datatype correctness fixes, Windows API call error checks,
and adding the volume serial number to the VFD cmp functions.
(DER - 2011/10/12)
- Converted post-checks for the appropriate POSIX I/O sizes to pre-checks
in order to avoid platform-specific or undefined behavior.
(DER - 2011/10/12)
- #ifdef _WIN32 instances have been changed to #ifdef H5_HAVE_WIN32_API.
H5_HAVE_VISUAL_STUDIO checks have been added where necessary. This is in
CMake only as configure never sets _WIN32. (ADB - 2011/09/12)
- CLANG compiler with the options -fcatch-undefined-behavior and -ftrapv
discovered 3 problems in tests and tools' library:
1. In dsets.c, left shifting an unsigned int for 32 bits or more
caused undefined behavior.
2. In dt_arith.c, the INIT_INTEGER macro definition has an overflow
when the value is a negative minimal and is being subtracted from one.
3. In tools/lib/h5tools_str.c, right shifting an int value for 32 bits
or more caused undefined behavior.
All the problems have been corrected. (SLU - 2011/09/02 - HDFFV-7674)
- H5Epush2() now has the correct trace functionality (this is related to the
bin/trace Perl script bug noted in the configure section).
(DER - 2011/08/25)
- Corrected mismatched function name typo of h5pget_dxpl_mpio_c and
h5pfill_value_defined_c. (AKC - 2011/08/22 - HDFFV-7641)
- Corrected an internal error in the library where objects that use committed
(named) datatypes and were accessed from two different file IDs could confuse
the two and cause erroneous failures. (QAK - 2011/07/18 - HDFFV-7638)
- In v1.6 of the library, there was an EOA for the whole MULTI file saved in the
super block. We took it out in v1.8 of the library because it's meaningless
for the MULTI file. v1.8 of the library saves the EOA for the metadata file
instead, but this caused a backward compatibility problem.
A v1.8 library couldn't open the file created with the v1.6 library. We
fixed the problem by checking the EOA value to detect the file
created with v1.6 library. (SLU - 2011/06/22)
- When a dataset had filters and reading data failed, the error message
didn't say which filter wasn't registered. It's fixed now. (SLU - 2011/06/03)
Parallel Library
----------------
- The Special Collective IO (IO when some processes do not contribute to the
IO) and the Complex Derived Datatype MPI functionalities are no longer
conditionally enabled in the library by configure. They are always
enabled in order to take advantage of performance boosts from these
behaviors. Older MPI implementations that do not allow for these
functionalities can no longer by used by HDF5.
(MAM - 2011/07/08 - HDFFV-7639).
Tools
-----
- h5diff: fixed segfault over non-comparable attribute with different
dimention or rank, along with '-c' option to display details.
(JKM - 2011/10/24 - HDFFV-7770)
- Fixed h5diff to display all the comparable objects and attributes
regardless of detecting non-comparables. (JKM - 2011/09/16 - HDFFV-7693)
- Fixed h5repack to update the values of references(object and region) of
attributes in h5repack for 1) references, 2) arrays of references,
3) variable-length references, and 4) compound references.
(PC - 2011/09/14 - HDFFV-5932)
- h5diff: fixed a segfault over a dataset with container types
array and variable-length (vlen) along with multiple nested compound types.
Example: compound->array->compound, compound->vlen->compound.
(JKM - 2011/09/01 - HDFFV-7712)
- h5repack: added macro to handle a failure in H5Dread/write when memory
allocation failed inside the library. (PC - 2011/08/19)
- Fixed h5jam to not to allow the specifying of an HDF5 formatted file as
an input file for the -u (user block file) option. The original HDF5 file
would not be accessible if this behavior was allowed.
(JKM - 2011/08/19 - HDFFV-5941)
- Revised the command help pages of h5jam and h5unjam. The descriptions
were not up to date and some were missing.
(JKM - 2011/08/15 - HDFFV-7515)
- Fixed h5dump to correct the schema location:
<hdf5:HDF5-File
xmlns:hdf5="http://hdfgroup.org/HDF5/XML/schema/HDF5-File"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://hdfgroup.org/HDF5/XML/schema/HDF5-File
http://www.hdfgroup.org/HDF5/XML/schema/HDF5-File.xsd">
(ADB - 2011/08/10)
- h5repack: h5repack failed to copy a dataset if the layout is changed
from chunked with unlimited dimensions to contiguous.
(PC - 2011/07/15 - HDFFV-7649)
- Fixed h5diff: the "--delta" option considers two NaN of the same type
are different. This is wrong based on the h5diff description in the
Reference Manual. (PC - 2011/07/15 - HDFFV-7656)
- Fixed h5diff to display an instructive error message and exit with
an instructive error message when mutually exclusive options
(-d, -p and --use-system-epsilon) are used together.
(JKM - 2011/07/07 - HDFFV-7600)
- Fixed h5dump so that it displays the first line of each element in correct
position for multiple dimention array types. Before this fix,
the first line of each element in an array was
displayed after the last line of previous element without
moving to the next line (+indentation).
(JKM - 2011/06/15 - HDFFV-5878)
- Fixed h5dump so that it will display the correct value for
H5T_STD_I8LE datasets on the Blue-gene system (ppc64, linux, Big-Endian,
clustering). (AKC & JKM - 2011/05/12 - HDFFV-7594)
- Fixed h5diff to compare a file to itself correctly. Previously h5diff
reported either the files were different or not compatible in certain
cases even when comparing a file to itself. This fix also improves
performance when comparing the same target objects through verifying
the object and file addresses before comparing the details
in the objects. Examples of details are datasets and attributes.
(XCAO & JKM - 2011/05/06 - HDFFV-5928)
F90 API
-------
- Modified the h5open_f and h5close_f subroutines to not to call H5open
and H5close correspondingly. While the H5open call just adds overhead,
the H5close call called by a Fortran application shuts down the HDF5
library. This makes the library inaccessible to the application.
(EIP & SB - 2011/10/13 - HDFFV-915)
- Fixed h5tget_tag_f where the length of the C string was used to
repack the C string into the Fortran string. This lead to memory
corruption in the calling program. (SB - 2011/07/26)
- Added defined constants:
H5T_ORDER_MIXED_F (HDFFV-2767)
H5Z_SO_FLOAT_DSCALE_F
H5Z_SO_FLOAT_ESCALE_F
H5Z_SO_INT_F
H5Z_SO_INT_MINBITS_DEFAULT_F
H5O_TYPE_UNKNOWN_F
H5O_TYPE_GROUP_F
H5O_TYPE_DATASET_F
H5O_TYPE_NAMED_DATATYPE_F
H5O_TYPE_NTYPES_F
C++ API
------
- None
High-Level APIs:
------
- Fixed the H5LTdtype_to_text function. It had some memory problems when
dealing with some complicated data types. (SLU - 2011/10/19 - HDFFV-7701)
- Fixed H5DSset_label seg faulting when retrieving the length of a
dimension label that was not set. (SB - 2011/08/07 - HDFFV-7673)
- Fixed a dimension scale bug where if you create a dimscale, attach two
datasets to it, and then unattach them, you get an error if they are
unattached in order, but no error if you unattach them in reverse order.
(SB - 2011/06/07 - HDFFV-7605)
Fortran High-Level APIs:
------
- None
Platforms Tested
================
The following platforms and compilers have been tested for this release.
AIX 5.3 xlc 10.1.0.5
(NASA G-ADA) xlC 10.1.0.5
xlf90 12.1.0.6
FreeBSD 8.2-STABLE i386 gcc 4.2.1 [FreeBSD] 20070719
(loyalty) g++ 4.2.1 [FreeBSD] 20070719
gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
FreeBSD 8.2-STABLE amd64 gcc 4.2.1 [FreeBSD] 20070719
(freedom) g++ 4.2.1 [FreeBSD] 20070719
gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
IBM Blue Gene/P bgxlc 9.0.0.9
(LLNL uDawn) bgxlf90 11.1.0.7
bgxlC 9.0.0.9
Linux 2.6.16.60-0.54.5-smp Intel(R) C, C++, Fortran Compilers
x86_64 Version 11.1 20090630
(INL Icestorm)
Linux 2.6.18-194.el5 x86_64 Intel(R) C, C++, Fortran Compilers
(INL Fission) Version 12.0.2 20110112
Linux 2.6.18-108chaos x86_64 Intel(R) C, C++, Fortran Compilers
(LLNL Aztec) Version 11.1 20090630
Linux 2.6.18-194.3.1.el5PAE gcc (GCC) 4.1.2 and 4.4.2
#1 SMP i686 i686 i386 GNU Fortran (GCC) 4.1.2 20080704
(jam) (Red Hat 4.1.2-48) and 4.4.2
PGI C, Fortran, C++ 10.4-0 32-bit
PGI C, Fortran, C++ 10.6-0 32-bit
Intel(R) C Compiler for 32-bit
applications, Version 11.1
Intel(R) C++ Compiler for 32-bit
applications, Version 11.1
Intel(R) Fortran Compiler for 32-bit
applications, Version 11.1
MPICH mpich2-1.3.1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.18-238.12.1.el5 gcc 4.1.2 and 4.4.2
#1 SMP x86_64 GNU/Linux GNU Fortran (GCC) 4.1.2 20080704
(koala) (Red Hat 4.1.2-46) and 4.4.2
tested for both 32- and 64-bit binaries
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64,
Version 11.1.
PGI C, Fortran, C++ Version 9.0-4
for 64-bit target on x86-64
MPICH mpich2-1.3.1 compiled with
gcc 4.1.2 and gfortran 4.1.2
SGI Altix UV Intel(R) C, Fortran Compilers
SGI ProPack 7 Linux Version 11.1 20100806
2.6.32.24-0.2.1.2230.2.PTF- SGI MPT 2.02
default #1 SMP
(NCSA ember)
Dell NVIDIA Cluster Intel(R) C, Fortran Compilers
Red Hat Enterprise Linux 6 Version 12.0.4 20110427
2.6.32-131.4.1.el6.x86_64 mvapich2 1.7rc1-intel-12.0.4
(NCSA forge)
SunOS 5.10 32- and 64-bit Sun C 5.11 SunOS_sparc 2010/08/13
Sun Fortran 95 8.5 SunOS_sparc 2010/08/13
Sun C++ 5.11 SunOS_sparc 2010/08/13
Windows XP Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 (cmake)
Cygwin(1.7.9 native gcc(4.5.3) compiler and gfortran)
Windows XP x64 Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 (cmake)
Cygwin(1.7.9 native gcc(4.5.3) compiler and gfortran)
Windows Vista Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows Vista x64 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows 7 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Mac OS X 10.8.0 (Intel 64-bit) i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5666) (dot 3)
Darwin Kernel Version 10.8.0 GNU Fortran (GCC) 4.6.1
Intel C, C++ and Fortran compilers 12.1.0
Mac OS X 10.8.0 (Intel 32-bit) i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5666) (dot 3)
Darwin Kernel Version 10.8.0 GNU Fortran (GCC) version 4.6.1
Intel C, C++ and Fortran compilers 12.1.0
Fedora 12 2.6.32.16-150.fc12.ppc64 #1 SMP ppc64 GNU/Linux
gcc (GCC) 4.4.4 20100630 (Red Hat 4.4.4-10)
GNU Fortran (GCC) 4.4.4 20100630 (Red Hat 4.4.4-10)
Debian6.0.3 2.6.32-5-686 #1 SMP i686 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
Debian6.0.3 2.6.32-5-amd64 #1 SMP x86_64 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
Fedora15 2.6.40.6-0.fc15.i686.PAE #1 SMP i686 i686 i386 GNU/Linux
gcc (GCC) 4.6.1 20110908 (Red Hat 4.6.1-9)
GNU Fortran (GCC) 4.6.1 20110908 (Red Hat 4.6.1-9)
Fedora15 2.6.40.6-0.fc15.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.6.1 20110908 (Red Hat 4.6.1-9)
GNU Fortran (GCC) 4.6.1 20110908 (Red Hat 4.6.1-9)
SUSE 11.4 2.6.37.6-0.7-desktop #1 SMP PREEMPT i686 i686 i386 GNU/Linux
gcc (SUSE Linux) 4.5.1 20101208
GNU Fortran (SUSE Linux) 4.5.1 20101208
SUSE 11.4 2.6.37.6-0.7-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.5.1 20101208
GNU Fortran (SUSE Linux) 4.5.1 20101208
Ubuntu 11.10 3.0.0-12-generic #20-Ubuntu SMP i686 GNU/Linux
gcc (Ubuntu/Linaro 4.6.1-9ubuntu3) 4.6.1
GNU Fortran (Ubuntu/Linaro 4.6.4-9ubuntu3) 4.6.1
Ubuntu 11.10 3.0.0-12-generic #20-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.6.1-9ubuntu3) 4.6.1
GNU Fortran (Ubuntu/Linaro 4.6.1-9ubuntu3) 4.6.1
OpenVMS Alpha 8.3 HP C V7.3-009
HP Fortran V8.2-104679-48H9K
HP C++ V7.3-009
Cray Linux Environment (CLE) PrgEnv-pgi 2.2.74
hopper.nersc.gov pgcc 11.7-0 64-bit target on x86-64 Linux -tp k8e
franklin.nersc.gov pgf90 11.7-0 64-bit target on x86-64 Linux -tp k8e
pgCC 11.7-0 64-bit target on x86-64 Linux -tp k8e
Tested Configuration Features Summary
=====================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
Solaris2.10 32-bit n y n y y y
Solaris2.10 64-bit n y n y y y
Windows XP n y(4) n y y y
Windows XP x64 n y(4) n y y y
Windows Vista n y(4) n y y y
Windows Vista x64 n y(4) n y y y
OpenVMS Alpha n y n y y n
Mac OS X 10.8 Intel 32-bit n y n y y y
Mac OS X 10.8 Intel 64-bit n y n y y y
AIX 5.3 32- and 64-bit n y n y y y
FreeBSD 8.2-STABLE 32&64 bit n x n x y y
CentOS 5.5 Linux 2.6.18-194 i686 GNU (1)W y y(2) y y y y
CentOS 5.5 Linux 2.6.18-194 i686 Intel W n y n y y n
CentOS 5.5 Linux 2.6.18-194 i686 PGI W n y n y y n
CentOS 5.5 Linux 2.6.16 x86_64 GNU (1) W y y(3) y y y y
CentOS 5.5 Linux 2.6.16 x86_64 Intel W n y n y y n
CentOS 5.5 Linux 2.6.16 x86_64 PGI W n y n y y y
Fedora 12 Linux 2.6.32.16-150.fc12.ppc64 n y n y y y
SGI ProPack 7 Linux 2.6.32.24 y y y y y y
Red Hat Enterprise Linux 6 y y y y y y
CLE hopper.nersc.gov y y(3) y y y n
CLE franklin.nersc.gov y y(3) y y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit y y y y
Windows XP y y(4) y n
Windows XP x64 y y(4) y n
Windows Vista y y(4) y y
Windows Vista x64 y y(4) y y
OpenVMS Alpha n n n n
Mac OS X 10.8 Intel 32-bit y(5) n y n
Mac OS X 10.8 Intel 64-bit y(5) n y n
AIX 5.3 32- and 64-bit n n n y
FreeBSD 8.2-STABLE 32&64 bit y x x y
CentOS 5.5 Linux 2.6.18-128 i686 GNU (1)W y y(2) y y
CentOS 5.5 Linux 2.6.18-128 i686 Intel W y y y n
CentOS 5.5 Linux 2.6.18-128 i686 PGI W y y y n
CentOS 5.5 Linux 2.6.16 x86_64 GNU (1) W y y y y
CentOS 5.5 Linux 2.6.16 x86_64 Intel W y y y n
CentOS 5.5 Linux 2.6.16 x86_64 PGI W y y y n
Fedora 12 Linux 2.6.32.16-150.fc12.ppc64 y y y y
SGI ProPack 7 Linux 2.6.32.24 y y y n
Red Hat Enterprise Linux 6 y y y n
CLE hopper.nersc.gov n n n n
CLE franklin.nersc.gov n n n n
(1) Fortran compiled with gfortran.
(2) With PGI and Absoft compilers.
(3) With PGI compiler for Fortran.
(4) Using Visual Studio 2008 w/ Intel Fortran 10.1 (Cygwin shared libraries are not supported)
(5) C and C++ shared libraries will not be built when Fortran is enabled.
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* The STDIO VFD does not work on some architectures, possibly due to 32/64
bit or large file issues. The basic STDIO VFD test is known to fail on
64-bit SunOS 5.10 on SPARC when built with -m64 and 32-bit OS X/Darwin
10.7.0. The STDIO VFD test has been disabled while we investigate and
a fix should appear in a future release, possibly 1.8.9.
(DER - 2011/10/14)
* h5diff can report inconsistent results when comparing datasets of enum type
that contain invalid values. This is due to how enum types are handled in
the library and will be addressed in the next release.
(DER - 2011/10/14 - HDFFV-7527)
* The links test can fail under the stdio VFD due to some issues with external
links. This will be investigated and fixed in a future release.
(DER - 2011/10/14 - HDFFV-7768)
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 - HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value. (AKC - 2011/05/07)
* The C++ and FORTRAN bindings are not currently working on FreeBSD.
(QAK - 2011/04/26)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however. (MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2. (AKC - 2011/03/10)
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
(NAF - 2011/01/19)
* The library's test dt_arith.c showed a compiler's rounding problem on
Cygwin when converting from unsigned long long to long double. The
library's own conversion works fine. We defined a macro for Cygwin to
skip this test until we can solve the problem.
(SLU - 2010/05/05 - HDFFV-1264)
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
(SLU - 2010/02/02)
* MinGW has a missing libstdc++.dll.a library file and will not successfully link
C++ applications/tests. Do not use the enable-cxx configure option. Read all of
the INSTALL_MINGW.txt file for all restrictions. (ADB - 2009/11/11)
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
(MAM - 2009/11/04)
* The PathScale MPI implementation, accessing a Panasas file system, would
cause H5Fcreate() with H5F_ACC_EXCL to fail even when the file does not
exist. This is due to the MPI_File_open() call failing if the mode has
the MPI_MODE_EXCL bit set. (AKC - 2009/08/11 - HDFFV-988)
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO. (CMC - 2009/04/28)
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh and
tools/h5copy/testh5copy.sh will fail some of its sub-tests. These sub-tests
are expected to fail and should exit with a non-zero code but the yod
command does not propagate the exit code of the executables. Yod always
returns 0 if it can launch the executable. The test suite shell expects
a non-zero for this particular test, therefore it concludes the test has
failed when it receives 0 from yod. Skip all the "failing" test for now
by changing them as following.
======== Original tools/h5ls/testh5ls.sh =========
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Change to ===============================
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
==================================================
======== Original tools/h5copy/testh5copy.sh =========
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
H5LSTEST $FILEOUT
======== Change to ===============================
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
echo SKIP H5LSTEST $FILEOUT
==================================================
(AKC - 2008/11/10)
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message, "yod allocation delayed for node recovery". This interferes with
test suites that do not expect to see this message. See the section of "Red
Storm" in file INSTALL_parallel for a way to deal with this problem.
(AKC - 2008/05/28)
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
(SLU - 2005/06/30)
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use",
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
(AKC - 2004/12/08)
%%%%1.8.7%%%%
HDF5 version 1.8.7 released on Tue May 10 09:24:44 CDT 2011
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.6 and
HDF5 1.8.7, and contains information on the platforms tested and
known problems in HDF5-1.8.7.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.7 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.7 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.7 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.7 (current
release) versus Release 1.8.6":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.6
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- Configure now generates Makefiles that build in "silent make mode"
by default in which compile and link lines are significantly
simplified for clarity. To override this and view actual compile and
link lines during building, the --disable-silent-rules flag can be used
at configure time, or the 'make' command can be followed by V=1, to
indicate a "verbose" make. (MAM - 2011/4/14).
- Added mpicc and mpif90 as the default C and Fortran compilers for Linux
systems when --enable-parallel is specified but no $CC or $FC is defined.
(AKC - 2011/2/7)
- Added a new configure option, "--enable-unsupported", which can
be used to stop configure from preventing the use of unsupported
configure option combinations, such as c++ in parallel or Fortran
with threadsafe. Use at your own risk, as it may result in a
library that won't compile or run as expected!
(MAM - 2010/11/17 - Bug 2061)
Library
-------
- The library allows the dimension size of a dataspace to be zero. In
the past, the library would allow this only if the maximal dimension
size was unlimited. Now there is no such restriction, but no data
can be written to this kind of dataset. (SLU - 2011/4/20)
- We added two new macros, H5_VERSION_GE and H5_VERSION_LE, to let users
compare certain version numbers with the library being used. (SLU -
2011/4/20)
- Added ability to cache files opened through external links. Added new
public functions H5Pset_elink_file_cache_size(),
H5Pget_elink_file_cache_size(), and H5Fclear_elink_file_cache().
(NAF - 2011/02/17)
- Finished implementing all options for 'log' VFD. (QAK - 2011/1/25)
- Removed all old code for Metrowerks compilers, bracketed by
__MWERKS__). Metrowerks compiler is long gone. (AKC - 2010/11/17)
Parallel Library
----------------
- None
Tools
-----
- h5diff: Added new "verbose with levels" option, '-vN, --verbose=N'.
The old '-v, --verbose' option is deprecated but remains available;
it is exactly equivalent to '-v0, --verbose=0'.
The new levels 1 ('-v1' or '--verbose=1') and 2 ('-v2' or
'--verbose=2') can be specified to view more information regarding
attributes differences. Bug #2121 (JKM 2011/3/23)
- h5dump: Added new option --enable-error-stack. This option will
display error stack information in the output stream. This is
useful when the "h5dump: Unable to print data" message is output.
(ADB - 2011/03/03)
High-Level APIs
---------------
- Fortran LT make datasets routines (H5LTmake_dataset_f,
h5ltmake_dataset_int_f, h5ltmake_dataset_float_f, h5ltmake_dataset_double_f)
and LT read datasets routines (h5ltread_dataset_f,h5ltread_dataset_int_f,
h5ltread_dataset_float_f, 5ltread_dataset_double_f) can now handle
4-dimensional to 7-dimensional rank datasets. HDFFV-1217 (MSB-2011/4/24/2011)
F90 API
-------
- None
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- Intel V11.1 uses now -O3 optimization in production mode (EIP - 2010/10/08)
Bug Fixes since HDF5-1.8.6
==========================
Configuration
-------------
- Shared C++ and HL libraries on AIX should now be working correctly.
