Bug fix (sorta)
Description:
Change reading of "missing" chunks from datasets with undefined fill
values to not overwrite the application buffer with random garbage from
memory. Note that this is almost the same, since whatever garbage the
application had in those locations will still be there...
Platforms tested:
FreeBSD 4.10 (sleipnir)
IRIX64 6.5 (modi4)
h5committested
Bug fix
Description:
Allow I/O on extendible chunked datasets with (currently) zero-sized
dimensions to proceed harmlessly instead of dumping core on an assertion.
Solution:
Removed assertion and added checks to avoid problem situation in H5TB_end
Platforms tested:
FreeBSD 4.10 (sleipnir) w/ & w/o parallel
Too minor to require h5committest
Purpose: Improvement
Description: HDF5 Library set pixels_per_scanline parameter to the size of the chunk's
fastest changing dimension. As a result, fastest changing dimension
of the chunk could not be bigger than 4K and smaller than pixels_per_block
value and szip compression couldn't be used for many real datasets.
Solution: Reworked algorithm how HDF5 sets pixels_per_scanline value; only chunks
with the total number of elements less than pixels_per_block value are rejected.
There is no restriction on the size of the chunk's fastest changing
dimension anymore.
Modified the test according to the new algorithm.
Platforms tested: verbena, copper, sol
Misc. update:
Code cleanup
Description:
Clean up almost all warnings from Windows builds.
Platforms tested:
FreeBSD 4.10 (sleipnir) w/parallel
Too minor to require h5committest
Purpose:
HDF5 now supports SZIP with no encoder.
Description:
SZIP can be configured to have both encoder and decoder or just to have the decoder. HDF5 can now query the configuration of any filter, and will throw errors if users try to write using a filter with encoding disabled.
Solution:
Added H5Zget_filter_info function, changed API for H5Pget_filter and H5P_get_filter_by_id. See SZIP RFC.
Platforms tested:
Copper (fortran, C++, parallel), Sleipnir (C++), Arabica (fortran, C++), Verbena (fortran, C++)
Misc. update:
Code optimization
Description:
Don't recompute the internal index value for looking up the chunk in the
hash table, just use the value already computed from iterating through the
chunks.
Platforms tested:
Solaris 2.7 (arabica)
FreeBSD 4.9 (sleipnir) w/parallel
Code optimization & bug fix
Description:
When dimension information is being stored in the storage layout message
on disk, it is stored as 32-bit quantities, possibly truncating the dimension
information, if a dimension is greater than 32-bits in size.
Solution:
Fix the storage layout message problem by revising file format to not store
dimension information, since it is already available in the dataspace.
Also revise the storage layout data structures to be more compartmentalized
for the information for contiguous, chunked and compact storage.
Platforms tested:
FreeBSD 4.9 (sleipnir) w/parallel
Solaris 2.7 (arabica)
h5committest
Code cleanup
Description:
Clean up lots of warnings based on those reported from the SGI compilers
as well as gcc.
Platforms tested:
SGI O3900, IRIX64 6.5 (Cheryl's SGI machine)
FreeBSD 4.9 (sleipnir) w/ & w/o parallel
h5committest
bug fix
Description:
the test that reads a pre-saved file with filters must only be run
when that filter is available on the current configuration,
otherwise the library trigers a read error
Solution:
added a #ifdef for the filter, that prints -SKIP- in case we do not have the filter
Platforms tested:
linux
Misc. update:
bug fix
Description:
the added binary files to read did not have the construction of the "srcdir" path
Solution:
add it
Platforms tested:
linux
Misc. update:
bug fix
new test
Description:
the fletcher filter used a temporary 2 byte word buffer to compute the checksum.
this is non portable between big-endian/little endian.
added a test that reads 2 pre-saved files (one LE, other BE) with that filter enabled
Solution:
replaced with a buffer of 1 byte type
Platforms tested:
linux
solaris
solaris 64 bit
AIX
windows
Misc. update:
Purpose:
replaced name of delete filter with remove filter for the new function H5Premove_filter
Description:
Solution:
Platforms tested:
linux
Misc. update:
Code cleanup
Description:
Add C++ and FORTRAN wrappers for new H5Pdelete_filter routine, along with
documentation and a note in the release notes.