Note that Fortran shared libraries are still not working on AIX.
(See the Known Problems section, below). (MAM - 2011/4/20)
- Removed config/ibm-aix6.x. All IBM-AIX settings are in one file,
ibm-aix. (AKC - 2011/4/14)
- Shared C libraries are no longer disabled on Mac when Fortran
is enabled. Shared Fortran libraries are still not supported on Mac,
so configure will disable them by default, but this is overrideable
with the new --enable-unsupported configure option. The configure
summary has been updated to reflect the fact that the shared-ness of
the C++/Fortran wrapper libraries may not align with the C library.
(MAM - 2011/04/11 - HDFFV-4353).
Library
-------
- Changed assertion failure when decoding a compound datatype with no
fields into a normal error failure. Also prohibit using this sort
of datatype for creating an attribute (as is already the case for
datasets and committed (named) datatypes). (QAK - 2011/04/15, Jira
issue #HDFFV-2766)
- Tell the VFL flush call that the file will be closing, allowing
the VFDs to avoid sync'ing the file (particularly valuable in parallel).
(QAK - 2011/03/09)
- The datatype handler created with H5Tencode/decode used to have the
reference count 0 (zero); it now has the reference count 1 (one).
(SLU - 2011/2/18)
- Fixed the definition of H5_HAVE_GETTIMEOFDAY on Windows so that
HDgettimeofday() is defined and works properly. Bug HDFFV-5931
(DER - 2011/04/14)
- Added basic VFD tests for the Windows, STDIO and log VFD tests.
(DER - 2011/04/11)
Parallel Library
----------------
- None
Tools
-----
- Updated h5dump test case script to prevent entire test failure when
source directory is read-only. Bug #HDFFV-4342 (JKM 2011/4/12)
- Fixed h5dump displaying incorrect values for H5T_STD_I8BE type data in
attribute on Big-Endian machine. H5T_STD_I8BE is unsigned 8bit type,
so h5dump is supposed to display -2 instead of 254. It worked correctly
on Little-Endian system , but not on Big-Endian system. Bug #HDFFV-4358
(JKM 04/08/2011)
- Updated some HDF5 tools to standardize the option name as
'--enable-error-stack' for printing HDF5 error stack messages. h5ls and
h5dump have been updated. For h5ls, this replaces "-e/--errors" option,
which is deprecated. For h5dump, this is a new option. Bug #2182
(JKM 2011/3/30)
- Fixed the h5diff --use-system-epsilon option. The formula used in the
calculation was changed from ( |a - b| / b ) to ( |a - b| ).
This was done to improve performance. Bug #2184 (JKM 2011/3/24)
- Fixed output for H5T_REFERENCE in h5dump. According to the BNF document
the output of a H5T_REFERENCE should be followed by the type;
<reference> ::= H5T_REFERENCE { <ref_type> }
<ref_type> ::= H5T_STD_REF_OBJECT | H5T_STD_REF_DSETREG
Previously this was only displayed if the -R option was used.
Bug #1725 (ADB 2011/3/28)
- Fixed two h5diff issues. 1) h5diff compared attributes correctly only
when two objects had the same number of attributes and the attribute
names were identical. 2) h5diff did not display useful information about
attribute differences. Bug #2121 (JKM 2011/3/17)
- Fixed a memory leak in h5diff that occurred when accessing symbolic links
with the --follow-symlink option. Bug #2214 (JKM 2011/3/18)
- Fixed a memory leak in h5diff that occurred when accessing variable length
string data. Bug #2216 (JKM 2011/3/18)
- Fixed and improved the help page for h5ls -a, --address option.
Bug #1904 (JKM 2011/3/11)
- Fixed h5copy to enable copying an object into the same HDF5 file.
Previously h5copy displayed an error message when the target file
was the same as the source file. (XCAO 2011/3/8)
- Fixed an h5dump problem that caused the tool to skip some data elements
in large datasets with a large array datatype on Windows. This issue
arose only on Windows due to the different return behavior of the
_vsnprintf() function. Bug #2161 (JKM 2011/3/3)
- Fixed h5dump which was skipping some array indices in large datasets
with a relatively large array datatype. The interval of skipped indices
varied according to the size of the array. Bug #2092 (JKM 2011/2/15)
- Fixed h5diff which was segfaulting when comparing compound datasets
with a combination of fixed-length string datatypes and variable-length
string datatypes in certain orders. Bug #2089 (JKM 2010/12/28)
- Improved h5diff performance. 1) Now use HDmemcmp() before comparing two
elements. 2) Replace expensive H5Tequals() calls. 3) Retrieve datatype
information at dataset level, not at each element level for compound
datasets. HDFFV-7516 (JKM 2011/4/18)
- Fixed h5ls to display nested compound types with curly brackets
when -S (--simple) option is used with -l (--label), so it shows
which members (in curly brackets) belong to which nested compound type,
making the output clearer. Bug #1979 (JKM 2010/11/09)
- Fixed h5diff to handle variable-length strings in a compound dataset
and variable-length string arrays in a compound dataset correctly.
Garbage values were previously displayed when h5diff compared multiple
variable-length strings in a compound type dataset.
Bug #1989 (JKM 2010/10/28)
- Fixed h5copy to fail gracefully when copying an object to a non-
existing group without the -p option. Bug #2040 (JKM 2010/10/18)
F90 API
------
- None
C++ API
------
- None
High-Level APIs:
------
- None
Fortran High-Level APIs:
------
- h5tbmake_table_f: Fixed error in passing an array of characters with different
length field names.
- h5tget_field_info_f: Fixed error with packing the C strings into a Fortran
array of strings. Added optional argument called 'maxlen_out' which returns
the maximum string character length in a field name element.
Bug HDFFV-1255 (MSB- 4/17/2011)
Platforms Tested
================
The following platforms and compilers have been tested for this release.
AIX 6.1 xlc 11.1.0.3
(NCSA BP) xlC 11.1.0.3
xlf90 13.1.0.3
mpcc_r 11.1.0.3
mpxlf90_r 13.1.0.3
FreeBSD 8.2-STABLE i386 gcc 4.2.1 [FreeBSD] 20070719
(loyalty) g++ 4.2.1 [FreeBSD] 20070719
gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
FreeBSD 8.2-STABLE amd64 gcc 4.2.1 [FreeBSD] 20070719
(freedom) g++ 4.2.1 [FreeBSD] 20070719
gcc 4.6.1 20110422
g++ 4.6.1 20110422
gfortran 4.6.1 20110422
Linux 2.6.18-194.3.1.el5PAE gcc (GCC) 4.1.2 and 4.4.2
#1 SMP i686 i686 i386 G95 (GCC 4.0.3 (g95 0.93!) Apr 21 2010)
(jam) GNU Fortran (GCC) 4.1.2 20080704
(Red Hat 4.1.2-48) and 4.4.2
PGI C, Fortran, C++ 10.4-0 32-bit
PGI C, Fortran, C++ 10.6-0 32-bit
Intel(R) C Compiler for 32-bit
applications, Version 11.1
Intel(R) C++ Compiler for 32-bit
applications, Version 11.1
Intel(R) Fortran Compiler for 32-bit
applications, Version 11.1
Absoft 32-bit Fortran 95 10.0.7
MPICH mpich2-1.3.1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.18-194.17.1.el5 gcc 4.1.2 and 4.4.2
#1 SMP x86_64 GNU/Linux G95 (GCC 4.0.3 (g95 0.93!) Apr 21 2010)
(amani) tested for both 32- and 64-bit binaries
GNU Fortran (GCC) 4.1.2 20080704
(Red Hat 4.1.2-46) and 4.4.2
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64,
Version 11.1.
PGI C, Fortran, C++ Version 9.0-4
for 64-bit target on x86-64
MPICH mpich2-1.3.1 compiled with
gcc 4.1.2 and gfortran 4.1.2
SGI ProPack 7 Linux Intel(R) C++ Version 11.1 20100806
2.6.32.24-0.2.1.2230.2.PTF- Intel(R) Fortran Version 11.1 20100806
default #1 SMP SGI MPT 2.01
SGI Altix UV
(NCSA ember)
SunOS 5.10 32- and 64-bit Sun C 5.9 Sun OS_sparc Patch 124867-16
(linew) Sun Fortran 95 8.3 Sun OS_sparc Patch 127000-13
Sun C++ 5.9 Sun OS_sparc Patch 124863-26
Sun C 5.10 SunOS_sparc Patch 141861-07
Sun Fortran 95 8.4 SunOS_sparc Patch 128231-06
Sun C++ 5.10 SunOS_sparc 128228-11
Intel Xeon Linux 2.6.18- gcc 4.2.4
92.1.10.el5_lustre.1.6.6smp- Intel(R) C++ Version 10.1.017
perfctr #8 SMP Intel(R) Fortran Compiler Version 10.1.017
(NCSA abe) Open MPI 1.3.2
MVAPICH2-1.5.1_pgi-10.8
Windows XP Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 (cmake)
Cygwin(1.7.7 native gcc(4.3.4) compiler and gfortran)
Windows XP x64 Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 (cmake)
Cygwin(1.7.7 native gcc(4.3.4) compiler and gfortran)
Windows Vista Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows Vista x64 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows 7 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Mac OS X 10.7.0 (Intel 64-bit) i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1
Darwin Kernel Version 10.7.0 GNU Fortran (GCC) 4.6.0 20101106 (experimental)
Intel C, C++ and Fortran compilers 12.0.1.122 20101110
Mac OS X 10.7.0 (Intel 32-bit) i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5666) (dot 3)
Darwin Kernel Version 10.7.0 GNU Fortran (GCC) version 4.4.0 20090123 (experimental)
[trunk revision 143587]
Fedora 12 2.6.32.16-150.fc12.ppc64 #1 SMP ppc64 GNU/Linux
gcc (GCC) 4.4.4 20100630 (Red Hat 4.4.4-10)
GNU Fortran (GCC) 4.4.4 20100630 (Red Hat 4.4.4-10)
Debian6.01 2.6.32-5-686 #1 SMP i686 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
Debian6.01 2.6.32-5-amd64 #1 SMP x86_64 GNU/Linux
gcc (Debian 4.4.5-8) 4.4.5
GNU Fortran (Debian 4.4.5-8) 4.4.5
Fedora14 2.6.35.12-88.fc14.i686.PAE #1 SMP i686 i686 i386 GNU/Linux
gcc (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
GNU Fortran (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
Fedora14 2.6.35.12-88.fc14.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
GNU Fortran (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
SUSE 11.4 2.6.37.1-1.2-desktop #1 SMP PREEMPT i686 i686 i386 GNU/Linux
gcc (SUSE Linux) 4.5.1 20101208
GNU Fortran (SUSE Linux) 4.5.1 20101208
SUSE 11.4 2.6.37.1-1.2-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.5.1 20101208
GNU Fortran (SUSE Linux) 4.5.1 20101208
Ubuntu 10.10 2.6.35-28-generic #50-Ubuntu SMP i686 GNU/Linux
gcc (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
GNU Fortran (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
Ubuntu 10.10 2.6.35-28-generic #50-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
GNU Fortran (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
OpenVMS Alpha 8.3 HP C V7.3-009
HP Fortran V8.2-104679-48H9K
HP C++ V7.3-009
Tested Configuration Features Summary
========================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
Solaris2.10 32-bit n y n y y y
Solaris2.10 64-bit n y n y y y
Windows XP n y(4) n y y y
Windows XP x64 n y(4) n y y y
Windows Vista n y(4) n y y y
Windows Vista x64 n y(4) n y y y
OpenVMS Alpha n y n y y n
Mac OS X 10.7 Intel 32-bit n y n y y y
Mac OS X 10.7 Intel 64-bit n y n y y y
AIX 6.1 32- and 64-bit y y y y y y
FreeBSD 8.2-STABLE 32&64 bit n x n x y y
CentOS 5.5 Linux 2.6.18-194 i686 GNU (1)W y y(2) y y y y
CentOS 5.5 Linux 2.6.18-194 i686 Intel W n y n y y n
CentOS 5.5 Linux 2.6.18-194 i686 PGI W n y n y y n
CentOS 5.5 Linux 2.6.16 x86_64 GNU (1) W y y(3) y y y y
CentOS 5.5 Linux 2.6.16 x86_64 Intel W n y n y y n
CentOS 5.5 Linux 2.6.16 x86_64 PGI W n y n y y y
RedHat EL4 2.6.18 Xeon Lustre C y y y y y n
Fedora 12 Linux 2.6.32.16-150.fc12.ppc64 n y n y y y
SGI Linux 2.6.32.19 y y y y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit y y y y
Windows XP y y(4) y n
Windows XP x64 y y(4) y n
Windows Vista y y(4) y y
Windows Vista x64 y y(4) y y
OpenVMS Alpha n n n n
Mac OS X 10.7 Intel 32-bit y(5) n y n
Mac OS X 10.7 Intel 64-bit y(5) n y n
AIX 6.1 32- and 64-bit n n n y
FreeBSD 8.2-STABLE 32&64 bit y x x y
CentOS 5.5 Linux 2.6.18-128 i686 GNU (1)W y y(2) y y
CentOS 5.5 Linux 2.6.18-128 i686 Intel W y y y n
CentOS 5.5 Linux 2.6.18-128 i686 PGI W y y y n
CentOS 5.5 Linux 2.6.16 x86_64 GNU (1) W y y y y
CentOS 5.5 Linux 2.6.16 x86_64 Intel W y y y n
CentOS 5.5 Linux 2.6.16 x86_64 PGI W y y y n
RedHat EL4 2.6.18 Xeon Lustre C y y y n
Fedora 12 Linux 2.6.32.16-150.fc12.ppc64 y y y y
SGI Linux 2.6.32.19 y y y y
(1) Fortran compiled with gfortran.
(2) With PGI and Absoft compilers.
(3) With PGI compiler for Fortran.
(4) Using Visual Studio 2008 w/ Intel Fortran 10.1 (Cygwin shared libraries are not supported)
(5) C and C++ shared libraries will not be built when Fortran is enabled.
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* After the shared library support was fixed for some bugs, it was discovered
that "make prefix=XXX install" no longer works for shared libraries. It
still works correctly for static libraries. Therefore, if you want to
install the HDF5 shared libraries in a location such as /usr/local/hdf5,
you need to specify the location via the --prefix option during configure
time. E.g, ./configure --prefix=/usr/local/hdf5 ...
(AKC - 2011/05/07 HDFFV-7583)
* The parallel test, t_shapesame, in testpar/, may run for a long time and may
be terminated by the alarm signal. If that happens, one can increase the
alarm seconds (default is 1200 seconds = 20 minutes) by setting the
environment variable, $HDF5_ALARM_SECONDS, to a larger value such as 3600
(60 minutes). Note that the t_shapesame test may fail in some systems
(see the "While working on the 1.8.6 release..." problem below). If
it does, it will waste more time if $HDF5_ALARM_SECONDS is set
to a larger value. (AKC - 2011/05/07)
* The C++ and FORTRAN bindings are not currently working on FreeBSD.
(QAK - 2011/04/26)
* Shared Fortran libraries are not quite working on AIX. While they are
generated when --enable-shared is specified, the fortran and hl/fortran
tests fail. We are looking into the issue. HL and C++ shared libraries
should now be working as intended, however. (MAM - 2011/04/20)
* The --with-mpe configure option does not work with Mpich2. AKC - 2011/03/10
* If parallel gmake (e.g., gmake -j 4) is used, the "gmake clean" command
sometimes fails in the perform directory due to the attempt to remove the
executable of h5perf or h5perf_serial by two "parallel" commands. This error
has no consequence on the functionality of the HDF5 library or install. It
is fixed in the next release. AKC - 2011/01/25
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or file systems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
* The library's test dt_arith.c showed a compiler's rounding problem on
Cygwin when converting from unsigned long long to long double. The
library's own conversion works fine. We defined a macro for Cygwin to
skip this test until we can solve the problem. Please see bug #1813.
SLU - 2010/5/5
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added in. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug in 1.8 library. Because there's only one user
complaining about it, we (Elena, Quincey, and I) decided to leave it as
it is (see bug report #1279). Quincey will make a plan for 1.10.
SLU - 2010/2/2
* MinGW has a missing libstdc++.dll.a library file and will not successfully link
C++ applications/tests. Do not use the enable-cxx configure option. Read all of
the INSTALL_MINGW.txt file for all restrictions. ADB - 2009/11/11
* The PathScale MPI implementation, accessing a Panasas file system, would
cause H5Fcreate() with H5F_ACC_EXCL to fail even when the file does not
exist. This is due to the MPI_File_open() call failing if the mode has
the MPI_MODE_EXCL bit set. (See bug 1468 for details.) AKC - 2009/8/11
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO. CMC - 2009/04/28
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh and
tools/h5copy/testh5copy.sh will fail some of its sub-tests. These sub-tests
are expected to fail and should exit with a non-zero code but the yod
command does not propagate the exit code of the executables. Yod always
returns 0 if it can launch the executable. The test suite shell expects
a non-zero for this particular test, therefore it concludes the test has
failed when it receives 0 from yod. Skip all the "failing" test for now
by changing them as following.
======== Original tools/h5ls/testh5ls.sh =========
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Change to ===============================
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
==================================================
======== Original tools/h5copy/testh5copy.sh =========
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
H5LSTEST $FILEOUT
======== Change to ===============================
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
echo SKIP H5LSTEST $FILEOUT
==================================================
AKC - 2008/11/10
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message, "yod allocation delayed for node recovery". This interferes with
test suites that do not expect to see this message. See the section of "Red
Storm" in file INSTALL_parallel for a way to deal with this problem.
AKC - 2008/05/28
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
SLU - 2005/6/30
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use",
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
* There is also a configure error on Altix machines that incorrectly reports
when a version of Szip without an encoder is being used.
* On cobalt, an SGI Altix SMP ia64 system, Intel compiler version 10.1 (which
is the default on that system) does not work properly and results in
failures during make check (in a static build) and make installcheck (during
a shared build). This appears to be a compiler optimization problem.
Reducing optimization by setting CFLAGS to -O1 or below resolves the issue.
Alternatively, using a newer version of the compiler (11.0) also works as
intended. MAM - 2010/06/01
* h5diff will not report enum value differences when one or both of the values
is not a valid enumeration value. The source of this bug has been identified
and it will be fixed in 1.8.8. DER - 2011/04/27
%%%%1.8.6%%%%
HDF5 version 1.8.6 released on Mon Feb 14 10:26:30 CST 2011
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.5 and
HDF5 1.8.6, and contains information on the platforms tested and
known problems in HDF5-1.8.6.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.6 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.6 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.6 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.6 (current
release) versus Release 1.8.5":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.5
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- CMake: Improved CPack packaging, added parallel commands, improved
configuration options (better similarity to configure), added more
tests, better support for use in external cmake projects.
(ADB - 2010/10/07)
- The default configuration setting for official releases is
--enable-production. For unofficial releases, the default configuration
setting has been --disable-production. (AKC - 2010/05/28)
Library
-------
- Added support for thread safety on Windows using the Windows threads
library. Use the HDF5_ENABLE_THREADSAFE option in CMake on a Windows
platform to enable this functionality. This is supported on Windows
Vista and newer Windows operating systems. (MAM - 2010/09/10)
- H5Tset_order and H5Tget_order now support all datatypes. A new byte
order, H5T_ORDER_MIXED, has been added specifically for a compound
datatype and its derived type. (SLU - 2010/8/23)
- Improved performance of metadata I/O by changing the default algorithm
to perform I/O from all processes (instead of just process 0) when using
parallel I/O drivers. (QAK - 2010/07/19)
- Improved performance of I/O on datasets with the same shape, but
different rank. (QAK - 2010/07/19)
- Improved performance of the chunk cache by avoiding unnecessary b-tree
lookups of chunks already in cache. (NAF - 2010/06/15)
Parallel Library
----------------
- None
Tools
-----
- h5diff: Added a new flag: --exclude-path. The specified path to an
object will be excluded when comparing two files or two groups. If a
group is specified to be excluded, all member objects of that group
will be excluded. (JKM - 2010/09/16).
- h5ls: Added a new flag: --no-dangling-links. See --help output for
details. (JKM - 2010/06/15)
- h5ls: Added a new flag --follow-symlinks. See --help output for
details. (JKM - 2010/05/25)
High-Level APIs
---------------
- None
F90 API
-------
- None
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- Sun C and C++ 5.10 and Sun Fortran 95 8.4.
- Mac OS X 10.6.4 with gcc 4.2.1 and gfortran 4.6
Bug Fixes since HDF5-1.8.5
==========================
Configuration
-------------
- The default number of MPI processes for testing purposes has been
changed from 3 to 6. (AKC - 2010/11/11)
- Some tests in tools/h5repack may fail in AIX systems when -q32 mode is
used. The error is caused by not requesting enough memory in default.
Added "env LDR_CNTRL=MAXDATA=0x20000000@DSA" into the $RUNSERIAL and
$RUNPARALLE in the AIX config file so that executables are tested with
more memory. (AKC - 2010/11/11)
- Removed recognition of the parallel compilers of LAM(hcc) and
ChMPIon(cmpicc) since we have no access to these two MPI implementations
and cannot verify their correctness. (AKC - 2010/07/14 - Bug 1921)
- PHDF5 was changed to use "mpiexec" instead of mpirun as the default
MPI applications startup command as defined in the MPI-2 definition,
section 4.1. (AKC - 2010/06/11 - Bug 1921)
Library
-------
- Fixed a bug that caused big endian machines to generate corrupt files
when using the scale-offset filter with floating point data or fill
values. Note that such datasets will no longer be readable by any
by any machine after this patch. (NAF - 2010/02/02 - Bug 2131)
- Retrieving a link's name by index in the case where the link is external
and the file that the link refers to doesn't exist will now fail
gracefully rather than cause a segmentation fault. (MAM - 2010/11/17)
- Modified metadata accumulator to better track accumulated dirty metadata
in an effort to reduce unnecessary I/O in certain situations and to
fix some other corner cases which were prone to error. (MAM - 2010/10/15)
- Added a new set of unit tests that are run during 'make check' to verify
the behavior of the metadata accumulator. (MAM - 2010/10/15)
- Modified library to always cache symbol table information. Libraries
from version 1.6.3 and earler have a bug which causes them to require
this information for some operations. (NAF - 2010/09/21 - Bug 1864)
- Fixed a bug where the library could generate an assertion/core dump when
a file that had been created with H5Pset_libver_bounds(fapl,
H5F_LIBVER_LATEST, H5F_LIBVER_LATEST) but didn't have a superblock
extension was later reopened. (QAK - 2010/09/16 - Bug 1968)
- Fixed a bug that could occur when getting information for a new-style
group that was previously opened through a file handle that was later
closed. (NAF - 2010/09/15)
- Added define check in H5public.h if stdint.h is supported by the C++
compiler. This define is only available on Windows with VS2010 and using
CMake to build the library. (ADB - 2010/09/13 - Bug 1938)
- When a mandatory filter failed to write data chunks, the dataset
couldn't close (bug 1260). The fix releases all resources and closes
the dataset but returns a failure. (SLU - 2010/09/08)
- H5Eset_current_stack now also closes the error stack set as the
default. This is to avoid a potential problem.
(SLU - 2010/09/07 - Bug 1799)
- Corrected situation where 1-D chunked dataset could get created by an
application without calling H5Pset_chunk(). H5Pset_chunk is now
required for creating all chunked datasets. (QAK - 2010/09/02)
- Fixed many memory issues that valgrind exposed. (QAK - 2010/08/24)
- Fixed the bug in the filter's public CAN_APPLY function. The return
value should be htri_t not herr_t. (SLU - 2010/08/05 - Bug 1239)
- Fixed the STDIO VFD to use fseeko64 instead of fseek64 for 64-bit I/O
support. (AKC - 2010/7/30)
- Fixed a bug in the direct I/O driver that could render files with certain
kinds of unaligned data unreadable or corrupt them. (NAF - 2010/07/28)
- valgrind reported an error of copying data to itself when a new attribute
is written. Fixed by taking out the memcpy step in the attribute code.
(SLU - 2010/07/28 - Bug 1956)
- Corrected various issues in the MPI datatype creation code which could
cause resource leaks or incorrect behavior (and may improve the
performance as well). (QAK - 2010/07/19)
- Fixed a bug that could cause file corruption when using non-default sizes
of addresses and/or lengths. This bug could also cause uncorrupted files
with this property to be unreadable. This bug was introduced in 1.8.5.
(NAF - 2010/07/16 - Bug 1951)
Parallel Library
----------------
- None
Tools
-----
- Fixed h5diff to compare member objects and groups recursively when
two files or groups are compared. (JKM - 2010/9/16 - Bug 1975)
- Fixed h5repack to be able to convert a dataset to COMPACT layout.
(JKM - 2010/09/15 - Bug 1896)
- Changed h5ls to not interpret special characters in object or attribute
names for output. (JKM - 2010/06/28 - Bug 1784)
- Revised the order of arguments for h5cc, h5fc, h5c++, h5pcc and h5pfc.
CPPFLAGS, CFLAGS, LDFLAGS, and LIBS have been duplicated with an H5BLD_
prefix to put the flags and paths from the hdf5 build in the correct
places and allow the script user to add entries in CPPFLAGS, CFLAGS,
LDFLAGS, and LIBS that will take precedence over those from the hdf5
build. The user can make these entries persistent by editing
CFLAGSBASE, CPPFLAGSBASE, LDFLAGSBASE, and LIBSBASE near the top of
the script or temporary by setting HDF5_CFLAGS, HDF5_CPPFLAGS,
HDF5_LDFLAGS, or HDF5_LIBS in the environment. The new order of
arguments in these scripts is $CLINKER $H5BLD_CPPFLAGS $CPPFLAGS
$H5BLD_CFLAGS $CFLAGS $LDFLAGS $clibpath $link_objs $LIBS $link_args
$shared_link. (LRK - 2010/10/25 - Bug 1973)
F90 API
------
- None
C++ API
------
- None
High-Level APIs:
------
- None
Fortran High-Level APIs:
------
- None
Platforms Tested
================
The following platforms and compilers have been tested for this release.
AIX 6.1 xlc 11.1.0.3
(NCSA BP) xlC 11.1.0.3
xlf 13.1.0.3
mpcc_r 11.1.0.3
mpxlf_r 13.1.0.3
FreeBSD 6.3-STABLE i386 gcc 3.4.6 [FreeBSD] 20060305
(duty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.4.5 20100803
g++ 4.4.5 20100803
gfortran 4.4.5 20100803
FreeBSD 6.3-STABLE amd64 gcc 3.4.6 [FreeBSD] 20060305
(liberty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.4.5 20100803
g++ 4.4.5 20100803
gfortran 4.4.5 20100803
Linux 2.6.18-194.3.1.el5PAE gcc (GCC) 4.1.2 and 4.4.2
#1 SMP i686 i686 i386 G95 (GCC 4.0.3 (g95 0.93!) Apr 21 2010)
(jam) GNU Fortran (GCC) 4.1.2 20080704
(Red Hat 4.1.2-48) and 4.4.2
PGI C, Fortran, C++ 10.4-0 32-bit
PGI C, Fortran, C++ 10.6-0 32-bit
Intel(R) C Compiler for 32-bit
applications, Version 11.1
Intel(R) C++ Compiler for 32-bit
applications, Version 11.1
Intel(R) Fortran Compiler for 32-bit
applications, Version 11.1
Absoft 32-bit Fortran 95 10.0.7
MPICH mpich2-1.3.1 compiled with
gcc 4.1.2 and gfortran 4.1.2
Linux 2.6.18-194.17.1.el5 gcc 4.1.2 and 4.4.2
#1 SMP x86_64 GNU/Linux G95 (GCC 4.0.3 (g95 0.93!) Apr 21 2010)
(amani) tested for both 32- and 64-bit binaries
GNU Fortran (GCC) 4.1.2 20080704
(Red Hat 4.1.2-46) and 4.4.2
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64,
Version 11.1.
PGI C, Fortran, C++ Version 9.0-4
for 64-bit target on x86-64
MPICH mpich2-1.3.1 compiled with
gcc 4.1.2 and gfortran 4.1.2
SGI ProPack 7 Linux Intel(R) C++ Version 11.1 20100806
2.6.32.19-0.3.1.1982.0.PTF- Intel(R) Fortran Version 11.1 20100806
default #1 SMP SGI MPT 2.01
SGI Altix UV
(NCSA ember)
SunOS 5.10 32- and 64-bit Sun C 5.9 Sun OS_sparc Patch 124867-16
(linew) Sun Fortran 95 8.3 Sun OS_sparc Patch 127000-13
Sun C++ 5.9 Sun OS_sparc Patch 124863-62
Sun C 5.10 SunOS_sparc Patch 141861-07
Sun Fortran 95 8.4 SunOS_sparc Patch 128231-06
Sun C++ 5.10 SunOS_sparc 128228-11
Intel Xeon Linux 2.6.18- gcc 4.2.4
92.1.10.el5_lustre.1.6.6smp- Intel(R) C++ Version 10.1.017
perfctr #8 SMP Intel(R) Fortran Compiler Version 10.1.017
(NCSA abe) Open MPI 1.3.2
MVAPICH2-1.5.1_pgi-10.8
Windows XP Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 (cmake)
Cygwin(1.7.7 native gcc(4.3.4) compiler and gfortran)
Windows XP x64 Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Visual Studio 2010 (cmake)
Cygwin(1.7.7 native gcc(4.3.4) compiler and gfortran)
Windows Vista Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows Vista x64 Visual Studio 2008 w/ Intel Fortran 10.1 (project files)
Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows 7 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Windows 7 x64 Visual Studio 2008 w/ Intel Fortran 11.1 (cmake)
Mac OS X 10.6.3 (Intel 64-bit) i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1
Darwin Kernel Version 10.3.1 GNU Fortran (GCC) 4.5.0 20090910
Intel C, C++ and Fortran compilers 11.1 20100806
Mac OS X 10.6.4 (Intel 32-bit) i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1
Darwin Kernel Version 10.4.0 GNU Fortran (GCC) 4.6.0 20101106
Intel C, C++ and Fortran compilers 12.0.0 20101110
Mac OS X 10.6.4 (Intel 64-bit) i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5659)
Darwin Kernel Version 10.6.0 GNU Fortran (GCC) 4.5.0 20090910
Intel C, C++ and Fortran compilers 11.1 20100806
Fedora 12 2.6.32.16-150.fc12.ppc64 #1 SMP ppc64 GNU/Linux
gcc (GCC) 4.4.4 20100630 (Red Hat 4.4.4-10)
GNU Fortran (GCC) 4.4.4 20100630 (Red Hat 4.4.4-10)
Debian5.06 2.6.26-2-686 #1 SMP i686 GNU/Linux
gcc (Debian 4.3.2-1.1) 4.3.2
GNU Fortran (Debian 4.3.2-1.1) 4.3.2
Debian5.06 2.6.26-2-amd64 #1 SMP x86_64 GNU/Linux
gcc (Debian 4.3.2-1.1) 4.3.2
GNU Fortran (Debian 4.3.2-1.1) 4.3.2
Fedora14 2.6.35.6-48.fc14.i686.PAE #1 SMP i686 i686 i386 GNU/Linux
gcc (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
GNU Fortran (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
Fedora14 2.6.35.6-48.fc14.x86_64 #1 SMP x86_64 x86_64 x86_64 GNU/Linux
gcc (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
GNU Fortran (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
SUSE 11.3 2.6.34.7-0.7-desktop #1 SMP PREEMPT i686 i686 i386 GNU/Linux
gcc (SUSE Linux) 4.5.0 20100604 [gcc-4_5-branch revision 160292]
GNU Fortran (SUSE Linux) 4.5.0 20100604 [gcc-4_5-branch revision 160292]
SUSE 11.3 2.6.34.7-0.7-desktop #1 SMP PREEMPT x86_64 x86_64 x86_64 GNU/Linux
gcc (SUSE Linux) 4.5.0 20100604 [gcc-4_5-branch revision 160292]
GNU Fortran (SUSE Linux) 4.5.0 20100604 [gcc-4_5-branch revision 160292]
Ubuntu 10.10 2.6.35-25-generic #44-Ubuntu SMP i686 GNU/Linux
gcc (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
GNU Fortran (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
Ubuntu 10.10 2.6.35-25-generic #44-Ubuntu SMP x86_64 GNU/Linux
gcc (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
GNU Fortran (Ubuntu/Linaro 4.4.4-14ubuntu5) 4.4.5
OpenVMS Alpha 8.3 HP C V7.3-009
HP Fortran V8.2-104679-48H9K
HP C++ V7.3-009
Tested Configuration Features Summary
========================================
In the tables below
y = tested
n = not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
Solaris2.10 32-bit n y n y y y
Solaris2.10 64-bit n y n y y y
Windows XP n y(4) n y y y
Windows XP x64 n y(4) n y y y
Windows Vista n y(4) n y y y
Windows Vista x64 n y(4) n y y y
OpenVMS Alpha n y n y y n
Mac OS X 10.6 Intel n y n y y y
AIX 6.1 32- and 64-bit y y y y y y
FreeBSD 6.3-STABLE 32&64 bit n y n y y y
CentOS 5.5 Linux 2.6.18-194 i686 GNU (1)W y y(2) y y y y
CentOS 5.5 Linux 2.6.18-194 i686 Intel W n y n y y n
CentOS 5.5 Linux 2.6.18-194 i686 PGI W n y n y y n
CentOS 5.5 Linux 2.6.16 x86_64 GNU (1) W y y(3) y y y y
CentOS 5.5 Linux 2.6.16 x86_64 Intel W n y n y y n
CentOS 5.5 Linux 2.6.16 x86_64 PGI W n y n y y y
RedHat EL4 2.6.18 Xeon Lustre C y y y y y n
Fedora 12 Linux 2.6.32.16-150.fc12.ppc64 n y n y y y
SGI Linux 2.6.32.19 y y y y y y
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit y y y y
Windows XP y y(4) y n
Windows XP x64 y y(4) y n
Windows Vista y y(4) y y
Windows Vista x64 y y(4) y y
OpenVMS Alpha n n n n
Mac OS X 10.6 y(5) n y n
AIX 6.1 32- and 64-bit n n n y
FreeBSD 6.3-STABLE 32&64 bit y n y y
CentOS 5.5 Linux 2.6.18-128 i686 GNU (1)W y y(2) y y
CentOS 5.5 Linux 2.6.18-128 i686 Intel W y y y n
CentOS 5.5 Linux 2.6.18-128 i686 PGI W y y y n
CentOS 5.5 Linux 2.6.16 x86_64 GNU (1) W y y y y
CentOS 5.5 Linux 2.6.16 x86_64 Intel W y y y n
CentOS 5.5 Linux 2.6.16 x86_64 PGI W y y y n
RedHat EL4 2.6.18 Xeon Lustre C y y y n
Fedora 12 Linux 2.6.32.16-150.fc12.ppc64 y y y y
SGI Linux 2.6.32.19 y y y y
(1) Fortran compiled with gfortran.
(2) With PGI and Absoft compilers.
(3) With PGI compiler for Fortran.
(4) Using Visual Studio 2008 w/ Intel Fortran 10.1 (Cygwin shared libraries are not supported)
(5) C and C++ shared libraries will not be built when Fortran is enabled.
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* examples/run-all-ex.sh does not work on Cygwin. (NAF - 2011/02/11)
* Parallel test, t_shapesame in testpar, is rather unstable as it continues to
have occasional errors in AIX and quite often in NCSA Abe. It is being built
but it is not run automatically in the "make check" command. One would have to
run it by hand to see if it works in a particular machine. AKC - 2011/01/28
* Although OpenVMS Alpha is supported, there are several problems with the C
test suite - getname.c, lheap.c, lheap.c, mtime.c, and stab.c. The test
suite for h5diff also fails. These failures are from the tests, not the
library. We have fixed these failures. But it's too late to put the fixes
into this release. If you install the 1.8.6 library, it should still work
despite of these test failures. If you want the working copy without any
test failure, you can request it from us. SLU - 2011/01/26
* If parallel gmake (e.g., gmake -j 4) is used, the "gmake clean" command
sometimes fails in the perform directory due to the attempt to remove the
executable of h5perf or h5perf_serial by two "parallel" commands. This error
has no consequence on the functionality of the HDF5 library or install. It
is fixed in the next release. AKC - 2011/01/25
* While working on the 1.8.6 release of HDF5, a bug was discovered that can
occur when reading from a dataset in parallel shortly after it has been
written to collectively. The issue was exposed by a new test in the parallel
HDF5 test suite, but had existed before that. We believe the problem lies with
certain MPI implementations and/or filesystems.
We have provided a pure MPI test program, as well as a standalone HDF5
program, that can be used to determine if this is an issue on your system.
They should be run across multiple nodes with a varying number of processes.
These programs can be found at:
http://www.hdfgroup.org/ftp/HDF5/examples/known_problems/
* The h5diff tool can display garbage values when variable-length strings in
a compound type dataset are compared. This also occurs with variable-length
string arrays in a compound type dataset. See bug #1989. This will be fixed
in the next release. JKM - 2010/11/05
* The AIX --enable-shared setting does not quite work. It can produce a shared
library, but there cannot be more than one shared library that is
interlinked. This means that the high level APIs will not work which is not
very useful. We hope to have a solution in the next release.
(AKC - 2010/10/15)
* H5Eset_auto can cause a seg fault for a library API call if the application
compiles with -DH5_USE_16_API (see bug 1707). It will be fixed in the
next release. SLU - 2010/10/5
* The library's test dt_arith.c showed a compiler's rounding problem on
Cygwin when converting an unsigned long long to a long double. The
library's own conversion works fine. We defined a macro for Cygwin to
skip this test until we can solve the problem. Please see bug #1813.
SLU - 2010/5/5
* All the VFL drivers aren't backwardly compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. A new parameter was added to the
get_eoa and set_eoa callback functions, and a new callback function
get_type_map was added. The public function H5FDrealloc was taken out in
1.8. The problem only happens when users define their own driver for 1.6
and try to plug in a 1.8 library. This will be fixed in 1.10. SLU - 2010/2/2
* MinGW has a missing libstdc++.dll.a library file and will not successfully link
C++ applications/tests. Do not use the enable-cxx configure option. Read all of
the INSTALL_MINGW.txt file for all restrictions. ADB - 2009/11/11
* The PathScale MPI implementation, accessing a Panasas file system, would
cause H5Fcreate() with H5F_ACC_EXCL to fail even when the file does not
exist. This is due to the MPI_File_open() call failing if the amode has
the MPI_MODE_EXCL bit set. (See bug 1468 for details.) AKC - 2009/8/11
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO. CMC - 2009/04/28
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh and
tools/h5copy/testh5copy.sh will fail some of their sub-tests. These
sub-tests are expected to fail and should exit with a non-zero code but
the yod command does not propagate the exit code of the executables. Yod
always returns 0 if it can launch the executable. The test suite shell
expects a non-zero for this particular test. Therefore, it concludes the
test has failed when it receives 0 from yod. To skip all the "failing"
tests for now, change them as shown below.
======== Original tools/h5ls/testh5ls.sh =========
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Change to ===============================
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
==================================================
======== Original tools/h5copy/testh5copy.sh =========
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
H5LSTEST $FILEOUT
======== Change to ===============================
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
echo SKIP H5LSTEST $FILEOUT
==================================================
AKC - 2008/11/10
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message "yod allocation delayed for node recovery." This interferes
with test suites that do not expect to see this message. See the "Red Storm"
section in file INSTALL_parallel for a way to deal with this problem.
AKC - 2008/05/28
* On an Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use the -mp -O1 compilation flags to build the libraries. A higher level
of optimization causes failures in several HDF5 library tests.
* On mpich 1.2.5 and 1.2.6 on a system using four processors, if more than
two processes contribute no I/O and the application asks to do collective
I/O, we have found that a simple collective write will sometimes hang. This
can be verified with the t_mpi test under testpar.
* A dataset created or rewritten with a v1.6.3 or later library cannot be
read with the v1.6.2 or earlier library when the Fletcher32 EDC filter
is enabled. There was a bug in the calculation of the Fletcher32 checksum
in the library before v1.6.3; the checksum value was not consistent
between big-endian and little-endian systems. This bug was fixed in
Release 1.6.3. However, after fixing the bug, the checksum value was no
longer the same as before on little-endian system. Library releases after
1.6.4 can still read datasets created or rewritten with an HDF5 library of
v1.6.2 or earlier. SLU - 2005/6/30
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'. To
work around this, set the environment variable MP_INFOLEVEL to 0 to
minimize the messages and run the tests again. The tests may fail with
messages like "The socket name is already in use", but HDF5 does not use
sockets. This failure is due to problems with the poe command trying to
set up the debug socket. To resolve this problem, check to see whether
there are any old /tmp/s.pedb.* files around. These are sockets used by
the poe command and left behind if the command failed at some point. To
resolve this, ask your system administrator to remove the
old/tmp/s.pedb.* files, and then ask IBM to provide a means to run poe
without the debug socket.
* The --enable-static-exec configure flag will only statically link
libraries if the static version of that library is present. If only the
shared version of a library exists (i.e., most system libraries on
Solaris, AIX, and Mac, for example, only have shared versions), the flag
should still result in a successful compilation, but note that the
installed executables will not be fully static. Thus, the only guarantee
on these systems is that the executable is statically linked with just
the HDF5 library.
* On an SGI Altix SMP ia64 system, the Intel compiler version 10.1 (which
is the default on that system) does not work properly and results in
failures during the make check (in a static build) and the make
installcheck (in a shared build). This appears to be a compiler
optimization problem. Reducing the optimization by setting CFLAGS to
-O1 or below resolves the issue. Using a newer version of the compiler
(11.0) avoids the issue. MAM - 2010/06/01
* On solaris systems, when running the examples with the scripts installed in
.../share/hdf5_examples, two of the c tests, h5_extlink and h5_elink_unix2win
may fail or generate HDF5 errors because the script commands in c/run-c-ex.sh
fail to create test directories red, blue, and u2w. Moving the '!' in lines
67, 70, 73 of run-c-ex.sh will fix the problem. For example the script command
"if ! test -d red; then" will work on solaris if changed to
"if test ! -d red; then".
%%%%1.8.5%%%%
HDF5 version 1.8.5 released on Fri Jun 4 13:27:31 CDT 2010
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.4 and HDF5 1.8.5, and
contains information on the platforms tested and known problems in HDF5-1.8.5.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt and HISTORY-1_8.txt
in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.5 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.5 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.5 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.5 (current
release) versus Release 1.8.4":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.4
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- CMake Early Access: This release adds support for building HDF5 using
the CMake system. Initial work has targeted Windows, but other platforms
can be used. See the CMake.TXT file for more information. Version
2.8.1 of CMake is required.
- Configure now adds appropriate defines for supporting large (64-bit)
files on all systems, where supported, by default, instead of only Linux.
This large file support is controllable with the --enable-largefile
configure option. The Linux-specific --enable-linux-lfs option has been
deprecated in favor of this new option. Please note that specifying
--disable-large does NOT attempt to "turn off" largefile support if it
is natively supported by the compiler, but rather just disables
configure from actively trying to add any additional compiler flags.
(MAM - 2010/05/05 - Bug # 1772/1434)
- Fixed an signal handling mask error in H5detect that might result in
SIGBUS or SIGSEGV failures in some platforms such as Linux on Sparc.
(AKC - 2010/4/28 - Bug # 1764)
- Fixed various "strict aliasing" problems, allowing higher levels
of compiler optimization (in particular, allowing '-O3' to work
with recent versions of GCC). (QAK - 2010/04/26)
- Upgraded versions of autotools used to generate configuration suite.
We now use Automake 1.11.1, Autoconf 2.65, and Libtool 2.2.6b.
(MAM - 2010/04/15)
- Added the xlc-* and mpcc_r-* BASENAME patterns to be recognized as IBM
compilers so that the IBM compiler options can be added properly. This
allows non-system-default compiler command names (e.g. xlc-m.n.k.l) be
recognized. (AKC - 2009/11/26)
Library
-------
- Performance is substantially improved when extending a dataset with early
allocation. (NAF - 2010/03/24 - Bug # 1637)
- Added support for filtering densely stored groups. Many of the API
functions related to filters have been extended to support dense groups
as well as datasets. Pipeline messages can now be stored in a group's
object header. (NAF/QAK - 2009/11/3)
Parallel Library
----------------
- None
Tools
-----
- h5dump: Added the new packed bits feature which prints packed bits stored
in an integer dataset. (AKC/ADB - 2010/5/7)
- h5diff: Fixed incorrect behavior (hang) in parallel mode when specifying
invalid options (ex: -v and -q). (JKM - 2010/02/17)
- h5diff: Added new flag --no-dangling-links (see --help for details).
(JKM - 2010/02/10)
- h5diff: Added new flag --follow-symlinks (see --help for details).