Platforms tested:
FreeBSD 4.9 (sleipnir)
Linux 2.4 (verbena) w/ C++ and FORTRAN
Too minor for full h5committest
new library function H5Pdelete_filter
deletes one or all filters from a dataset creation property list
this was done for the NONE option of h5repack, added tests for this feature
added a test for the new function in /test/dsets.c
Description:
Solution:
Platforms tested:
linux
solaris
AIX
Misc. update:
Bug fix
Description:
When two property lists are compared, the H5Pequal routine was just
comparing the raw information for the property values. This causes problems
when the raw information contains pointers to other information.
Solution:
Allow a 'compare' callback to be registered for properties, so that a user
application get perform the comparison itself, allowing for "deep" compares of
the property value.
This was exported to the H5Pregister & H5Pinsert routines in the development
branch, but not the release branch.
Platforms tested:
FreeBSD 4.9 (sleipnir)
h5committest
Bug fix
Description:
Range check the szip 'pixels per block' against the chunk size of the
dataset when attempting to create a new dataset, since the szip library
requires the PPB to be at least the size of the fastest changing dimension
in the chunk.
Platforms tested:
FreeBSD 4.9 (sleipnir)
too minor for h5committest
Code cleanup
Description:
Add in rest of szip "options mask" macros that were missing. Also made
"raw" options mask set by the library, instead of requiring users to always set
it.
Platforms tested:
FreeBSD 4.8 (sleipnir)
Minor tweaks too small fo h5committest
Code cleanup
Description:
Remove debugging information and match bug-fix changes to szip filter
code.
Platforms tested:
FreeBSD 4.8 (sleipnir)
h5committest not needed.
Description: when a datatype element size is bigger than I/O datatype
conversion buffer(1MB), library reports failure.
Solution: adjust conversion buffer to be at least equal to one element size
Platforms tested: h5committest
New feature/enhancement
Description:
Chunked datasets are handled poorly in several circumstances involving
certain selections and chunks that are too large for the chunk cache and/or
chunks with filters, causing the chunk to be read from disk multiple times.
Solution:
Rearrange raw data I/O infrastructure to handle chunked datasets in a much
more friendly way by creating a selection in memory and on disk for each chunk
in a chunked dataset and performing all of the I/O on that chunk at one time.
There are still some scalability (the current code attempts to
create a selection for all the chunks in the dataset, instead of just the
chunks that are accessed, requiring portions of the istore.c and fillval.c
tests to be commented out) and performance issues, but checking this in will
allow the changes to be tested by a much wider audience while I address the
remaining issues.
Platforms tested:
h5committested, FreeBSD 4.8 (sleipnir) serial & parallel, Linux 2.4 (eirene)
Code restructure.
Description:
Restructured tests so they shouldn't depend on shared open()'s and can be
tested with CodeWarrior
Platforms tested:
FreeBSD 4.8 (sleipnir)
Triple check not needed.
code warrior port
Description:
added a flag H5_NO_SHARED_WRITING, it is defined only for CW in the H5pubconf.h
this avoids doing some tests for CW that fail on shared writing
Solution:
Platforms tested:
windows (cw and ms)
Linux 2.4 (rockaway)
SunOS 5.7 (arabica)
IRIX 6.5 (modi4)
Misc. update:
Purpose: Windows maintenance and code cleanup
Description: There were a lot of places in the source code where varibale
was declared but never used.
Solution: Cleaned the source code to avoid warnings on Windows and Unix
platforms.
Platforms tested: arabica, modi4, Windows
Probably my environment is wrong on Linux, cannot
compile at all when fortran is enabled
If daily tests fails tomorrow, this will be my fault
but I want to go home now :-)
Misc. update:
Bug Fix
Description:
When using the STDIO driver, the dsets test would fail.
Solution:
Need to flush the data to disk before doing an HDopen call.