(JKM - 2010/01/25)
- h5diff: Added a fix to correct the display of garbage values when
displaying big-endian data on a little-endian machine. (JKM - 2009/11/20)
High-Level APIs
---------------
- None
F90 API
-------
- None
C++ API
-------
- New member functions
+ Overloaded CommonFG::getObjnameByIdx to take char* for name.
+ Overloaded CommonFG::getObjTypeByIdx to return type name as a char*.
(BMR - 2010/05/10)
+ Added DataSet::getInMemDataSize() to simplify getting the dataset's
data size in memory. (BMR - 2009/07/26)
Support for New Platforms, Languages, and Compilers
===================================================
- AIX 6.1 has been added. (AKC - 2010/1/4)
Bug Fixes since HDF5-1.8.4
==========================
Configuration
-------------
- Fixed various "strict aliasing" problems, allowing higher levels
of compiler optimization (in particular, allowing '-O3' to work
with recent versions of GCC). (QAK - 2010/04/26)
Library
-------
- Fixed a file corruption bug that could happen when shrinking a compressed
dataset. (NAF - 2010/05/20)
- Fixed some memory leaks in VL datatype conversion when strings are
used as fill values. (MAM - 2010/05/12 - Bug # 1826)
- Fixed an H5Rcreate failure when passing in a -1 for the dataspace
identifier. (ADB - 2010/4/28)
- Fixed a bug when copying objects with NULL references with the
H5O_COPY_EXPAND_REFERENCE_FLAG flag set. (NAF - 2010/04/08 - Bug # 1815)
- Added a mechanism to the H5I interface to save returned object identifier
structures for immediate re-use if needed. This addresses a potential
performance issue by delaying the case when the next identifier to be
registered has grown so large that it wraps around and needs to be
checked to see whether it is available for distribution.
(MAM - 2010/03/15 - Bug # 1730)
- Files can now be concurrently opened more than once using the core file
driver, as long as the backing store is used. (NAF - 2010/03/09)
- Added support for H5O_COPY_EXPAND_EXT_LINK_FLAG to H5Ocopy. External
links will now be expanded if this flag is set.
(NAF - 2010/03/05 - Bug # 1733)
- Fixed a bug where the library, when traversing an external link, would
reopen the source file if nothing else worked. (NAF - 2010/03/05)
- Fixed a bug where fractal heap identifiers for attributes and shared
object header messages could be incorrectly encoded in the file for
files created on big-endian platforms.
Please see http://www.hdfgroup.org/HDF5/release/known_problems if you
suspect you have a file with this problem.
(QAK - 2010/02/23 - Bug # 1755)
- Fixed an intermittent bug in the b-tree code which could be triggered
by expanding and shrinking chunked datasets in certain ways.
(NAF - 2010/02/16)
- H5Tdetect_class said a VL string is a string type. But when it's in
a compound type, it said it's a VL type. THis has been fixed to be
consistent; it now always returns a string type.
(SLU - 2009/12/10 - Bug # 1584)
- Allow "child" files from external links to be correctly located when
relative to a "parent" file that is opened through a symbolic link.
(QAK - 2009/12/01)
Parallel Library
----------------
- Parallel mode in AIX will fail some of the testcheck_version.sh tests
where it treats "exit(134) the same as if process 0 had received an abort
signal. Fixed. (AKC - 2009/11/3)
Tools
-----
- Fixed h5ls to return exit code 1 (error) when a non-existent file is
specified. (JKM - 2010/04/27 - Bug # 1793)
- Fixed h5copy failure when copying a dangling link that is specified
directly. (JKM - 2010/04/22 - Bug # 1817)
- Fixed an h5repack failure that lost attributes from a dataset of
reference type. (JKM - 2010/3/25 - Bug # 1726)
- Fixed h5repack error that set NULL for object reference values for
datasets, groups, or named datatypes. (JKM - 2010/03/19 - Bug # 1814)
F90 API
------
- None
C++ API
------
- The constructor PropList::PropList(id) was fixed to act properly
according to the nature of 'id'. When 'id' is a property class
identifier, a new property list will be created. When 'id' is a
property list identifier, a copy of the property list will be made.
(BMR - 2010/5/9)
- The parameters 'size' and 'bufsize' in CommonFG::getLinkval and
CommonFG::getComment, respectively, now have default values for the
user's convenience. (BMR - 2009/10/23)
- NULL pointer accessing was fixed. (BMR - 2009/10/05 - Bug # 1061)
- Read/write methods of DataSet and Attribute classes were fixed
to handle string correctly. (BMR - 2009/07/26)
High-Level APIs:
------
- Fixed a bug in H5DSattach_scale, H5DSis_attached, and H5DSdetach_scale
caused by using the H5Tget_native_type function to determine the native
type for reading the REFERENCE_LIST attribute. This bug was exposed
on Mac PPC. (EIP - 2010/05/22 - Bug # 1851)
- Fixed a bug in the H5DSdetach_scale function when 0 bytes were
allocated after the last reference to a dimension scale was removed
from the list of references in a VL element of the DIMENSION_LIST
attribute. Modified the function to comply with the specification:
the DIMENSION_LIST attribute is now deleted when no dimension scales
are left attached. (EIP - 2010/05/14 - Bug # 1822)
Fortran High-Level APIs:
------
- None
Platforms Tested
================
The following platforms and compilers have been tested for this release.
AIX 5.3 xlc 7.0.0.9, 8.0.0.20, 9.0.0.4
(LLNL Up) xlC 7.0.0.9, 8.0.0.20, 9.0.0.4
xlf 9.1.0.9, 10.1.0.9, 11.1.0.7
mpcc_r 7.0.0.9
mpxlf_r 09.01.0000.0008
AIX 6.1 xlc 10.1.0.6
(NCSA BP) xlC 10.1.0.6
xlf 12.1.0.7
Cray XT3 (2.1.56) cc (pgcc) 10.0-0
(SNL red storm) ftn (pgf90) 10.0-0
CC (pgCC) 10.0-0
FreeBSD 6.3-STABLE i386 gcc 3.4.6 [FreeBSD] 20060305
(duty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.3.4 20090419
g++ 4.3.4 20090419
gfortran 4.3.4 20090419
FreeBSD 6.3-STABLE amd64 gcc 3.4.6 [FreeBSD] 20060305
(liberty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.4.1 20090421
g++ 4.4.1 20090421
gfortran 4.4.1 20090421
Linux 2.6.18-128.1.6.el5xen gcc (GCC) 4.1.2 20080704 and 4.4.2
#1 SMP i686 i686 i386 GNU Fortran (GCC) 4.1.2 20080704 and 4.4.2
(jam) g++ (GCC) 4.1.2 20080704 and 4.4.2
G95 (GCC 4.0.3 (g95 0.93!) Apr 21 2010)
Absoft 32-bit Fortran 95 10.0.7
PGI C, Fortran, C++ 10.4-0 32-bit
Intel(R) C, C++, Fortran Compilers for 32-bit
applications, Version 11.1 Build 20090827
MPICH mpich2-1.0.8 compiled with
gcc 4.1.2 and GNU Fortran (GCC) 4.1.2
Linux 2.6.18-164.el5 #1 SMP gcc 4.1.2 20080704 and gcc 4.4.2
x86_64 GNU/Linux GNU Fortran (GCC) 4.1.2 20080704 and 4.4.2
(amani) g++ (GCC) 4.1.2 20080704 and 4.4.2
G95 (GCC 4.0.3 (g95 0.93!) Apr 21 2010)
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64,
Version 11.1 Build 20090827.
PGI C, Fortran, C++ Version 10.4-0
for 32 & 64-bit target on x86-64
MPICH mpich2-1.0.8 compiled with
gcc 4.1.2 and GNU Fortran (GCC) 4.1.2
Linux 2.6.16.54-0.2.5 #1 Intel(R) C++ Version 11.0.074
SGI Altix SMP ia64 Intel(R) Fortran Itanium(R) Version 11.0.074
(cobalt) SGI MPI 1.38
SunOS 5.10 32- and 64-bit Sun C 5.9 SunOS_sparc Patch 124867-14
(linew) Sun Fortran 95 8.3 SunOS_sparc
Patch 127000-13
Sun C++ 5.9 SunOS_sparc Patch 124863-23
Intel Xeon Linux 2.6.18- Intel(R) C++ Version 10.0.026
92.1.10.el5_lustre.1.6.6smp- Intel(R) Fortran Compiler Version 10.0.026
perfctr #7 SMP Open MPI 1.2.2
(abe) MVAPICH2-0.9.8p28p2patched-intel-ofed-1.2
compiled with icc v10.0.026 and ifort 10.0.026
Linux 2.6.18-76chaos #1 SMP Intel(R) C, C++, Fortran Compilers for
SMP x86_64 GNU/Linux applications running on Intel(R) 64,
(SNL Glory) Versions 11.1.
Windows XP Visual Studio 2008 w/ Intel Fortran 10.1
Cygwin(1.7.5 native gcc(4.3.4) compiler and
gfortran)
Windows XP x64 Visual Studio 2008 w/ Intel Fortran 10.1
Windows Vista Visual Studio 2008 w/ Intel Fortran 10.1
Windows Vista x64 Visual Studio 2008 w/ Intel Fortran 10.1
MAC OS 10.6.3 (Intel) i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1
(pahra) GNU Fortran (GCC) 4.5.0 20090910
i686-apple-darwin10-g++-4.2.1 (GCC) 4.2.1
Intel C, C++ and Fortran compilers 11.1
MAC OS 10.5.8 (Intel) i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1
(tejeda)
MAC OS 10.5 (PPC) powerpc-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1
(juniper-w)
OpenVMS Alpha V8.3 HP C V7.3-009
HP C++ V7.3-009
HP Fortran V8.0-1-104669-48GBT
Supported Configuration Features Summary
========================================
In the tables below
y = tested and supported
n = not supported or not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
Solaris2.10 32-bit n y n y y y
Solaris2.10 64-bit n y n y y y
Windows XP n y(4) n(4) y y y
Windows XP x64 n y(4) n(4) y y y
Windows Vista n y(4) n(4) y y y
Windows Vista x64 n y(4) n(4) y y y
Mac OS X 10.5 PPC n n n n y n
Mac OS X 10.5 Intel n y n y y y
Mac OS X 10.6 Intel n y n y y y
AIX 5.3 32- and 64-bit n y n y y n
AIX 6.1 32- and 64-bit n y n y y n
FreeBSD 6.3-STABLE 32&64 bit n y n y y y
RedHat EL4 2.6.9-42 i686 GNU (1) W y y y y y y
RedHat EL5 2.6.18-128 i686 GNU (1)W y y(2) y y y y
RedHat EL5 2.6.18-128 i686 Intel W n y n y y n
RedHat EL5 2.6.18-128 i686 PGI W n y n y y n
SuSe Linux 2.6.16 x86_64 GNU (1) W y y(3) y y y y
SuSe Linux 2.6.16 x86_64 Intel W n y n y y n
SuSe Linux 2.6.16 x86_64 PGI W n y n y y y
SuSe Linux 2.6.16 SGI Altix ia64 C y y y y y y
RedHat EL4 2.6.18 Xeon Lustre C y y y y y n
Cray XT3 2.1.56 y y y y y n
OpenVMS Alpha V8.3 n y n y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit y y y y
Windows XP y y(4) y n
Windows XP x64 y y(4) y n
Windows Vista y y(4) y n
Windows Vista x64 y y(4) y n
Mac OS X 10.5 PPC y n n n
Mac OS X 10.5 (Intel) y(5) n y n
Mac OS X 10.6 (Intel) y(5) n y n
AIX 5.3 32- and 64-bit n n n n
AIX 6.1 32- and 64-bit n n n n
FreeBSD 6.3-STABLE 32&64 bit y n y y
RedHat EL4 2.6.9-42 i686 GNU (1) W y y y y
RedHat EL5 2.6.18-128 i686 GNU (1)W y y(2) y y
RedHat EL5 2.6.18-128 i686 Intel W y y y n
RedHat EL5 2.6.18-128 i686 PGI W y y y n
SuSe Linux 2.6.16 x86_64 GNU (1) W y y y y
SuSe Linux 2.6.16 x86_64 Intel W y y y n
SuSe Linux 2.6.16 x86_64 PGI W y y y n
SuSe Linux 2.6.16 SGI Altix ia64 C y n
RedHat EL4 2.6.18 Xeon Lustre C y y y n
Cray XT3 2.1.56 n n n n
OpenVMS Alpha V8.3 n n n n
(1) Fortran compiled with g95.
(2) With PGI and Absoft compilers.
(3) With PGI compiler for Fortran.
(4) Using Visual Studio 2008. (Cygwin shared libraries are not
supported.)
(5) Shared C and C++ are disabled when Fortran is configured in.
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* The library's test dt_arith.c exposed a compiler's rounding problem on
Cygwin when converting from unsigned long long to long double. The
library's own conversion works correctly. A macro is defined for Cygwin
to skip this test until we can solve the problem. (Please see bug #1813.)
SLU - 2010/5/5
* All the VFL drivers aren't backward compatible. In H5FDpublic.h, the
structure H5FD_class_t changed in 1.8. There is a new parameter added to
get_eoa and set_eoa callback functions. A new callback function
get_type_map was added. The public function H5FDrealloc was taken
out in 1.8. The problem only happens when users define their own driver
for 1.6 and try to plug it into a 1.8 library. This affects a very small
number of users. (See bug report #1279.) SLU - 2010/2/2
* MinGW has a missing libstdc++.dll.a library file and will not successfully
link C++ applications/tests. Do not use the enable-cxx configure option.
Read all of the INSTALL_MINGW.txt file for all restrictions.
ADB - 2009/11/11
* Some tests in tools/h5repack may fail in AIX systems when -q32 mode is used.
The error is due to insufficient memory requested. Request a large amount
of runtime memory by setting the following environment variable for more
memory.
LDR_CNTRL=MAXDATA=0x20000000@DSA
AKC - 2009/10/31
* The PathScale MPI implementation, accessing a Panasas file system, would
cause H5Fcreate() with H5F_ACC_EXCL to fail even when the file is not
existing. This is due to the MPI_File_open() call failing if the amode has
the MPI_MODE_EXCL bit set. (See bug 1468 for details.) AKC - 2009/8/11
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO. CMC - 2009/04/28
* For gcc v4.3 and v4.4, with production mode, if -O3 is used, H5Tinit.c
would fail to compile. Actually bad H5Tinit.c is produced. If -O (same
as -O1) is used, H5Tinit.c compiled okay but test/dt_arith would fail.
When -O0 (no optimizatio) is used, H5Tinit.c compilete okay and all
tests passed. Therefore, -O0 is imposed for v4.3 and v4.4 of gcc.
AKC - 2009/04/20
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh and
tools/h5copy/testh5copy.sh will fail some of its sub-tests. These sub-tests
are expected to fail and should exit with a non-zero code but the yod
command does not propagate the exit code of the executables. Yod always
returns 0 if it can launch the executable. The test suite shell expects
a non-zero for this particular test, therefore it concludes the test has
failed when it receives 0 from yod. Skip all the "failing" test for now
by changing them as following.
======== Original tools/h5ls/testh5ls.sh =========
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Change to ===============================
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
==================================================
======== Original tools/h5copy/testh5copy.sh =========
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
H5LSTEST $FILEOUT
======== Change to ===============================
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
echo SKIP H5LSTEST $FILEOUT
==================================================
AKC - 2008/11/10
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message, "yod allocation delayed for node recovery". This interferes with
test suites that do not expect seeing this message. See the section of "Red
Storm" in file INSTALL_parallel for a way to deal with this problem.
AKC - 2008/05/28
* On Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
SLU - 2005/6/30
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use,"
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
* There is also a configure error on Altix machines that incorrectly reports
when a version of Szip without an encoder is being used.
* On FREE-BSD systems when shared libraries are disabled, make install fails
in install-examples with the error '"Makefile", line 635: Need an operator'.
When this error occurs removing or commenting out the line "export
LD_LIBRARY_PATH=$(LL_PATH)" (line 635 in examples/Makefile) will allow make
install to finish installing examples. The problem will be fixed in the
next release. LRK - 2010/05/26
* On cobalt, an SGI Altix SMP ia64 system, Intel compiler version 10.1 (which
is the default on that system) does not work properly and results in
failures during make check (in a static build) and make installcheck (during
a shared build). This appears to be a compiler optimization problem.
Reducing optimization by setting CFLAGS to -O1 or below resolves the issue.
Alternatively, using a newer version of the compiler (11.0) also works as
intended. MAM - 2010/06/01
%%%%1.8.4%%%%
HDF5 version 1.8.4 released on Tue Nov 10 15:33:14 CST 2009
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.3 and
HDF5 1.8.4, and contains information on the platforms tested and
known problems in HDF5-1.8.4
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.4 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.4 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.4 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.4 (current
release) versus Release 1.8.3":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.3
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- Configuration suite now uses Automake 1.11 and Autoconf 2.64.
MAM 2009/08/31.
- Changed default Gnu fortran compiler from g95 to gfortran since
gfortran is more likely installed with gcc now. -AKC 2009/07/19-
Library
-------
- The embedded library information is displayed by H5check_version() if a
version mismatch is detected. Also changed H5check_version() to
suppress the warning message totally if $HDF5_DISABLE_VERSION_CHECK is 2
or higher. (Old behavior treated 3 or higher the same as 1, that is
print a warning and allows the program to continue. (AKC - 2009/9/28)
- If a user does not care for the extra library information insert
in the executables, he may turn it off by --disable-embedded-libinfo
during configure. (AKC - 2009/9/15)
Parallel Library
----------------
- None
Tools
-----
- h5diff: h5diff treats two INFINITY values different. Fixed by checking
(value==expect) before call ABS(...) at h5diff_array.c. This will make
that (INF==INF) is true (INF is treated as an number instead of NaN)
(PC -- 2009/07/28)
- h5diff: add option "--use-system-epsilon" to print difference if
(|a-b| > EPSILON).
Change default to use strict equality (PC -- 2009/09/12)
High-Level APIs
---------------
- None
F90 API
-------
- Added H5Oopen_by_addr_f MSB - 9/14/09
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- PathScale compilers are recognized and can build the HDF5 library
properly. AKC - 2009/7/28 -
Bug Fixes since HDF5-1.8.3
==========================
Configuration
-------------
- Removed the following config files, as we no longer support them:
config/dec-osf*, config/hpux11.00, config/irix5.x,
config/powerpc-ibm-aix4.x config/rs6000-ibm-aix5.x config/unicos*
MAM - 2009/10/08
- Modified configure and make process to properly preserve user's CFLAGS
(and company) environment variables. Build will now properly use
automake's AM_CFLAGS for any compiler flags set by the configure
process. Configure will no longer modify CFLAGS directly, nor will
setting CFLAGS during make completely replace what configure has set up.
MAM - 2009/10/08
- Support for TFLOPS, config/intel-osf1, is removed since the TFLOPS
machine has long retired. AKC - 2009/10/06.
- Added $(EXEEXT) extension to H5detect when it's executed in the
src/Makefile to generate H5Tinit.c so it works correctly on platforms
that require the full extension when running executables.
MAM - 2009/10/01 - BZ #1613
- Configure will now set FC and CXX to "no" when fortran and c++
are not being compiled, respectively, so configure will not run
some of the compiler tests for these languages when they are not
being used. MAM - 2009/10/01
- The --enable-static-exec flag will now properly place the -static flag
on the link line of all installed executables. This will force the
executable to link with static libraries over shared libraries, provided
the static libraries are available. MAM - 2009/08/31 - BZ #1583
- The PathScale compiler (v3.2) was mistaken as gcc v4.2.0 but it fails to
recognize some gcc options. Fixed. (see bug 1301). AKC - 2009/7/28 -
Library
-------
- Fixed a bug where writing and deleting many global heap objects (i.e.
variable length data) would render the file unreadable. Previously
created files exhibiting this problem should now be readable.
NAF - 2009/10/27 - 1483
- Fixed error in library's internal caching mechanisms which could cause
an assertion failure (and attendent core dump) when encountering an
unusually formatted file. (QAK - 2009/10/13)
- Fixed incorrect return value for H5Pget_preserve. AKC - 2009/10/08 - 1628
- Fixed an assertion failure that occurred when H5Ocopy was called on a
dataset using a vlen inside a compound. NAF - 2009/10/02 - 1597
- Fixed incorrect return value for H5Pget_filter_by_id1/2 in H5Ppublic.h.
NAF - 2009/09/25 - 1620
- Fixed a bug where properties weren't being compared with the registered
compare callback. NAF - 2009/09/25 - 1555
- Corrected problem where library would re-write the superblock in a file
opened for R/W access, even when no changes were made to the file.
(QAK - 2009/08/20, Bz#1473)
- Fixed a bug where H5Pget_filter_by_id would succeed when called for a
filter that wasn't present. NAF - 2009/06/25 - 1250
- Fixed an issue with committed compound datatypes containing a vlen. Also
fixed memory leaks involving committed datatypes. NAF - 2009/06/10 - 1593
Parallel Library
----------------
- None
Tools
-----
- h5dump/h5ls display buffer resize fixed in tools library.
ADB - 2009/7/21 - 1520
- perf_serial test added to Windows projects and check batch file.
ADB - 2009/06/11 -1504
F90 API
------
- Fixed bug in h5lget_info_by_idx_f by adding missing arguments,
consequently changing the API. New API is:
SUBROUTINE h5lget_info_by_idx_f(loc_id, group_name, index_field, order, n, &
link_type, f_corder_valid, corder, cset, address, val_size, hdferr, lapl_id)
MSB - 2009/9/17 - 1652
- Corrected the values for the H5L_flags FORTRAN constants:
H5L_LINK_ERROR_F, H5L_LINK_HARD_F, H5L_LINK_SOFT_F, H5L_LINK_EXTERNAL_F
MSB - 2009-09-17 - 1653
- Added FORTRAN equivalent of C constant H5T_ORDER_NONE: H5T_ORDER_NONE_F
MSB - 2009-9-24 - 1471
C++ API
------
- None
High-Level APIs:
------
- Fixed a bug where the H5TB API would forget the order of fields when added
out of offset order. NAF - 2009/10/27 - 1582
- H5DSis_attached failed to account for different platform types. Added a
get native type call. ADB - 2009/9/29 - 1562
Fortran High-Level APIs:
------
- Lite: the h5ltread_dataset_string_f and h5ltget_attribute_string_f functions
had memory problems with the g95 fortran compiler. (PVN <20> 5/13/2009) 1522
Platforms Tested
================
The following platforms and compilers have been tested for this release.
AIX 5.3 xlc 7.0.0.8
(LLNL Up) xlf 09.01.0000.0008
xlC 7.0.0.8
mpcc_r 7.0.0.8
mpxlf_r 09.01.0000.0008
Cray XT3 (2.0.41) cc (pgcc) 7.1-4
(SNL red storm) ftn (pgf90) 7.1-4
CC (pgCC) 7.1-4
FreeBSD 6.3-STABLE i386 gcc 3.4.6 [FreeBSD] 20060305
(duty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.3.5 20091004
g++ 4.3.5 20091004
gfortran 4.3.5 20091004
FreeBSD 6.3-STABLE amd64 gcc 3.4.6 [FreeBSD] 20060305
(liberty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.4.2 20091006
g++ 4.4.2 20091006
gfortran 4.4.2 20091006
Linux 2.6.18-164.el5 gcc (GCC) 4.1.2 20080704
#1 SMP i686 i686 i386 G95 (GCC 4.0.3 (g95 0.92!) Jun 24 2009)
(jam) GNU Fortran (GCC) 4.1.2 20080704
(Red Hat 4.1.2-46)
PGI C, Fortran, C++ 8.0-5 32-bit
PGI C, Fortran, C++ 8.0-1 32-bit
Intel(R) C Compiler for 32-bit
applications, Versions 11.0, 11.1
Intel(R) C++ Compiler for 32-bit
applications, Version 11.0, 11.1
Intel(R) Fortran Compiler for 32-bit
applications, Version 11.0, 11.1
Absoft 32-bit Fortran 95 10.0.7
MPICH mpich2-1.0.8 compiled with
gcc (GCC) 4.1.2 and G95
(GCC 4.0.3 (g95 0.92!)
Linux 2.6.18-164.el5 #1 SMP gcc 4.1.2 20080704
x86_64 GNU/Linux G95 (GCC 4.0.3 (g95 0.92!) Jun 24 2009)
(amani) tested for both 32- and 64-bit binaries
Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64,
Versions 11.1.
PGI C, Fortran, C++ Version 9.0-4
for 64-bit target on x86-64
gcc 4.1.2 and G95 (GCC 4.0.3 (g95 0.92!)
MPICH mpich2-1.0.8 compiled with
gcc 4.1.2 and G95 (GCC 4.0.3 (g95 0.92!)
GNU Fortran (GCC) 4.1.2 20080704
(Red Hat 4.1.2-46)
Linux 2.6.16.60-0.42.5 #1 Intel(R) C++ Version 10.1.017
SGI Altix SMP ia64 Intel(R) Fortran Itanium(R) Version 10.1.017
(cobalt) SGI MPI 1.38
SunOS 5.10 32- and 64-bit Sun C 5.9 SunOS_sparc Patch 124867-11 2009/04/30
(linew) Sun Fortran 95 8.3 SunOS_sparc
Patch 127000-11 2009/10/06
Sun C++ 5.9 SunOS_sparc
Patch 124863-16 2009/09/15
Intel Xeon Linux 2.6.18- Intel(R) C++ Version 10.0.026
92.1.10.el5_lustre.1.6.6smp- Intel(R) Fortran Compiler Version 10.0.026
perfctr #6 SMP Open MPI 1.2.2
(abe) MVAPICH2-0.9.8p28p2patched-intel-ofed-1.2
compiled with icc v10.0.026 and ifort 10.0.026
IA-64 Linux 2.4.21-309.tg1 gcc (GCC) 3.2.2
#1 SMP ia64 Intel(R) C++ Version 8.1.037
(NCSA tg-login) Intel(R) Fortran Compiler Version 8.1.033
mpich-gm-1.2.7p1..16-intel-8.1.037-r1
Linux 2.6.9-55.0.9.EL_lustre Intel(R) C, C++, Fortran Compilers for
.1.4.11.1smp #1 SMP applications running on Intel(R) 64,
SMP x86_64 GNU/Linux Versions 10.1.
(SNL Thunderbird)
Linux 2.6.18-76chaos #1 SMP Intel(R) C, C++, Fortran Compilers for
SMP x86_64 GNU/Linux applications running on Intel(R) 64,
(SNL Glory) Versions 10.1.
Windows XP Visual Studio 2005 w/ Intel Fortran 9.1
Cygwin(native gcc compiler and g95)
Windows XP x64 Visual Studio 2005 w/ Intel Fortran 9.1
Windows Vista Visual Studio 2005 w/ Intel Fortran 9.1
Windows Vista x64 Visual Studio 2005 w/ Intel Fortran 9.1
MAC OS 10.5.6 (Intel) i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1
GNU Fortran (GCC) 4.3.0 20070810
G95 (GCC 4.0.3 (g95 0.91!) Apr 24 2008)
Intel C, C++ and Fortran compilers 10.1
Supported Configuration Features Summary
========================================
In the tables below
y = tested and supported
n = not supported or not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
Solaris2.10 32-bit n y n y y y
Solaris2.10 64-bit n y n y y y
Windows XP n y(4) n(4) y y y
Windows XP x64 n y(4) n(4) y y y
Windows Vista n n n y y y
Mac OS X 10.5 Intel n y n y y y
AIX 5.3 32- and 64-bit n y n y y n
FreeBSD 6.3-STABLE 32&64 bit n y n y y y
RedHat EL5 2.6.18-164 i686 GNU (1)W y y(2) y y y y
RedHat EL5 2.6.18-164 i686 Intel W n y n y y n
RedHat EL5 2.6.18-164 i686 PGI W n y n y y n
RedHat EL5 2.6.18-164 x86_64 GNU(1)W y y(3) y y y y
RedHat EL5 2.6.18-164 x86_64 IntelW n y n y y n
RedHat EL5 2.6.18-164 x86_64 PGI W n y n y y y
SuSe Linux 2.6.16 SGI Altix ia64 C y y y y y y
RedHat EL4 2.6.18 Xeon Lustre C y y y y y n
SuSe Linux 2.4.21 ia64 Intel C y y y y y n
Cray XT3 2.0.62 y y y y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit y y y y
Windows XP y y(4) y y
Windows XP x64 y y(4) y y
Windows Vista y n n y
Mac OS X 10.5 y n y n
AIX 5.3 32- and 64-bit n n n n
FreeBSD 6.3-STABLE 32&64 bit y y y y
RedHat EL5 2.6.18-164 i686 GNU (1)W y y(2) y y
RedHat EL5 2.6.18-164 i686 Intel W y y y n
RedHat EL5 2.6.18-164 i686 PGI W y y y n
RedHat EL5 2.6.18-164 x86_64 GNU(1)W y y y y
RedHat EL5 2.6.18-164 x86_64 IntelW y y y n
RedHat EL5 2.6.18-164 x86_64 PGI W y y y n
SuSe Linux 2.6.16 SGI Altix ia64 C y n
RedHat EL4 2.6.18 Xeon Lustre C y y y n
SuSe Linux 2.4.21 ia64 Intel C y y y n
Cray XT3 2.0.62 n n n n
(1) Fortran compiled with g95.
(2) With PGI and Absoft compilers.
(3) With PGI compiler for Fortran.
(4) Using Visual Studio 2005 or Cygwin
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* Parallel mode in AIX will fail some of the testcheck_version.sh tests where
it treats "exit(134) the same as if process 0 had received an abort signal.
This is fixed and will be available in the next release. AKC - 2009/11/3
* Some tests in tools/h5repack may fail in AIX systems when -q32 mode is used.
The error is due to insufficient memory requested. Request a large amount
of runtime memory by setting the following environment variable for more
memory.
LDR_CNTRL=MAXDATA=0x20000000@DSA
AKC - 2009/10/31
* The PathScale MPI implementation, accessing a Panasas file system, would
cause H5Fcreate() with H5F_ACC_EXCL to fail even when the file is not
existing. This is due to the MPI_File_open() call failing if the amode has
the MPI_MODE_EXCL bit set. (See bug 1468 for details.) AKC - 2009/8/11
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO. CMC - 2009/04/28
* There is a known issue in which HDF5 will change the timestamp on a file
simply by opening it with read/write permissions, even if the file is not
modified in any way. This is due to the way in which HDF5 manages the file
superblock. A fix is currently underway and should be included in the 1.8.4
release of HDF5. MAM - 2009/04/28
* For gcc v4.3 and v4.4, with production mode, if -O3 is used, H5Tinit.c
would fail to compile. Actually bad H5Tinit.c is produced. If -O (same
as -O1) is used, H5Tinit.c compiled okay but test/dt_arith would fail.
When -O0 (no optimizatio) is used, H5Tinit.c compilete okay and all
tests passed. Therefore, -O0 is imposed for v4.3 and v4.4 of gcc.
AKC - 2009/04/20
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh and
tools/h5copy/testh5copy.sh will fail some of its sub-tests. These sub-tests
are expected to fail and should exit with a non-zero code but the yod
command does not propagate the exit code of the executables. Yod always
returns 0 if it can launch the executable. The test suite shell expects
a non-zero for this particular test, therefore it concludes the test has
failed when it receives 0 from yod. Skip all the "failing" test for now
by changing them as following.
======== Original tools/h5ls/testh5ls.sh =========
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Change to ===============================
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
==================================================
======== Original tools/h5copy/testh5copy.sh =========
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
H5LSTEST $FILEOUT
======== Change to ===============================
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
echo SKIP H5LSTEST $FILEOUT
==================================================
AKC - 2008/11/10
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message, "yod allocation delayed for node recovery". This interferes with
test suites that do not expect seeing this message. See the section of "Red
Storm" in file INSTALL_parallel for a way to deal with this problem.
AKC - 2008/05/28
* On Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
SLU - 2005/6/30
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use",
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
* The --enable-static-exec configure flag will only statically link libraries
if the static version of that library is present. If only the shared version
of a library exists (i.e., most system libraries on Solaris, AIX, and Mac,
for example, only have shared versions), the flag should still result in a
successful compilation, but note that the installed executables will not be
fully static. Thus, the only guarantee on these systems is that the
executable is statically linked with just the HDF5 library.
* There is also a configure error on Altix machines that incorrectly reports
when a version of Szip without an encoder is being used.
%%%%1.8.3%%%%
HDF5 version 1.8.3 released on Mon May 4 09:21:00 CDT 2009
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.2 and
HDF5 1.8.3, and contains information on the platforms tested and
known problems in HDF5-1.8.3.
For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.3 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.3 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.3 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.3 (current
release) versus Release 1.8.2":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for New Platforms, Languages, and Compilers
- Bug Fixes since HDF5-1.8.2
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- Added libtool version numbers to generated c++, fortran, and
hl libraries. MAM 2009/04/19.
- Regenerated Makefile.ins using Automake 1.10.2. MAM 2009/04/19.
- Added a Make target of check-all-install to test the correctness of
installing via the prefix= or $DESTDIR options. AKC - 2009/04/14
Library
-------
- Embed the content of libhdf5.settings into the hdf5 executables
so that an "orphaned" executables can display (via the Unix
strings command, for example) the library settings used to build
the executables. This is a prototype implementation. Improvement will
be added in next release. AKC - 2009/04/20
- Separated "factory" free list class from block free lists. These free
lists are dynamically created and manage blocks of a fixed size.
H5set_free_list_limits() will use the same settings specified for block
free lists for factory free lists. NAF - 2009/04/08
- Added support for dense attributes to H5Ocopy. XCao/NAF - 2009/01/29
- Added H5Pset_elink_cb and H5Pget_elink_cb functions to support a
user-defined callback function for external link traversal.
NAF - 2009/01/08
- Added H5Pset_elink_acc_flags and H5Pget_elink_acc_flags functions to
allow the user to specify the file access flags used to open the target
file of an external link. NAF - 2009/01/08
- Added H5Pset_chunk_cache() and H5Pget_chunk_cache() functions to allow
individual rdcc configuration for each dataset. Added
H5Dget_access_plist() function to retrieve a dataset access property
list from a dataset. NAF - 2008/11/12
- Added H5Iis_valid() function to check if an id is valid without
producing an error message. NAF - 2008/11/5
- Added code to maintain a min_clean_fraction in the metadata cache when
in serial mode. MAM - 2009/01/9
Parallel Library
----------------
- Modified parallel tests to run with arbitrary number of processes. The
modified tests are testphdf5 (parallel dataset access), t_chunk_alloc
(chunk allocation), and t_posix_compliant (posix compliance). The rest of
the parallel tests already use in the code the number of processes
available in the communicator. (CMC - 2009/04/28)
Tools
-----
- h5diff new flag, -c, --compare, list objects that are not comparable.
PVN - 2009/4/2 - 1368
- h5diff new flag, -N, --nan, avoids NaNs detection. PVN - 2009/4/2
- h5dump correctly specifies XML dtd / schema urls ADB - 2009/4/3 - 1519
- h5repack now handles group creation order. PVN - 2009/4/2 - 1402
- h5repack: When user doesn't specify a chunk size, h5repack now
defines a default chunk size as the same size of the size of the
hyperslab used to read the chunks. The size of the hyperslabs are
defined as the size of each dimension or a predefined constant,
whatever is smaller. This assures that the chunk read fits in the
chunk cache. PVN - 2008/11/21
High-Level APIs
---------------
- Table: In version 3.0 of Table, the writing of the "NROWS" attribute
(used to store number of records) was deprecated. PVN - 2008/11/24
F90 API
-------
- Added for the C APIs the Fortran wrappers:
h5dget_access_plist_f
h5iis_valid_f
h5pset_chunk_cache_f
h5pget_chunk_cache_f
MSB - 2009/04/17
C++ API
-------
- None
Support for New Platforms, Languages, and Compilers
===================================================
Bug Fixes since HDF5-1.8.2
==========================
Configuration
-------------
- The --includedir=DIR configuration option now works as intended, and
can be used to specify the location to install C header files. The
default location remains unchanged, residing at ${prefix}/include.
MAM - 2009/03/10 - BZ #1381
- Configure no longer removes the '-g' flag from CFLAGS when in production
mode if it has been explicitly set in the CFLAGS environment variable
prior to configuration. MAM - 2009/03/09 - BZ #1401
Library
-------
- Added versioning to H5Z_class_t struct to allow compatibility with 1.6
API. NAF - 2009/04/20 - 1533
- Fixed a problem with using data transforms with non-native types in the
file. NAF - 2009/04/20 - 1548
- Added direct.h include file to windows section of H5private.h
to fix _getcwd() warning. ADB - 2009/04/14 - 1536
- Fixed a bug that prevented external links from working after calling
H5close(). NAF - 2009/04/10 - 1539
- Modified library to write cached symbol table information to the
superblock, to allow library versions 1.3.0 to 1.6.3 to read files created
by this version. NAF - 2009/04/08 - 1423
- Changed skip lists to use a deterministic algorithm. The library should
now never call rand() or srand(). NAF - 2009/04/08 - 503
- Fixed a bug where H5Lcopy and H5Lmove wouldn't create intermediate groups
when that property was set. NAF - 2009/04/07 - 1526
- Fixed a bug that caused files with a user block to grow by the size of the
user block every time they were opened. NAF - 2009/03/26 - 1499
- Fixed a rare problem that could occur with files using the old (pre 1.4)
array datatype. NAF - 2009/03/23
- Modified library to be able to open files with corrupt root group symbol
table messages, and correct these errors if they are found. Such files
can only be successfully opened with write access. NAF - 2009/03/23 - 1189
- Removed the long_long #define and replaced all instances with
"long long". This caused problems with third party products. All
currently supported compliers support the type. ADB - 2009/03/05
- Fixed various bugs that could prevent the fill value from being written
in certain rare cases. NAF - 2009/02/26 - 1469
- Fixed a bug that prevented more than one dataset chunk from being cached
at a time. NAF - 2009/02/12 - 1015
- Fixed an assertion failure caused by opening an attribute multiple times
through multiple file handles. NAF - 2009/02/12 - 1420
- Fixed a problem that could prevent the user from adding attributes (or any
object header message) in some circumstances. NAF - 2009/02/12 - 1427
- Fixed a bug that could cause problems when an attribute was added to a
committed datatype using the committed datatype's datatype.
NAF - 2009/02/12
- Fixed a bug that could cause problems when copying an object with a shared
message in its own object header. NAF - 2009/01/29
- Changed H5Tset_order to properly reject H5T_ORDER_NONE for most datatypes.
NAF - 2009/01/27 - 1443
- Fixed a bug where H5Tpack wouldn't remove trailing space from an otherwise
packed compound type. NAF - 2009/01/14
- Fixed up some old v2 btree assertions that get run in debug mode that
were previously failing on compilation, and removed some of the
more heavily outdated and non-rewritable ones. MAM - 2008/12/15
- Fixed a bug that could cause problems when "automatically" unmounting
multiple files. NAF - 2008/11/17
- H5Dset_extent: when shrinking dimensions, some chunks were not deleted.
PVN - 2009/01/8
Parallel Library
----------------
- None
Tools
-----
- Fixed many problems that could occur when using h5repack with named
datatypes. NAF - 2009/4/20 - 1516/1466
- h5dump, h5diff, h5repack were not reading (by hyperslabs) datasets
that have a datatype datum size greater than H5TOOLS_BUFSIZE, a constant
defined as 1024Kb, such as array types with large dimensions.
PVN - 2009/4/1 - 1501
- h5import: By selecting a compression type, a big endian byte order
was being selected. PVN - 2009/3/11 - 1462
- zip_perf.c had missing argument on one of the open() calls. Fixed.
AKC - 2008/12/9
F90 API
------
- None
C++ API
------
- None
High-Level APIs:
------
- Dimension scales: The scale index return value in H5DSiterate_scales
was not always incremented. PVN - 2009/4/8 - 1538
Fortran High-Level APIs:
------
- Lite: The h5ltget_dataset_info_f function (gets information about
a dataset) was not correctly returning the dimension array
PVN - 2009/3/23
Platforms Tested
================
The following platforms and compilers have been tested for this release.
AIX 5.3 xlc 7.0.0.8
(LLNL Up) xlf 09.01.0000.0008
xlC 7.0.0.8
mpcc_r 7.0.0.8
mpxlf_r 09.01.0000.0008
Cray XT3 (2.0.41) cc (pgcc) 7.1-4
(SNL red storm) ftn (pgf90) 7.1-4
CC (pgCC) 7.1-4
FreeBSD 6.3-STABLE i386 gcc 3.4.6 [FreeBSD] 20060305
(duty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.3.4 20090419
g++ 4.3.4 20090419
gfortran 4.3.4 20090419
FreeBSD 6.3-STABLE amd64 gcc 3.4.6 [FreeBSD] 20060305
(liberty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.4.1 20090421
g++ 4.4.1 20090421
gfortran 4.4.1 20090421
IRIX64 6.5 (64 & n32) MIPSpro cc 7.4.4m
F90 MIPSpro 7.4.4m
C++ MIPSpro cc 7.4.4m
Linux 2.6.18-128.1.6.el5xen gcc (GCC) 4.1.2
#1 SMP i686 i686 i386 G95 (GCC 4.0.3 (g95 0.92!) Feb 4 2009)
(jam) PGI C, Fortran, C++ 7.2-1 32-bit
PGI C, Fortran, C++ 8.0-1 32-bit
Intel(R) C Compiler for 32-bit
applications, Versions 10.1, 11.0
Intel(R) C++ Compiler for 32-bit
applications, Version 10.1, 11.0
Intel(R) Fortran Compiler for 32-bit
applications, Version 10.1, 11.0
Absoft 32-bit Fortran 95 10.0.7
MPICH mpich2-1.0.8 compiled with
gcc 4.1.2 and G95 (GCC 4.0.3 (g95 0.92!)
Linux 2.6.9-42.0.10.ELsmp #1 gcc (GCC) 3.4.6
SMP i686 i686 i386 G95 (GCC 4.0.3 (g95 0.92!) Feb 4 2009)
(kagiso) MPICH mpich2-1.0.8 compiled with
gcc 3.4.6 and G95 (GCC 4.0.3 (g95 0.92!)
Linux 2.6.16.60-0.37-smp #1 gcc 4.1.2
SMP x86_64 GNU/Linux G95 (GCC 4.0.3 (g95 0.92!) Feb 4 2009)
(smirom) Intel(R) C, C++, Fortran Compilers for
applications running on Intel(R) 64,
Versions 10.1, 11.0.
PGI C, Fortran, C++ Version 7.2-1, 8.0-1
for 64-bit target on x86-64
gcc 4.1.2 and G95 (GCC 4.0.3 (g95 0.92!)
MPICH mpich2-1.0.8 compiled with
gcc 4.1.2 and G95 (GCC 4.0.3 (g95 0.92!)
tested for both 32- and 64-bit binaries
Linux 2.6.16.54-0.2.5 #1 Intel(R) C++ Version 10.1.017
SGI Altix SMP ia64 Intel(R) Fortran Itanium(R) Version 10.1.017
(cobalt) SGI MPI 1.38
SunOS 5.10 32- and 64-bit Sun WorkShop 6 update 2 C 5.9 Patch 124867-09
(linew) Sun WorkShop 6 update 2 Fortran 95 8.3
Patch 127000-07
Sun WorkShop 6 update 2 C++ 5.8
Patch 124863-11
Intel Xeon Linux 2.6.18- gcc 3.4.6 20060404
92.1.10.el5_lustre.1.6.6smp- Intel(R) C++ Version 10.0.026
perfctr #2 SMP Intel(R) Fortran Compiler Version 10.0.026
(abe) Open MPI 1.2.2
MVAPICH2-0.9.8p28p2patched-intel-ofed-1.2
compiled with icc v10.0.026 and ifort 10.0.026
IA-64 Linux 2.4.21-309.tg1 gcc (GCC) 3.2.2
#1 SMP ia64 Intel(R) C++ Version 8.1.037
(NCSA tg-login) Intel(R) Fortran Compiler Version 8.1.033
mpich-gm-1.2.7p1..16-intel-8.1.037-r1
Linux 2.6.9-55.0.9.EL_lustre Intel(R) C, C++, Fortran Compilers for
.1.4.11.1smp #1 SMP applications running on Intel(R) 64,
SMP x86_64 GNU/Linux Versions 9.1.
(SNL Spirit)
Linux 2.6.9-55.0.9.EL_lustre Intel(R) C, C++, Fortran Compilers for
.1.4.11.1smp #1 SMP applications running on Intel(R) 64,
SMP x86_64 GNU/Linux Versions 10.1.