Platforms tested:
Modi4 (Parallel & Fortran)
Arabica (Fortran)
Verbena (Fortran & C++)
Misc. update:
Purpose: Windows port
Description: Windows cannot convert from _uint64_t to double. One of
the tests (test_set_local) used conversion to fill a data
buffer. Compilation failed on Windows.
Solution: Fixed the code not to use conversion
Platforms tested: CRAY T90IEEE, arabica, modi4 with parallel, Windows 2000
Only static tests (both debug and release) passed on Windows.
DLLs have multiple problems ;-)
Misc. update:
Code cleanup
Description:
Switched some system/library calls to use wrappers.
Platforms tested:
FreeBSD 4.8 (sleipnir) w/C++
Linux 2.4 (burrwhite) w/FORTRAN
Solaris 2.7 (arabica) w/FORTRAN
IRIX64 6.5 (modi4) w/parallel & FORTRAN
(h5committest not run due to my ongoing difficulties with C++ on burrwhite).
Code cleanup & new tests
Description:
Updated szip and shuffle testing to reflect changes to API functions.
Added testing for new API functions: H5Pmodify_filter, H5Pget_filter_by_id
and H5Pall_filters_avail.
Added testing for "can apply" and "set local callbacks.
Cleaned up various things, also.
Solution:
Platforms tested:
FreeBSD 4.8 (sleipnir) w/szip
Linux 2.4 (sleipnir) w/szip
Solaris 2.7 (arabica) w/FORTRAN
IRIX64 6.5 (modi4) w/szip, FORTRAN & parallel
Misc. update:
Code cleanup
Description:
Updated copyright.
Removed szip-specific testing in favor of standard testing, now that bugs
in the szip filter are fixed.
Updated to latest version of H5Pset_szip API
Solution:
Platforms tested:
FreeBSD 4.8 (sleipnir)
IRIX64 6.5 (modi4) w/parallel
Misc. update:
To add tests for szip compression in HDF5
Description:
Three tests were added:
1) szip filter itself
3) szip + shuffle + fletch32
2) fletch32 + shuffle + szip
Solution:
Since H5Dget_storagesize returns 0(Quincey can fix it or it may be gone),
I create another function called test_internal_filter_szip, which is the almost the same
as test_internal_filter but comments out H5Dget_storage_size function.
Platforms tested:
Since there are changes of configure.in and configure,I didn't use h5committest.
I tested with four platforms.
1) Linux 2.4 (eirene) 2) Solaris 2.7(arabic
3) windows 2000(VS 6.0) 4) SGI IRIX6.5-64(modi4)
For test 1)-3), only basic C tests were done For modi4 test, I tested 64-bit C,parallel and fortran.
All tests passed, except a warning message from szip library when checksum is used in some order, which doesn't cause any real problems.
Misc. update:
Update
Description:
Updated the Copyright statement
Platforms tested:
Linux (This change is only in the comments, so I just check that the
modules still compile)
Misc. update:
Code cleanup
Description:
Clean up miscellaneous warnings which have crept into the code.
Fix "_POSIX_C_SOURCE not defined" warning on FreeBSD.
Adjust gcc compiler flags to be more concise for production mode.
Refactor the H5O code so that there is a stronger boundary between code
in the H5O package and code in the library which just calls H5O routines.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
FreeBSD 4.7 (sleipnir) serial & parallel and gcc 2.95.4 & gcc 3.2.2
Misc. update:
Update MANIFEST if you add or remove any file.
Lots of performance improvements & a couple new internal API interfaces.
Description:
Performance Improvements:
- Cached file offset & length sizes in shared file struct, to avoid
constantly looking them up in the FCPL.
- Generic property improvements:
- Added "revision" number to generic property classes to speed
up comparisons.
- Changed method of storing properties from using a hash-table
to the TBBT routines in the library.
- Share the propery names between classes and the lists derived
from them.
- Removed redundant 'def_value' buffer from each property.
- Switching code to use a "copy on write" strategy for
properties in each list, where the properties in each list
are shared with the properties in the class, until a
property's value is changed in a list.