(SNL Thunderbird)
Linux 2.6.18-63chaos #1 SMP Intel(R) C, C++, Fortran Compilers for
SMP x86_64 GNU/Linux applications running on Intel(R) 64,
(SNL Glory) Versions 10.1.
Linux 2.6.18-63chaos #1 SMP Intel(R) C, C++, Fortran Compilers for
SMP x86_64 GNU/Linux applications running on Intel(R) 64,
(LLNL Zeus) Versions 9.1.
gcc/gfortran/g++ (GCC) 4.1.2.
Windows XP Visual Studio .NET
Visual Studio 2005 w/ Intel Fortran 9.1
Cygwin(native gcc compiler and g95)
Windows XP x64 Visual Studio 2005 w/ Intel Fortran 9.1
Windows Vista Visual Studio 2005
MAC OS 10.5.6 (Intel) i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1
GNU Fortran (GCC) 4.3.0 20070810
G95 (GCC 4.0.3 (g95 0.91!) Apr 24 2008)
Intel C, C++ and Fortran compilers 10.1
Supported Configuration Features Summary
========================================
In the tables below
y = tested and supported
n = not supported or not tested in this release
C = Cluster
W = Workstation
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
Solaris2.10 32-bit n y n y y y
Solaris2.10 64-bit n y n y y y
IRIX64_6.5 32-bit n n n n y y
IRIX64_6.5 64-bit n y y y y y
Windows XP n y(4) n(4) y y y
Windows XP x64 n y(4) n(4) y y y
Windows Vista n n n y y y
Mac OS X 10.5 Intel n y n y y y
AIX 5.3 32- and 64-bit n y n y y n
FreeBSD 6.3-STABLE 32&64 bit n y n y y y
RedHat EL4 2.6.9-42 i686 GNU (1) W y y y y y y
RedHat EL5 2.6.18-128 i686 GNU (1)W y y(2) y y y y
RedHat EL5 2.6.18-128 i686 Intel W n y n y y n
RedHat EL5 2.6.18-128 i686 PGI W n y n y y n
SuSe Linux 2.6.16 x86_64 GNU (1) W y y(3) y y y y
SuSe Linux 2.6.16 x86_64 Intel W n y n y y n
SuSe Linux 2.6.16 x86_64 PGI W n y n y y y
SuSe Linux 2.6.16 SGI Altix ia64 C y y y y y y
RedHat EL4 2.6.18 Xeon Lustre C y y y y y n
SuSe Linux 2.4.21 ia64 Intel C y y y y y n
Cray XT3 2.0.41 y y y y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit y y y y
IRIX64_6.5 32-bit y dna y y
IRIX64_6.5 64-bit y y n y
Windows XP y y(4) y y
Windows XP x64 y y(4) y y
Windows Vista y n n y
Mac OS X 10.5 y n y n
AIX 5.3 32- and 64-bit n n n n
FreeBSD 6.3-STABLE 32&64 bit y n y y
RedHat EL4 2.6.9-42 i686 GNU (1) W y y y y
RedHat EL5 2.6.18-128 i686 GNU (1)W y y(2) y y
RedHat EL5 2.6.18-128 i686 Intel W y y y n
RedHat EL5 2.6.18-128 i686 PGI W y y y n
SuSe Linux 2.6.16 x86_64 GNU (1) W y y y y
SuSe Linux 2.6.16 x86_64 Intel W y y y n
SuSe Linux 2.6.16 x86_64 PGI W y y y n
SuSe Linux 2.6.16 SGI Altix ia64 C y n
RedHat EL4 2.6.18 Xeon Lustre C y y y n
SuSe Linux 2.4.21 ia64 Intel C y y y n
Cray XT3 2.0.41 n n n n
(1) Fortran compiled with g95.
(2) With PGI and Absoft compilers.
(3) With PGI compiler for Fortran.
(4) Using Visual Studio 2005 or Cygwin
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* Parallel tests failed with 16 processes with data inconsistency at testphdf5
/ dataset_readAll. Parallel tests also failed with 32 and 64 processes with
collective abort of all ranks at t_posix_compliant / allwrite_allread_blocks
with MPI IO. CMC - 2009/04/28
* There is a known issue in which HDF5 will change the timestamp on a file
simply by opening it with read/write permissions, even if the file is not
modified in any way. This is due to the way in which HDF5 manages the file
superblock. A fix is currently underway and should be included in the 1.8.4
release of HDF5. MAM - 2009/04/28
* For gcc v4.3 and v4.4, with production mode, if -O3 is used, H5Tinit.c
would fail to compile. Actually bad H5Tinit.c is produced. If -O (same
as -O1) is used, H5Tinit.c compiled okay but test/dt_arith would fail.
When -O0 (no optimizatio) is used, H5Tinit.c compilete okay and all
tests passed. Therefore, -O0 is imposed for v4.3 and v4.4 of gcc.
AKC - 2009/04/20
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh and
tools/h5copy/testh5copy.sh will fail some of its sub-tests. These sub-tests
are expected to fail and should exit with a non-zero code but the yod
command does not propagate the exit code of the executables. Yod always
returns 0 if it can launch the executable. The test suite shell expects
a non-zero for this particular test, therefore it concludes the test has
failed when it receives 0 from yod. Skip all the "failing" test for now
by changing them as following.
======== Original tools/h5ls/testh5ls.sh =========
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Change to ===============================
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
==================================================
======== Original tools/h5copy/testh5copy.sh =========
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
H5LSTEST $FILEOUT
======== Change to ===============================
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
echo SKIP H5LSTEST $FILEOUT
==================================================
AKC - 2008/11/10
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message, "yod allocation delayed for node recovery". This interferes with
test suites that do not expect seeing this message. See the section of "Red
Storm" in file INSTALL_parallel for a way to deal with this problem.
AKC - 2008/05/28
* We have discovered two problems when running collective IO parallel HDF5
tests with chunking storage on the ChaMPIon MPI compiler on tungsten, a
Linux cluster at NCSA.
Under some complex selection cases:
1) MPI_Get_element returns the wrong value.
2) MPI_Type_struct also generates the wrong derived datatype and corrupt
data may be generated.
These issues arise only when turning on collective IO with chunking storage
with some complex selections. We have not found these problems on other
MPI-IO compilers. If you encounter these problems, you may use independent
IO instead.
To avoid this behavior, change the following line in your code
H5Pset_dxpl_mpio(xfer_plist, H5FD_MPIO_COLLECTIVE);
to
H5Pset_dxpl_mpio(xfer_plist, H5FD_MPIO_INDEPENDENT);
KY - 2007/08/24
* On Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* For LLNL, uP: both serial and parallel tests pass.
Zeus: Serial tests pass but parallel tests fail with a known problem in MPI.
ubgl: Serial tests pass but parallel tests fail.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* On IRIX6.5, when the C compiler version is greater than 7.4, complicated
MPI derived datatype code will work. However, the user should increase
the value of the MPI_TYPE_MAX environment variable to some appropriate value
to use collective irregular selection code. For example, the current
parallel HDF5 test needs to raise MPI_TYPE_MAX to 200,000 to pass the test.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
SLU - 2005/6/30
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use",
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
* The --enable-static-exec configure flag fails to compile for Solaris
platforms. This is due to the fact that not all of the system libraries on
Solaris are available in a static format.
The --enable-static-exec configure flag also fails to correctly compile
on IBM SP2 platforms for serial mode. The parallel mode works fine with
this option.
It is suggested that you do not use this option on these platforms
during configuration.
* There is also a configure error on Altix machines that incorrectly reports
when a version of Szip without an encoder is being used.
* Information about building with PGI and Intel compilers is available in
the INSTALL file sections 4.7 and 4.8.
%%%%1.8.2%%%%
HDF5 version 1.8.2 released on Mon Nov 10 15:43:09 CST 2008
================================================================================
INTRODUCTION
============
This document describes the differences between HDF5-1.8.1 and HDF5 1.8.2,
and contains information on the platforms tested and known problems in
HDF5-1.8.2. For more details, see the files HISTORY-1_0-1_8_0_rc3.txt
and HISTORY-1_8.txt in the release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.2 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.2 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.2 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.2 (current
release) versus Release 1.8.1":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for new platforms and languages
- Bug Fixes since HDF5-1.8.1
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- Upgraded libtool to version 2.2.6a. (MAM - 2008/10/15).
Library
-------
- Added two new public routines: H5Pget_elink_fapl() and
H5Pset_elink_fapl(). (see bug #1247) (VC - 2008/10/13)
- Improved free space tracking in file to be faster. (QAK - 2008/10/06)
- Added 'mounted' field to H5G_info_t struct. (QAK - 2008/07/15)
Parallel Library
----------------
- None
Tools
-----
- h5repack: added new options -u and -b to add a userblock to an HDF5
file during the repack. (PVN - 2008/08/26)
- h5repack: added options -t and -a to call H5Pset_alignment while
creating a repacked file. (PVN - 2008/08/29)
- h5ls: added capability to traverse through external links when the -r
(recursive) flag is given. (NAF - 2008/09/16)
- h5ls: added -E option to enable traversal of external links.
h5ls will not traverse external links without this flag being set.
(NAF - 2008/10/06)
- h5dump: when -b flag is used without a keyword after it, binary
output defaults to NATIVE. MEMORY keyword was deprecated
and replaced by NATIVE keyword. (PVN - 2008/10/30)
- h5diff: returns 1 when file graphs differ by any object.
Error return code was changed to 2 from -1. (PVN - 2008/10/30)
- h5import: TEXTFPE (scientific format) was deprecated. Use TEXTFP
instead (PVN - 2008/10/30)
F90 API
------
- Added optional parameter 'mounted' to H5Gget_info_f,
H5Gget_info_by_idx_f, H5Gget_info_by_name_f (MSB - 2008/09/24)
- Added H5Tget_native_type_f (MSB - 2008/09/30)
C++ API
------
- These member functions were added as wrapper for H5Rdereference to
replace the incorrect IdComponent::dereference().
void H5Object::dereference(H5Object& obj, void* ref,
H5R_type_t ref_type=H5R_OBJECT)
void H5Object::dereference(H5File& h5file, void* ref,
H5R_type_t ref_type=H5R_OBJECT)
void H5Object::dereference(Attribute& obj, void* ref,
H5R_type_t ref_type=H5R_OBJECT)
In addition, these constructors were added to create the associated
objects by way of dereference:
DataSet(H5Object& obj, void* ref, H5R_type_t ref_type=H5R_OBJECT)
DataSet(H5File& file, void* ref, H5R_type_t ref_type=H5R_OBJECT)
DataSet(Attribute& attr, void* ref, H5R_type_t ref_type=H5R_OBJECT)
Group(H5Object& obj, void* ref, H5R_type_t ref_type=H5R_OBJECT)
Group(H5File& obj, void* ref, H5R_type_t ref_type=H5R_OBJECT)
Group(Attribute& attr, void* ref, H5R_type_t ref_type=H5R_OBJECT)
DataType(H5Object& obj, void* ref, H5R_type_t ref_type=H5R_OBJECT)
DataType(H5File& file, void* ref, H5R_type_t ref_type=H5R_OBJECT)
DataType(Attribute& attr, void* ref, H5R_type_t ref_type=H5R_OBJECT)
(BMR - 2008/10/29)
Support for New Platforms, Languages, and Compilers
===================================================
- Intel 10.1 is supported on Mac OS X 10.5.4.
Note:
When Fortran is enabled, configure automatically
disables the build of shared libraries (i.e., only
static C and C++ HDF5 libraries will be built
along with the static HDF5 Fortran library).
Intel 10.1 C and C++ compilers require
"-no-multibyte-chars" compilation flag due to the known
bug in the compilers.
(EIP - 2008/10/30)
Bug Fixes since HDF5-1.8.1
==========================
Configuration
-------------
- Fixed error with 'make check install' failing due to h5dump
needing other tools built first. (MAM - 2008/10/15).
- When using shared szip, it is no longer necessary to specify
the path to the shared szip libraries in LD_LIBRARY_PATH.
(MAM - 2008/10/15).
- The file libhdf5_fortran.settings is not installed since its content
is included in libhdf5.settings now. (AKC - 2008/10/21)
- "make DESTDIR=xxx install" failed to install some tools and files
(e.g., h5cc and fortran modules). Fixed. (AKC - 2008/10/8).
Library
-------
- H5Ovisit and H5Ovisit_by_name will now properly terminate when the
callback function returns a positive value on the starting object.
(NAF - 2008/11/03)
- Fixed an error where a null message could be created that was larger
than could be written to the file. (NAF - 2008/10/23)
- Corrected error with family/split/multi VFD not updating driver info
when "latest" version of the file format used. (QAK - 2008/10/14)
- Corrected alignment+threshold errors to work correctly when metadata
aggregation is enabled. (QAK - 2008/10/06)
- Changed H5Fget_obj_count and H5Fget_obj_ids to ignore objects
registered by the library for internal library use.
(NAF - 2008/10/06)
- Fixed potential memory leak during compound conversion.
(NAF - 2008/10/06)
- Changed the return value of H5Fget_obj_count from INT to SSIZE_T.
Also changed the return value of H5Fget_obj_ids from HERR_T to
SSIZE_T and the type of the parameter MAX_OBJS from INT to SIZE_T.
(SLU - 2008/09/26)
- Fixed an issue that could cause data to be improperly overwritten
during compound type conversion. (NAF - 2008/09/19)
- Fixed pointer alignment violations that could occur during vlen
conversion. (NAF - 2008/09/16)
- Fixed problem where library could cause a segmentation fault when
an invalid location ID was given to H5Giterate(). (QAK - 2008/08/19)
- Fixed improper shutdown when objects have reference count > 1. The
library now tracks reference count due to the application separately
from that due to internal library routines. (NAF - 2008/08/19)
- Fixed assertion failure caused by incorrect array datatype version.
(NAF - 2008/08/08)
- Fixed an issue where mount point traversal would fail when using
multiple handles for the child. (NAF - 2008/08/07)
- Fixed an issue where mount points were inaccessible when using
multiple file handles for the parent. The mount table is now in
the shared file structure (the parent pointer is still in the
top structure). (NAF - 2008/08/07)
- Fixed assertion failure caused by incorrect array datatype version.
(NAF - 2008/08/04)
- Fixed issue where a group could have a file mounted on it twice.
(QAK - 2008/07/15)
- When an attribute was opened twice and data was written with
one of the handles, the file didn't have the data. It happened
because each handle had its own object structure, and the empty
one overwrote the data with fill value. This is fixed by making
some attribute information like the data be shared in the
attribute structure. (SLU - 2008/07/07)
- Fixed a Windows-specific issue in the ohdr test which was causing
users in some timezones to get false errors. This a deficiency in
the Windows mktime() function, and has been handled properly.
(SJW - 2008/06/19)
Parallel Library
----------------
- None
Tools
-----
- h5dump now checks for uniqueness of committed datatypes.
(NAF - 2008/10/15)
- Fixed unnecessary indentation of committed datatypes in h5dump.
(NAF - 2008/10/15)
- Fixed bugs in h5stat: segmemtation fault when printing groups and
print warning message when traversal of objects is unsuccessful.
(see bug #1253) (VC- 2008/10/13)
- Fixed bug in h5ls that prevented relative group listings (like
"h5ls foo.h5/bar") from working correctly (QAK - 2008/06/03)
- h5dump: when doing binary output (-b), the stdout printing of
attributes was done incorrectly. Removed printing of attributes
when doing binary output. (PVN - 2008/06/05)
F90 API
------
- h5sselect_elements_f: Added additional operators H5S_SELECT_APPEND
and H5S_SELECT_PREPEND (MSB - 2008/09/30)
- h5sget_select_elem_pointlist: Fixed list of returned points by
rearranging the point list correctly by accounting for C
conventions. (MSB - 2008/09/30)
- h5sget_select_hyper_blocklist_f: Fixed error in transposed dimension
of arrays.(MSB - 2008/9/30)
- h5sget_select_bounds_f: Swapped array bounds to account for C and
Fortran reversed array notation (MSB - 2008/9/30)
- Changed to initializing string to a blank character instead of a
null type in tH5P.f90 to fix compiling error using AIX 5.3.0
(MSB - 2008/7/29)
- Fixed missing commas in H5test_kind.f90 detected by NAG compiler
(MSB - 2008/7/29)
- Fixed passing and array to a scalar in tH5A_1_8.f90 detected by
NAG compiler (MSB - 2008/7/29)
- Added the ability of the test programs to use the status of
HDF5_NOCLEANUP to determine if the *.h5 files should be removed
or not after the tests are completed (MSB - 2008/10/1)
- In nh5tget_offset_c: (MSB 9/12/2008)
If offset was equal to 0 it returned the error code of -1,
this was changed to return an error code of -1 when the offset
value is < 0.
- Uses intrinsic Fortran function SIZEOF if available when detecting
type of INTEGERs and REALs in H5test_kind.f90 (MSB - 2008/9/3)
- Put the DOUBLE PRECISION interfaces in a separate module and
added a USE statement for the module. The interfaces are
included/excluded depending on the state of FORTRAN_DEFAULT_REAL
is DBLE_F which detects if the default REAL is DOUBLE PRECISION.
This allows the library to be compiled with -r8 Fortran flag
without the user needing to edit the source code.
(MSB - 200/8/27)
- Enable building shared library for fortran by adding the flag -fPIC
to the compile flags for versions of Intel Fortran compiler >=9
(MSB - 2008/8/26)
C++ API
------
- Fixed a design bug which allowed an Attribute object to create/modify
attributes (bugzilla #1068). The API class hierarchy was revised
to address the problem. Classes AbstractDS and Attribute are moved
out of H5Object. Class Attribute now multiply inherits from
IdComponent and AbstractDs and class DataSet from H5Object and
AbstractDs. In addition, the data member IdComponent::id was
moved into subclasses: Attribute, DataSet, DataSpace, DataType,
H5File, Group, and PropList. (BMR - 2008/05/20)
- IdComponent::dereference was incorrect and replaced as described
in "New Features" section.
(BMR - 2008/10/29)
Platforms Tested
================
The following platforms and compilers have been tested for this release.
AIX 5.3 xlc 7.0.0.8
xlf 09.01.0000.0008
xlC 7.0.0.8
mpcc_r 7.0.0.8
mpxlf_r 09.01.0000.0008
Cray XT3 (2.0.41) cc (pgcc) 7.1-4
(red storm) ftn (pgf90) 7.1-4
CC (pgCC) 7.1-4
FreeBSD 6.3-STABLE i386 gcc 3.4.6 [FreeBSD] 20060305
(duty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.2.5 20080702
g++ 4.2.5 20080702
gfortran 4.2.5 20080702
FreeBSD 6.3-STABLE amd64 gcc 3.4.6 [FreeBSD] 20060305
(liberty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.2.5 20080702
g++ 4.2.5 20080702
gfortran 4.2.5 20080702
IRIX64 6.5 (64 & n32) MIPSpro cc 7.4.4m
F90 MIPSpro 7.4.4m
C++ MIPSpro cc 7.4.4m
Linux 2.6.9-42.0.10.ELsmp #1 gcc (GCC) 3.4.6
SMP i686 i386 G95 (GCC 4.0.3 (g95 0.92!) April 18 2007)
(kagiso) PGI C, Fortran, C++ 7.2-1 32-bit
Intel(R) C Compiler for 32-bit
applications, Version 10.1
Intel(R) C++ Compiler for 32-bit
applications, Version 10.1
Intel(R) Fortran Compiler for 32-bit
applications, Version 10.1
Absoft 32-bit Fortran 95 10.0.4
MPICH mpich-1.2.7 compiled with
gcc 3.4.6 and G95 (GCC 4.0.3 (g95 0.92!)
MPICH mpich2-1.0.6p1 compiled with
gcc 3.4.6 and G95 (GCC 4.0.3 (g95 0.92!)
Linux 2.6.16.46-0.14-smp #1 Intel(R) C++ for Intel(R) EM64T
SMP x86_64 GNU/Linux Ver. 10.1.013
(smirom) Intel(R) Fortran Intel(R) EM64T
Ver. 10.1.013
PGI C, Fortran, C++ Version 7.2-1
for 64-bit target on x86-64
MPICH mpich-1.2.7 compiled with
gcc 4.1.2 and G95 (GCC 4.0.3 (g95 0.92!)
MPICH mpich2-1.0.7 compiled with
gcc 4.1.2 and G95 (GCC 4.0.3 (g95 0.92!)
tested for both 32- and 64-bit binaries
Linux 2.6.16.54-0.2.5 #1 Intel(R) C++ Version 10.1.017
Altix SMP ia64 Intel(R) Fortran Itanium(R) Version 10.1.017
(cobalt) SGI MPI 1.16
SunOS 5.10 32- and 64-bit Sun WorkShop 6 update 2 C 5.8
(linew) Sun WorkShop 6 update 2 Fortran 95 8.2
Sun WorkShop 6 update 2 C++ 5.8
Patch 121019-06
Xeon Linux 2.6.9-42.0.10.EL_lustre-1.4.10.1smp
(abe) Intel(R) C++ Version 10.0.026
Intel(R) Fortran Compiler Version 10.0.026
Open MPI 1.2.2
MVAPICH2-0.9.8p28p2patched-intel-ofed-1.2
compiled with icc v10.0.026 and
ifort 10.0.026
IA-64 Linux 2.4.21-309.tg1 #1 SMP
ia64 gcc (GCC) 3.2.2
(NCSA tg-login) Intel(R) C++ Version 8.1.037
Intel(R) Fortran Compiler Version 8.1.033
mpich-gm-1.2.7p1..16-intel-8.1.037-r1
Intel 64 Linux 2.6.9-42.0.10.EL_lustre-1.4.10.1smp
(abe) gcc 3.4.6 20060404
Intel(R) C++ Version 10.0
Intel (R) Fortran Compiler Version 10.0
mvapich2-0.9.8p2patched-intel-ofed-1.2
Windows XP Visual Studio .NET
Visual Studio 2005 w/ Intel Fortran 9.1
Cygwin(native gcc compiler and g95)
Windows XP x64 Visual Studio 2005 w/ Intel Fortran 9.1
Windows Vista Visual Studio 2005
MAC OS 10.5.4 (Intel) i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1
GNU Fortran (GCC) 4.3.0 20070810
G95 (GCC 4.0.3 (g95 0.91!) Apr 24 2008)
Intel C, C++ and Fortran compilers 10.1
Supported Configuration Features Summary
========================================
In the tables below
y = tested and supported
n = not supported or not tested in this release
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
Solaris2.10 32-bit n y n y y y
Solaris2.10 64-bit n y n y y y
IRIX64_6.5 32-bit n n n n y y
IRIX64_6.5 64-bit n y y y y y
Windows XP n y(15) n(15) y y y
Windows XP x64 n y(15) n(15) y y y
Windows Vista n n n y y y
Mac OS X 10.5 Intel n y n y y y
AIX 5.3 32- and 64-bit n y n y y n
FreeBSD 6.3-STABLE
32&64 bit n y n y y y
RedHat EL4 (3) W y(1) y(10) y(1) y y y
RedHat EL4 Intel (3) W n y n y y n
RedHat EL4 PGI (3) W n y n y y n
SuSe x86_64 gcc(3,12) W y(2) y(11) y(2) y y y
SuSe x86_64 Int(3,12) W n y(13) n y y n
SuSe x86_64 PGI(3,12) W n y(8) n y y y
Linux 2.6 SuSE ia64 C
Intel (3,7) y y y y y n
Linux 2.6 SGI Altix
ia64 Intel (3) y y y y y y
Linux 2.6 RHEL C
Lustre Intel (5) y(4) y y(4) y y n
Cray XT3 2.0.41 y y y y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 32-bit y y y y
Solaris2.10 64-bit y y y y
IRIX64_6.5 32-bit y dna y y
IRIX64_6.5 64-bit y y n y
Windows XP y y(15) y y
Windows XP x64 y y(15) y y
Windows Vista y n n y
Mac OS X 10.5 y n y n
AIX 5.3 32- and 64-bit n n n n
FreeBSD 6.2 32&64 bit y n y y
RedHat EL4 (3) W y y(10) y y
RedHat EL4 Intel (3) W y y y n
RedHat EL4 PGI (3) W y y y n
SuSe x86_64 GNU(3,12) W y y y y
SuSe x86_64 Int(3,12) W y y y n
SuSe x86_64 PGI(3,12) W y y y n
Linux 2.4 SuSE C
ia64 C Intel (7) y y y n
Linux 2.4 SGI Altix C
ia64 Intel y n
Linux 2.6 RHEL C
Lustre Intel (5) y y y n
Cray XT3 2.0.41 n n n n
Notes: (1) Using mpich2 1.0.6.
(2) Using mpich2 1.0.7.
(3) Linux 2.6 with GNU, Intel, and PGI compilers, as indicated.
W or C indicates workstation or cluster, respectively.
(4) Using mvapich2 0.9.8.
(5) Linux 2.6.9-42.0.10. Xeon cluster with ELsmp_perfctr_lustre
and Intel compilers
(6) Linux 2.4.21-32.0.1. Xeon cluster with ELsmp_perfctr_lustre
and Intel compilers
(7) Linux 2.4.21, SuSE_292.till. Ia64 cluster with Intel compilers
(8) pgf90
(9) With Compaq Visual Fortran 6.6c compiler.
(10) With PGI and Absoft compilers.
(11) PGI and Intel compilers for both C and Fortran
(12) AMD Opteron x86_64
(13) ifort
(14) Yes with C and Fortran, but not with C++
(15) Using Visual Studio 2005 or Cygwin
(16) Not tested for this release.
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh and
tools/h5copy/testh5copy.sh will fail some of its sub-tests. These sub-tests
are expected to fail and should exit with a non-zero code but the yod
command does not propagate the exit code of the executables. Yod always
returns 0 if it can launch the executable. The test suite shell expects
a non-zero for this particular test, therefore it concludes the test has
failed when it receives 0 from yod. Skip all the "failing" test for now
by changing them as following.
======== Original tools/h5ls/testh5ls.sh =========
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Change to ===============================
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
==================================================
======== Original tools/h5copy/testh5copy.sh =========
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
H5LSTEST $FILEOUT
======== Change to ===============================
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d grp_rename
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -v -s grp_dsets -d /grp_rename/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_dsets -d /E/F/grp_dsets
echo SKIP TOOLTEST_FAIL -i $TESTFILE -o $FILEOUT -vp -s /grp_nested -d /G/H/grp_nested
echo SKIP H5LSTEST $FILEOUT
==================================================
AKC - 2008/11/10
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message, "yod allocation delayed for node recovery". This interferes with
test suites that do not expect seeing this message. See the section of "Red
Storm" in file INSTALL_parallel for a way to deal with this problem.
AKC - 2008/05/28
* We have discovered two problems when running collective IO parallel HDF5
tests with chunking storage on the ChaMPIon MPI compiler on tungsten, a
Linux cluster at NCSA.
Under some complex selection cases:
1) MPI_Get_element returns the wrong value.
2) MPI_Type_struct also generates the wrong derived datatype and corrupt
data may be generated.
These issues arise only when turning on collective IO with chunking storage
with some complex selections. We have not found these problems on other
MPI-IO compilers. If you encounter these problems, you may use independent
IO instead.
To avoid this behavior, change the following line in your code
H5Pset_dxpl_mpio(xfer_plist, H5FD_MPIO_COLLECTIVE);
to
H5Pset_dxpl_mpio(xfer_plist, H5FD_MPIO_INDEPENDENT);
KY - 2007/08/24
* On Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* For LLNL, uP: both serial and parallel tests pass.
Zeus: Serial tests pass but parallel tests fail with a known problem in MPI.
ubgl: Serial tests pass but parallel tests fail.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* On IRIX6.5, when the C compiler version is greater than 7.4, complicated
MPI derived datatype code will work. However, the user should increase
the value of the MPI_TYPE_MAX environment variable to some appropriate value
to use collective irregular selection code. For example, the current
parallel HDF5 test needs to raise MPI_TYPE_MAX to 200,000 to pass the test.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
SLU - 2005/6/30
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use",
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
* The --enable-static-exec configure flag fails to compile for Solaris
platforms. This is due to the fact that not all of the system libraries on
Solaris are available in a static format.
The --enable-static-exec configure flag also fails to correctly compile
on IBM SP2 platforms for serial mode. The parallel mode works fine with
this option.
It is suggested that you do not use this option on these platforms
during configuration.
* There is also a configure error on Altix machines that incorrectly reports
when a version of Szip without an encoder is being used.
* Information about building with PGI and Intel compilers is available in
the INSTALL file sections 4.7 and 4.8.
%%%%1.8.1%%%%
HDF5 version 1.8.1 released on Thu May 29 15:28:55 CDT 2008
================================================================================
INTRODUCTION
============
This document describes the differences between the HDF5-1.8.1 release
and HDF5 1.8.0, and contains information on the platforms tested and known
problems in HDF5-1.8.1. For more details, see the files
HISTORY-1_0-1_8_0_rc3.txt and HISTORY-1_8.txt in the release_docs/ directory
of the HDF5 source.
Links to the HDF5 1.8.1 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.1 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.1 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in the HDF5-1.8.x release series, including brief general
descriptions of some new and modified APIs, are described in the "What's New
in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.1 (current
release) versus Release 1.8.0":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Support for new platforms and languages
- Bug Fixes since HDF5-1.8.0
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
Configuration
-------------
- The lib/libhdf5.settings file contains much more configure
information. (AKC - 2008/05/18)
- The new configure option "--disable-sharedlib-rpath" disables
embedding the '-Wl,-rpath' information into executables when
shared libraries are produced, and instead solely relies on the
information in LD_LIBRARY_PATH. (MAM - 2008/05/15)
- Configuration suite now uses Autoconf 2.61, Automake 1.10.1, and
Libtool 2.2.2 (MAM - 2008/05/01)
Source code distribution
========================
Library
-------
- None
Parallel Library
----------------
- None
Tools
-----
- h5repack: Reinstated the -i and -o command line flags to specify
input and output files. h5repack now understands both the old
syntax (with -i and -o) and the new syntax introduced in Release
1.8.0. (PVN - 2008/05/23)
- h5dump: Added support for external links, displaying the object that
an external link points to. (PVN - 2008/05/12)
- h5dump: Added an option, -m, to allow user-defined formatting in the
output of floating point numbers. (PVN - 2008/05/06)
- h5dump, in output of the -p option: Added effective data compression
ratio to the dataset storage layout output when a compression filter
has been applied to a dataset. (PVN - 2008/05/01)
F90 API
------
New H5A, H5G, H5L, H5O, and H5P APIs to enable 1.8 features were
added. See "Release 1.8.1 (current release) versus Release 1.8.0" in
the document "HDF5 Software Changes from Release to Release"
(http://hdfgroup.org/HDF5/doc/ADGuide/Changes.html) for the
complete list of the new APIs.
C++ API
------
- None
Support for New Platforms, Languages, and Compilers
===================================================
- Both serial and parallel HDF5 are supported for the Red Storm machine
which is a Cray XT3 system.
- The Fortran library will work correctly if compiled with the -i8
flag. This has been tested with the g95, PGI and Intel Fortran
compilers.
Bug Fixes since HDF5-1.8.0
==========================
Configuration
-------------
- None
Source code distribution
========================
Library
-------
- Chunking: Chunks greater than 4GB are disallowed.
(QAK - 2008/05/16)
- Fixed the problem with searching for a target file when following
an external link. The search pattern will depend on whether the
target file's pathname is an absolute or a relative path.
Please see the H5Lcreate_external description in the "HDF5
Reference Manual" (http://hdfgroup.org/HDF5/doc/RM/RM_H5L.html).
(VC - 2008/04/08)
- Fixed possible file corruption bug when encoding datatype
descriptions for compound datatypes whose size was between
256 and 511 bytes and the file was opened with the "use the
latest format" property enabled (with H5Pset_libver_bounds).
(QAK - 2008/03/13)
- Fixed bug in H5Aget_num_attrs() routine to correctly handle an
invalid location identifier. (QAK - 2008/03/11)
Parallel Library
----------------
- None
Tools
-----
- Fixed bug in h5diff that prevented datasets and attributes with
variable-length string elements from comparing correctly.
(QAK - 2008/02/28)
- Fixed bug in h5dump that caused binary output to be made only for
the first dataset, when several datasets were requested.
(PVN - 2008/04/07)
F90 API
------
- The h5tset(get)_fields subroutines were missing the parameter to
specify a sign position; fixed. (EIP - 2008/05/23)
- Many APIs were fixed to work with the 8-byte integers in Fortran vs.
4-byte integers in C. This change is trasparent to user applications.
C++ API
------
- The class hierarchy was revised to address the problem reported
in bugzilla #1068, Attribute should not be derived from base
class H5Object. Classes AbstractDS was moved out of H5Object.
Class Attribute now multiply inherits from IdComponent and
AbstractDs and class DataSet from H5Object and AbstractDs.
In addition, data member IdComponent::id was moved into subclasses:
Attribute, DataSet, DataSpace, DataType, H5File, Group, and PropList.
(BMR - 2008/05/20)
- IdComponent::dereference was incorrect; it was changed from:
void IdComponent::dereference(IdComponent& obj, void* ref)
to:
void H5Object::dereference(H5File& h5file, void* ref)
void H5Object::dereference(H5Object& obj, void* ref)
(BMR - 2008/05/20)
- Revised Attribute::write and Attribute::read wrappers to handle
memory allocation/deallocation properly. (bugzilla 1045)
(BMR - 2008/05/20)
Platforms Tested
================
The following platforms and compilers have been tested for this release.
Cray XT3 (2.0.41) cc (pgcc) 7.1-4
(red storm) ftn (pgf90) 7.1-4
CC (pgCC) 7.1-4
mpicc 1.0.2
mpif90 1.0.2
FreeBSD 6.2-STABLE i386 gcc 3.4.6 [FreeBSD] 20060305
(duty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.2.1 20080123
g++ 4.2.1 20080123
gfortran 4.2.1 20070620
FreeBSD 6.2-STABLE amd64 gcc 3.4.6 [FreeBSD] 20060305
(liberty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.2.1 20080123
g++ 4.2.1 20080123
gfortran 4.2.1 20080123
IRIX64 6.5 (64 & n32) MIPSpro cc 7.4.4m
F90 MIPSpro 7.4.4m
C++ MIPSpro cc 7.4.4m
Linux 2.6.9 (RHEL4) Intel 10.0 compilers
(abe.ncsa.uiuc.edu)
Linux 2.4.21-47 gcc 3.2.3 20030502
(osage)
Linux 2.6.9-42.0.10 gcc,g++ 3.4.6 20060404, G95 (GCC 4.0.3)
(kagiso) PGI 7.1-6 (pgcc, pgf90, pgCC)
Intel 9.1 (icc, ifort, icpc)
Linux 2.6.16.27 x86_64 AMD gcc 4.1.0 (SuSE Linux), g++ 4.1.0,
(smirom) g95 (GCC 4.0.3)
PGI 7.1-6 (pgcc, pgf90, pgCC)
Intel 9.1 (icc, ifort, icpc)
Linux 2.6.5-7.252.1-rtgfx #1 Intel(R) C++ Version 9.0
SMP ia64 Intel(R) Fortran Itanium(R) Version 9.0
(cobalt) SGI MPI
SunOS 5.8 32,46 Sun WorkShop 6 update 2 C 5.3
(Solaris 2.8) Sun WorkShop 6 update 2 Fortran 95 6.2
Sun WorkShop 6 update 2 C++ 5.3
SunOS 5.10 cc: Sun C 5.8
(linew) f90: Sun Fortran 95 8.2
CC: Sun C++ 5.8
Xeon Linux 2.4.21-32.0.1.ELsmp-perfctr-lustre
(tungsten) gcc 3.2.2 20030222
Intel(R) C++ Version 9.0
Intel(R) Fortran Compiler Version 9.0
IA-64 Linux 2.4.21.SuSE_309.tg1 ia64
(NCSA tg-login) gcc 3.2.2
Intel(R) C++ Version 8.1
Intel(R) Fortran Compiler Version 8.1
mpich-gm-1.2.6..14b-intel-r2
Intel 64 Linux 2.6.9-42.0.10.EL_lustre-1.4.10.1smp
(abe) gcc 3.4.6 20060404
Intel(R) C++ Version 10.0
Intel (R) Fortran Compiler Version 10.0
mvapich2-0.9.8p2patched-intel-ofed-1.2
Windows XP Visual Studio .NET
Visual Studio 2005 w/ Intel Fortran 9.1
Cygwin(native gcc compiler and g95)
MinGW(native gcc compiler and g95)
Windows XP x64 Visual Studio 2005 w/ Intel Fortran 9.1
Windows Vista Visual Studio 2005
MAC OS 10.5.2 (Intel) i686-apple-darwin9-gcc-4.0.1 (GCC) 4.0.1
GNU Fortran (GCC) 4.3.0 20070810
G95 (GCC 4.0.3 (g95 0.91!) Apr 24 2008)
Supported Configuration Features Summary
========================================
In the tables below
y = tested and supported
n = not supported or not tested in this release
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
SunOS5.10 64-bit n y n y y y
SunOS5.10 32-bit n y n y y y
IRIX64_6.5 64-bit n y y y y y
IRIX64_6.5 32-bit n n n n y y
Windows XP n y(15) n(15) y y y
Windows XP x64 n y(15) n(15) y y y
Windows Vista n n n y y y
Mac OS X 10.5 Intel n y n y y y
FreeBSD 4.11 n n n y y y
RedHat EL3 W (3) y(1) y(10) y(1) y y y
RedHat EL3 W Intel (3) n y n y y n
RedHat EL3 W PGI (3) n y n y y n
SuSe x86_64 gcc (3,12) y(2) y(11) y(2) y y y
SuSe x86_64 Int (3,12) n y(13) n y y n
SuSe x86_64 PGI (3,12) n y(8) n y y y
Linux 2.4 Xeon C
Lustre Intel (3,6) n y n y y n
Linux 2.6 SuSE ia64 C
Intel (3,7) y y y y y n
Linux 2.6 SGI Altix
ia64 Intel (3) y y y y y y
Linux 2.6 RHEL C
Lustre Intel (5) y(4) y y(4) y y n
Cray XT3 2.0.41 y y y y y n
Platform Shared Shared Shared Thread-
C libs F90 libs C++ libs safe
Solaris2.10 64-bit y y y y
Solaris2.10 32-bit y y y y
IRIX64_6.5 64-bit y y n y
IRIX64_6.5 32-bit y dna y y
Windows XP y y(15) y y
Windows XP x64 y y(15) y y
Windows Vista y n n y
Mac OS X 10.3 y n
FreeBSD 4.11 y n y y
RedHat EL3 W (3) y y(10) y y
RedHat EL3 W Intel (3) y y y n
RedHat EL3 W PGI (3) y y y n
SuSe x86_64 W GNU (3,12) y y y y
SuSe x86_64 W Int (3,12) y y y n
SuSe x86_64 W PGI (3,12) y y y n
Linux 2.4 Xeon C
Lustre Intel (6) y y y n
Linux 2.4 SuSE
ia64 C Intel (7) y y y n
Linux 2.4 SGI Altix
ia64 Intel y n
Linux 2.6 RHEL C
Lustre Intel (5) y y y n
Cray XT3 2.0.41 n n n n n
Notes: (1) Using mpich2 1.0.6.
(2) Using mpich2 1.0.7.
(3) Linux 2.6 with GNU, Intel, and PGI compilers, as indicated.
W or C indicates workstation or cluster, respectively.
(4) Using mvapich2 0.9.8.
(5) Linux 2.6.9-42.0.10. Xeon cluster with ELsmp_perfctr_lustre
and Intel compilers
(6) Linux 2.4.21-32.0.1. Xeon cluster with ELsmp_perfctr_lustre
and Intel compilers
(7) Linux 2.4.21, SuSE_292.till. Ia64 cluster with Intel compilers
(8) pgf90
(9) With Compaq Visual Fortran 6.6c compiler.
(10) With PGI and Absoft compilers.
(11) PGI and Intel compilers for both C and Fortran
(12) AMD Opteron x86_64
(13) ifort
(14) Yes with C and Fortran, but not with C++
(15) Using Visual Studio 2005 or Cygwin
(16) Not tested for this release.
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* For Red Storm, a Cray XT3 system, the yod command sometimes gives the
message, "yod allocation delayed for node recovery". This interferes with
test suites that do not expect seeing this message. See the section of "Red
Storm" in file INSTALL_parallel for a way to deal with this problem.
AKC - 2008/05/28
* For Red Storm, a Cray XT3 system, the tools/h5ls/testh5ls.sh will fail on
the test "Testing h5ls -w80 -r -g tgroup.h5" fails. This test is
expected to fail and exit with a non-zero code but the yod command does
not propagate the exit code of the executables. Yod always returns 0 if it
can launch the executable. The test suite shell expects a non-zero for
this particular test, therefore it concludes the test has failed when it
receives 0 from yod. To bypass this problem for now, change the following
lines in the tools/h5ls/testh5ls.sh.
======== Original =========
# The following combination of arguments is expected to return an error message
# and return value 1
TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== Skip the test =========
echo SKIP TOOLTEST tgroup-1.ls 1 -w80 -r -g tgroup.h5
======== end of bypass ========
AKC - 2008/05/28
* We have discovered two problems when running collective IO parallel HDF5
tests with chunking storage on the ChaMPIon MPI compiler on tungsten, a
Linux cluster at NCSA.
Under some complex selection cases:
1) MPI_Get_element returns the wrong value.
2) MPI_Type_struct also generates the wrong derived datatype and corrupt
data may be generated.
These issues arise only when turning on collective IO with chunking storage
with some complex selections. We have not found these problems on other
MPI-IO compilers. If you encounter these problems, you may use independent
IO instead.
To avoid this behavior, change the following line in your code
H5Pset_dxpl_mpio(xfer_plist, H5FD_MPIO_COLLECTIVE);
to
H5Pset_dxpl_mpio(xfer_plist, H5FD_MPIO_INDEPENDENT);
KY - 2007/08/24
* For SNL, spirit/liberty/thunderbird: The serial tests pass but parallel
tests failed with MPI-IO file locking message. AKC - 2007/6/25
* On Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* For LLNL, uP: both serial and parallel tests pass.
Zeus: Serial tests pass but parallel tests fail with a known problem in MPI.
ubgl: Serial tests pass but parallel tests fail.
* Configuring with --enable-debug=all produces compiler errors on most
platforms: Users who want to run HDF5 in debug mode should use
--enable-debug rather than --enable-debug=all to enable debugging
information on most modules.
* On Mac OS 10.4, test/dt_arith.c has some errors in conversion from long
double to (unsigned) long long and from (unsigned) long long to long double.
* On Altix SGI with Intel 9.0, testmeta.c would not compile with -O3
optimization flag.
* On VAX, the Scaleoffset filter is not supported. The Scaleoffset filter
supports only the IEEE standard for floating-point data; it cannot be applied
to HDF5 data generated on VAX.
* On Cray X1, a lone colon on the command line of h5dump --xml (as in
the testh5dumpxml.sh script) is misinterpereted by the operating system
and causes an error.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* On IRIX6.5, when the C compiler version is greater than 7.4, complicated
MPI derived datatype code will work. However, the user should increase
the value of the MPI_TYPE_MAX environment variable to some appropriate value
to use collective irregular selection code. For example, the current
parallel HDF5 test needs to raise MPI_TYPE_MAX to 200,000 to pass the test.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculation of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
SLU - 2005/6/30
* For version 6 (6.02 and 6.04) of the Portland Group compiler on the AMD
Opteron processor, there is a bug in the compiler for optimization(-O2).
The library failed in several tests, all related to the MULTI driver.
The problem has been reported to the vendor.
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use",
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
* The --enable-static-exec configure flag fails to compile for Solaris
platforms. This is due to the fact that not all of the system libraries on
Solaris are available in a static format.
The --enable-static-exec configure flag also fails to correctly compile
on IBM SP2 platforms for serial mode. The parallel mode works fine with
this option.
It is suggested that you do not use this option on these platforms
during configuration.
* With the gcc 2.95.2 compiler, HDF5 uses the `-ansi' flag during
compilation. The ANSI version of the compiler complains about not being
able to handle the `long long' datatype with the warning:
warning: ANSI C does not support `long long'
This warning is innocuous and can be safely ignored.
* The ./dsets tests fail on the TFLOPS machine if the test program,
dsets.c, is compiled with the -O option. The HDF5 library still works
correctly with the -O option. The test program works fine if it is
compiled with -O1 or -O0. Only -O (same as -O2) causes the test
program to fail.
* Not all platforms behave correctly with Szip's shared libraries. Szip is
disabled in these cases, and a message is relayed at configure time. Static
libraries should be working on all systems that support Szip and should be
used when shared libraries are unavailable.
There is also a configure error on Altix machines that incorrectly reports
when a version of Szip without an encoder is being used.
* On some platforms that use Intel and Absoft compilers to build the HDF5
Fortran library, compilation may fail for fortranlib_test.f90, fflush1.f90
and fflush2.f90 complaining about the exit subroutine. Comment out the line
IF (total_error .ne. 0) CALL exit (total_error).
* Information about building with PGI and Intel compilers is available in
the INSTALL file sections 4.7 and 4.8.
* On at least one system, SDSC DataStar, the scheduler (in this case
LoadLeveler) sends job status updates to standard error when you run
any executable that was compiled with the parallel compilers.
This causes problems when running "make check" on parallel builds, as
many of the tool tests function by saving the output from test runs,
and comparing it to an exemplar.
The best solution is to reconfigure the target system so it no longer
inserts the extra text. However, this may not be practical.
In such cases, one solution is to "setenv HDF5_Make_Ignore yes" prior to
the configure and build. This will cause "make check" to continue after
detecting errors in the tool tests. However, in the case of SDSC DataStar,
it also leaves you with some 150 "failed" tests to examine by hand.
A second solution is to write a script to run serial tests and filter
out the text added by the scheduler. A sample script used on SDSC
DataStar is given below, but you will probably have to customize it
for your installation.
Observe that the basic idea is to insert the script as the first item
on the command line which executes the the test. The script then
executes the test and filters out the offending text before passing
it on.
#!/bin/csh
set STDOUT_FILE=~/bin/serial_filter.stdout
set STDERR_FILE=~/bin/serial_filter.stderr
rm -f $STDOUT_FILE $STDERR_FILE
($* > $STDOUT_FILE) >& $STDERR_FILE
set RETURN_VALUE=$status
cat $STDOUT_FILE
tail +3 $STDERR_FILE
exit $RETURN_VALUE
You get the HDF5 make files and test scipts to execute your filter script
by setting the environment variable "RUNSERIAL" to the full path of the
script prior to running configure for parallel builds. Remember to
"unsetenv RUNSERIAL" before running configure for a serial build.
Note that the RUNSERIAL environment variable exists so that we can
prefix serial runs as necessary on the target system. On DataStar,
no prefix is necessary. However on an MPICH system, the prefix might
have to be set to something like "/usr/local/mpi/bin/mpirun -np 1" to
get the serial tests to run at all.
In such cases, you will have to include the regular prefix in your
filter script.
* H5Ocopy() does not copy reg_ref attributes correctly when shared-message
is turn on. The value of the reference in the destination attriubte is
wrong. This H5Ocopy problem will affect the h5copy tool.
* In the C++ API, it appears that there are bugs in Attribute::write/read
and DataSet::write/read for fixed- and variable-len strings. The problems
are being worked on and a patch will be provided when the fixes are
available.
%%%%1.8.0%%%%
HDF5 version 1.8.0 released on Tue Feb 12 20:41:19 CST 2008
================================================================================
INTRODUCTION
============
This document describes the differences between the HDF5-1.6.x release series
and HDF5 1.8.0, and contains information on the platforms tested and known
problems in HDF5-1.8.0. For more details, see the HISTORY-1_0-1_8_0_rc3.txt
file in the
release_docs/ directory of the HDF5 source.
Links to the HDF5 1.8.0 source code, documentation, and additional materials
can be found on the HDF5 web page at:
http://www.hdfgroup.org/products/hdf5/
The HDF5 1.8.0 release can be obtained from:
http://www.hdfgroup.org/HDF5/release/obtain5.html
User documentation for 1.8.0 can be accessed directly at this location:
http://www.hdfgroup.org/HDF5/doc/
New features in 1.8.0, including brief general descriptions of some new
and modified APIs, are described in the "What's New in 1.8.0?" document:
http://www.hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
All new and modified APIs are listed in detail in the "HDF5 Software Changes
from Release to Release" document, in the section "Release 1.8.0 (current
release) versus Release 1.6.x":
http://www.hdfgroup.org/HDF5/doc/ADGuide/Changes.html
If you have any questions or comments, please send them to the HDF Help Desk:
help@hdfgroup.org
CONTENTS
========
- New Features
- Removed Feature
- Support for new platforms and languages
- Bug Fixes since HDF5-1.6.0
- Platforms Tested
- Supported Configuration Features Summary
- Known Problems
New Features
============
HDF5 Release 1.8.0 is a major release with many changes and new features.
New format and interface features discussed in the "What's New in
HDF5 1.8.0" document include the following:
Enhanced group object management
Enhanced attribute management and more efficient meta data handling
Expanded datatype features
Creation order tracking and indexing
Improved meta data caching and cache control
UTF-8 encoding
New I/O filters: n-bit and scale+offset compression
New link (H5L) and object (H5O) interfaces and features
External and user-defined links
New high-level APIs:
HDF5 Packet Table (H5PT) and HDF5 Dimension Scale (H5DS)
C++ and Fortran interfaces for older high-level APIs:
H5Lite (H5LT), H5Image (H5IM), and H5Table (H5TB)
New and improved tools
And more...
http://hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
New APIs associated with these features, other interface changes
(e.g., ENUM and struct definitions), and new library configuration flags
are listed in the "Release 1.8.0 (current release) versus Release 1.6.x"
section of "HDF5 Software Changes from Release to Release."
http://hdfgroup.org/HDF5/doc/ADGuide/Changes.html
Compatibility
-------------
Many HDF5 users and user communities have existing applications that
they may wish to port to Release 1.8.0. Alternatively, some users may
wish to take advantage of Release 1.8.0's improved performance without
having to port such applications. To facilitate managing application
compatibility and porting applications from release to release, the HDF
Team has implemented the following features:
Individually-configurable macros that selectively map common
interface names to the old and new interfaces
Library configuration options to configure the macro mappings
Two related documents accompany this release:
"API Compatibility Macros in HDF5" discusses the specifics of the
new individually-configurable macros and library configuration
options.
http://hdfgroup.org/HDF5/doc/RM/APICompatMacros.html
"New Features in HDF5 Release 1.8.0 and Backward/Forward Format
Compatibility Issues" discusses each new feature with regard to
its impact on format compatibility.
http://hdfgroup.org/HDF5/doc/ADGuide/CompatFormat180.html
Referenced documents
--------------------
http://hdfgroup.org/HDF5/doc/ADGuide/WhatsNew180.html
"What's New in HDF5 1.8.0"
http://hdfgroup.org/HDF5/doc/ADGuide/Changes.html
The "Release 1.8.0 (current release) versus Release 1.6.x "
section in "HDF5 Software Changes from Release to Release"
http://hdfgroup.org/HDF5/doc/RM/APICompatMacros.html
"API Compatibility Macros in HDF5"
http://hdfgroup.org/HDF5/doc/ADGuide/CompatFormat180.html
"New Features in HDF5 Release 1.8.0 and Backward/Forward Format
Compatibility Issues"
Removed Feature
===============
The stream virtual file driver (H5FD_STREAM) have been removed in this
release. This affects the functions H5Pset_fapl_stream and H5Pget_fapl_stream
and the constant H5FD_STREAM.
This virtual file driver will be available at
http://hdf5-addons.origo.ethz.ch/. Note that at the time of this release,
the transition is still in progress; the necessary integration tools may
not be available when HDF5 Release 1.8.0 first comes out.
Support for New Platforms, Languages, and Compilers
===================================================
- Support for Open VMS 7.3 was added.
Bug Fixes since HDF5-1.6.0
==========================
This release contains numerous bug fixes. For details, see the
"Changes from 1.6.0 to 1.8.0-rc3" section of the HISTORY.txt file for
this release.
Platforms Tested
================
The following platforms and compilers have been tested for for this release.
AIX 5.2 (32/64 bit) xlc 8.0.0.11
xlC 8.0
xlf 10.01.0000.0
mpcc_r 6.0.0.8
mpxlf_r 8.1.1.7
FreeBSD 6.2-STABLE i386 gcc 3.4.6 [FreeBSD] 20060305
(duty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.2.1 20080123
g++ 4.2.1 20080123
gfortran 4.2.1 20070620
FreeBSD 6.2-STABLE amd64 gcc 3.4.6 [FreeBSD] 20060305
(liberty) g++ 3.4.6 [FreeBSD] 20060305
gcc 4.2.1 20080123
g++ 4.2.1 20080123
gfortran 4.2.1 20080123
IRIX64 6.5 (64 & n32) MIPSpro cc 7.4.4m
F90 MIPSpro 7.4.4m
C++ MIPSpro cc 7.4.4m
Linux 2.6.9 (RHEL4) Intel 10.0 compilers
(abe.ncsa.uiuc.edu)
Linux 2.4.21-47 gcc 3.2.3 20030502
(osage)
Linux 2.6.9-42.0.10 gcc 3.4.6 20060404
(kagiso) PGI 7.0-7 (pgcc, pgf90, pgCC)
Intel 9.1 (icc, ifort, icpc)
Linux 2.6.16.27 x86_64 AMD gcc 4.1.0 (SuSE Linux), g++ 4.1.0,
(smirom) g95 (GCC 4.0.3)
PGI 6.2-5 (pgcc, pgf90, pgCC)
Intel 9.1 (icc, iort, icpc)
Linux 2.6.5-7.252.1-rtgfx #1 Intel(R) C++ Version 9.0
SMP ia64 Intel(R) Fortran Itanium(R) Version 9.0
(cobalt) SGI MPI
SunOS 5.8 32,46 Sun WorkShop 6 update 2 C 5.3
(Solaris 2.8) Sun WorkShop 6 update 2 Fortran 95 6.2
Sun WorkShop 6 update 2 C++ 5.3
SunOS 5.10 cc: Sun C 5.8
(linew) f90: Sun Fortran 95 8.2
CC: Sun C++ 5.8
Xeon Linux 2.4.21-32.0.1.ELsmp-perfctr-lustre
(tungsten) gcc 3.2.2 20030222
Intel(R) C++ Version 9.0
Intel(R) Fortran Compiler Version 9.0
IA-64 Linux 2.4.21.SuSE_292.til1 ia64
(NCSA tg-login) gcc 3.2.2
Intel(R) C++ Version 8.1
Intel(R) Fortran Compiler Version 8.1
mpich-gm-1.2.5..10-intel-r2
Windows XP Visual Studio .NET
Visual Studio 2005 w/ Intel Fortran 9.1
Cygwin(native gcc compiler and g95)
MinGW(native gcc compiler and g95)
Windows XP x64 Visual Studio 2005 w/ Intel Fortran 9.1
Windows Vista Visual Studio 2005
MAC OS 10.4 (Intel) gcc i686-apple-darwin8-gcc-4.0.1 (GCC) 4.0.1
G95 (GCC 4.0.3 (g95 0.91!) Nov 21 2006)
Alpha Open VMS 7.3 Compaq C V6.5-001-48BCD
HP Fortran V7.6-3276
Compaq C++ V6.5-004
Supported Configuration Features Summary
========================================
In the tables below
y = tested and supported
n = not supported or not tested in this release
x = not working in this release
dna = does not apply
( ) = footnote appears below second table
<blank> = testing incomplete on this feature or platform
Platform C F90 F90 C++ zlib SZIP
parallel parallel
SunOS5.8 64-bit n y n y y y
SunOS5.8 32-bit n y n y y y
SunOS5.10 64-bit y(1) y n y y y
SunOS5.10 32-bit y(1) y n y y y
IRIX64_6.5 64-bit n y y y y y
IRIX64_6.5 32-bit n n n n y y
AIX-5.2 32-bit y y y y y y
AIX-5.2 64-bit y y y y y y
Windows XP n y(15) n(15) y y y
Windows XP x64 n y(15) n(15) y y y
Windows Vista n n n y y y
Mac OS X 10.4 PowerPC n n
Mac OS X 10.4 Intel n y n y y y
FreeBSD 4.11 n n n y y y
RedHat EL3 W (3) y(1a) y(10) y(1a) y y y
RedHat EL3 W Intel (3) n y n y y n
RedHat EL3 W PGI (3) n y n y y n
SuSe x86_64 gcc (3,12) y(1a) y(11) n y y y
SuSe x86_64 Int (3,12) n y(13) n y y n
SuSe x86_64 PGI (3,12) n y(8) n y y y
Linux 2.4 Xeon C
Lustre Intel (3,6) n y n y y n
Linux 2.6 SuSE ia64 C
Intel (3,7) y y y y y n
Linux 2.6 SGI Altix
ia64 Intel (3) y y y y y y
Alpha OpenVMS 7.3.2 n y n y n n
Platform Shared Shared Shared static- Thread-
C libs F90 libs C++ libs exec safe
Solaris2.8 64-bit y y y x y
Solaris2.8 32-bit y y y x y
Solaris2.10 64-bit y x y
Solaris2.10 32-bit y x y
IRIX64_6.5 64-bit y y n y y
IRIX64_6.5 32-bit y dna y y y
AIX-5.2 & 5.3 32-bit n n n y n
AIX-5.2 & 5.3 64-bit n n n y n
Windows XP y y(15) y y y
Windows XP x64 y y(15) y y y
Windows Vista y n n y y
Mac OS X 10.3 y y n
FreeBSD 4.11 y n y y y
RedHat EL3 W (3) y y(10) y y y
RedHat EL3 W Intel (3) y y y y n
RedHat EL3 W PGI (3) y y y y n
SuSe x86_64 W GNU (3,12) y y y y y
SuSe x86_64 W Int (3,12) y y y y(14) n
SuSe x86_64 W PGI (3,12) y y y y(14) n
Linux 2.4 Xeon C
Lustre Intel (6) y y y y n
Linux 2.4 SuSE
ia64 C Intel (7) y y y y n
Linux 2.4 SGI Altix
ia64 Intel y y n
Alpha OpenVMS 7.3.2 n n n y n
Notes: (1) Using mpich 1.2.6.
(1a) Using mpich2 1.0.6.
(2) Using mpt and mpich 1.2.6.
(3) Linux 2.6 with GNU, Intel, and PGI compilers, as indicated.
W or C indicates workstation or cluster, respectively.
(6) Linux 2.4.21-32.0.1. Xeon cluster with ELsmp_perfctr_lustre
and Intel compilers
(7) Linux 2.4.21, SuSE_292.till. Ia64 cluster with Intel
compilers
(8) pgf90
(9) With Compaq Visual Fortran 6.6c compiler.
(10) With PGI and Absoft compilers.
(11) PGI and Intel compilers for both C and Fortran
(12) AMD Opteron x86_64
(13) ifort
(14) Yes with C and Fortran, but not with C++
(15) Using Visual Studio 2005 or Cygwin
(16) Not tested for this release.
Compiler versions for each platform are listed in the preceding
"Platforms Tested" table.
Known Problems
==============
* We have discovered two problems when running collective IO parallel HDF5
tests with chunking storage on the ChaMPIon MPI compiler on tungsten, a
Linux cluster at NCSA.
Under some complex selection cases:
1) MPI_Get_element returns the wrong value.
2) MPI_Type_struct also generates the wrong derived datatype and corrupt
data may be generated.
These issues arise only when turning on collective IO with chunking storage
with some complex selections. We have not found these problems on other
MPI-IO compilers. If you encounter these problems, you may use independent
IO instead.
To avoid this behavior, change the following line in your code
H5Pset_dxpl_mpio(xfer_plist, H5FD_MPIO_COLLECTIVE);
to
H5Pset_dxpl_mpio(xfer_plist, H5FD_MPIO_INDEPENDENT);
KY - 2007/08/24
* For SNL, spirit/liberty/thunderbird: The serial tests pass but parallel
tests failed with MPI-IO file locking message. AKC - 2007/6/25
* On Intel 64 Linux cluster (RH 4, Linux 2.6.9) with Intel 10.0 compilers,
use -mp -O1 compilation flags to build the libraries. A higher level of
optimization causes failures in several HDF5 library tests.
* For SNL, Red Storm: Only parallel HDF5 is supported. The serial tests pass
when run against the parallel library; the parallel tests also pass, but
with lots of non-fatal error messages.
* For LLNL, uP: both serial and parallel tests pass.
Zeus: Serial tests pass but parallel tests fail with a known problem in MPI.
ubgl: Serial tests pass but parallel tests fail.
* On SUN 5.10 C++, testing fails in the "Testing Shared Datatypes with
Attributes" test.
* Configuring with --enable-debug=all produces compiler errors on most
platforms: Users who want to run HDF5 in debug mode should use
--enable-debug rather than --enable-debug=all to enable debugging
information on most modules.
* On Mac OS 10.4, test/dt_arith.c has some errors in conversion from long
double to (unsigned) long long and from (unsigned) long long to long double.
* On Altix SGI with Intel 9.0, testmeta.c would not compile with -O3
optimization flag.
* On VAX, the Scaleoffset filter is not supported. The filter cannot be
applied to HDF5 data generated on VAX. The Scaleoffset filter only supports
the IEEE standard for floating-point data.
* On Cray X1, a lone colon on the command line of h5dump --xml (as in
the testh5dumpxml.sh script) is misinterpereted by the operating system
and causes an error.
* On mpich 1.2.5 and 1.2.6, if more than two processes contribute no IO and
the application asks to do collective IO, we have found that when using 4
processors, a simple collective write will sometimes be hung. This can be
verified with t_mpi test under testpar.
* On IRIX6.5, when the C compiler version is greater than 7.4, complicated
MPI derived datatype code will work. However, the user should increase
the value of the MPI_TYPE_MAX environment variable to some appropriate value
to use collective irregular selection code. For example, the current
parallel HDF5 test needs to raise MPI_TYPE_MAX to 200,000 to pass the test.
* A dataset created or rewritten with a v1.6.3 library or after cannot be read
with the v1.6.2 library or before when the Fletcher32 EDC filter is enabled.
There was a bug in the calculating code of the Fletcher32 checksum in the
library before v1.6.3; the checksum value was not consistent between big-
endian and little-endian systems. This bug was fixed in Release 1.6.3.
However, after fixing the bug, the checksum value was no longer the same as
before on little-endian system. Library releases after 1.6.4 can still read
datasets created or rewritten with an HDF5 library of v1.6.2 or before.
SLU - 2005/6/30
* For version 6 (6.02 and 6.04) of the Portland Group compiler on the AMD
Opteron processor, there is a bug in the compiler for optimization(-O2).
The library failed in several tests, all related to the MULTI driver.
The problem has been reported to the vendor.
* On IBM AIX systems, parallel HDF5 mode will fail some tests with error
messages like "INFO: 0031-XXX ...". This is from the command `poe'.
Set the environment variable MP_INFOLEVEL to 0 to minimize the messages
and run the tests again.
The tests may fail with messages like "The socket name is already in use",
but HDF5 does not use sockets. This failure is due to problems with the
poe command trying to set up the debug socket. To resolve this problem,
check to see whether there are many old /tmp/s.pedb.* files staying around.
These are sockets used by the poe command and left behind due to failed
commands. First, ask your system administrator to clean them out.
Lastly, request IBM to provide a means to run poe without the debug socket.
* The --enable-static-exec configure flag fails to compile for Solaris
platforms. This is due to the fact that not all of the system libraries on
Solaris are available in a static format.
The --enable-static-exec configure flag also fails to correctly compile
on IBM SP2 platform for the serial mode. The parallel mode works fine with
this option.
It is suggested that you do not use this option on these platforms
during configuration.
* With the gcc 2.95.2 compiler, HDF5 uses the `-ansi' flag during
compilation. The ANSI version of the compiler complains about not being
able to handle the `long long' datatype with the warning:
warning: ANSI C does not support `long long'
This warning is innocuous and can be safely ignored.
* The ./dsets tests fail on the TFLOPS machine if the test program,
dsets.c, is compiled with the -O option. The HDF5 library still works
correctly with the -O option. The test program works fine if it is
compiled with -O1 or -O0. Only -O (same as -O2) causes the test
program to fail.
* Not all platforms behave correctly with Szip's shared libraries. Szip is
disabled in these cases, and a message is relayed at configure time. Static
libraries should be working on all systems that support Szip and should be
used when shared libraries are unavailable.
There is also a configure error on Altix machines that incorrectly reports
when a version of Szip without an encoder is being used.
* On some platforms that use Intel and Absoft compilers to build the HDF5
Fortran library, compilation may fail for fortranlib_test.f90, fflush1.f90
and fflush2.f90 complaining about the exit subroutine. Comment out the line
IF (total_error .ne. 0) CALL exit (total_error).
* Information about building with PGI and Intel compilers is available in
the INSTALL file sections 4.7 and 4.8.
* On at least one system, SDSC DataStar, the scheduler (in this case
LoadLeveler) sends job status updates to standard error when you run
any executable that was compiled with the parallel compilers.
This causes problems when running "make check" on parallel builds, as
many of the tool tests function by saving the output from test runs,
and comparing it to an exemplar.
The best solution is to reconfigure the target system so it no longer
inserts the extra text. However, this may not be practical.
In such cases, one solution is to "setenv HDF5_Make_Ignore yes" prior to
the configure and build. This will cause "make check" to continue after
detecting errors in the tool tests. However, in the case of SDSC DataStar,
it also leaves you with some 150 "failed" tests to examine by hand.
A second solution is to write a script to run serial tests and filter
out the text added by the scheduler. A sample script used on SDSC
DataStar is given below, but you will probably have to customize it
for your installation.
Observe that the basic idea is to insert the script as the first item
on the command line which executes the the test. The script then
executes the test and filters out the offending text before passing
it on.
#!/bin/csh
set STDOUT_FILE=~/bin/serial_filter.stdout
set STDERR_FILE=~/bin/serial_filter.stderr
rm -f $STDOUT_FILE $STDERR_FILE
($* > $STDOUT_FILE) >& $STDERR_FILE
set RETURN_VALUE=$status
cat $STDOUT_FILE
tail +3 $STDERR_FILE
exit $RETURN_VALUE
You get the HDF5 make files and test scipts to execute your filter script
by setting the environment variable "RUNSERIAL" to the full path of the
script prior to running configure for parallel builds. Remember to
"unsetenv RUNSERIAL" before running configure for a serial build.
Note that the RUNSERIAL environment variable exists so that we can
can prefix serial runs as necessary on the target system. On DataStar,
no prefix is necessary. However on an MPICH system, the prefix might
have to be set to something like "/usr/local/mpi/bin/mpirun -np 1" to
get the serial tests to run at all.
In such cases, you will have to include the regular prefix in your
filter script.
* H5Ocopy() does not copy reg_ref attributes correctly when shared-message
is turn on. The value of the reference in the destination attriubte is
wrong. This H5Ocopy problem will affect the h5copy tool.
Reference in New Issue View Git Blame Copy Permalink