- Fixed error in layout code which was allocating too many buffers.
- Redefined public macros of the form (H5open()/H5check, <variable>)
internally to only be (<variable>), avoiding innumerable useless
calls to H5open() and H5check_version().
- Reuse already zeroed buffers in H5F_contig_fill instead of
constantly re-zeroing them.
- Don't write fill values if writing entire dataset.
- Use gettimeofday() system call instead of time() system when
checking the modification time of a dataset.
- Added reference counted string API and use it for tracking the
names of objects opening in a file (for the ID->name code).
- Removed redundant H5P_get() calls in B-tree routines.
- Redefine H5T datatype macros internally to the library, to avoid
calling H5check redundantly.
- Keep dataspace information for dataset locally instead of reading
from disk each time. Added new module to track open objects
in a file, to allow this (which will be useful eventually for
some FPH5 metadata caching issues).
- Remove H5AC_find macro which was inlining metadata cache lookups,
and call function instead.
- Remove redundant memset() calls from H5G_namei() routine.
- Remove redundant checking of object type when locating objects
in metadata cache and rely on the address only.
- Create default dataset object to use when default dataset creation
property list is used to create datasets, bypassing querying
for all the property list values.
- Use default I/O vector size when performing raw data with the
default dataset transfer property list, instead of querying for
I/O vector size.
- Remove H5P_DEFAULT internally to the library, replacing it with
more specific default property list based on the type of
property list needed.
- Remove redundant memset() calls in object header message (H5O*)
routines.
- Remove redunant memset() calls in data I/O routines.
- Split free-list allocation routines into malloc() and calloc()-
like routines, instead of one combined routine.
- Remove lots of indirection in H5O*() routines.
- Simplify metadata cache entry comparison routine (used when
flushing entire cache out).
- Only enable metadata cache statistics when H5AC_DEBUG is turned
on, instead of always tracking them.
- Simplify address comparison macro (H5F_addr_eq).
- Remove redundant metadata cache entry protections during dataset
creation by protecting the object header once and making all
the modifications necessary for the dataset creation before
unprotecting it.
- Reduce # of "number of element in extent" computations performed
by computing and storing the value during dataspace creation.
- Simplify checking for group location's file information, when file
has not been involving in file-mounting operations.
- Use binary encoding for modification time, instead of ASCII.
- Hoist H5HL_peek calls (to get information in a local heap)
out of loops in many group routine.
- Use static variable for iterators of selections, instead of
dynamically allocation them each time.
- Lookup & insert new entries in one step, avoiding traversing
group's B-tree twice.
- Fixed memory leak in H5Gget_objname_idx() routine (tangential to
performance improvements, but fixed along the way).
- Use free-list for reference counted strings.
- Don't bother copying object names into cached group entries,
since they are re-created when an object is opened.
The benchmark I used to measure these results created several thousand
small (2K) datasets in a file and wrote out the data for them. This is
Elena's "regular.c" benchmark.
These changes resulted in approximately ~4.3x speedup of the
development branch when compared to the previous code in the
development branch and ~1.4x speedup compared to the release
branch.
Additionally, these changes reduce the total memory used (code and
data) by the development branch by ~800KB, bringing the development
branch back into the same ballpark as the release branch.
I'll send out a more detailed description of the benchmark results
as a followup note.
New internal API routines:
Added "reference counted strings" API for tracking strings that get
used by multiple owners without duplicating the strings.
Added "ternary search tree" API for text->object mappings.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
Other platforms/configurations tested?
FreeBSD 4.7 (sleipnir) serial & parallel
Solaris 2.6 (baldric) serial
Purpose:
bug fixed for windows
Description:
windows fopen functions by default opens a text mode file. It will fail for HDF5 file.
So to use HDopen HDread HDclose for test.
Solution:
Change from fopen, fread, fclose to HDopen, HDread, HDclose.
Platforms tested:
linux2.2.18smp, windows 2000
Misc. update:
Update MANIFEST if you add or remove any file.
Update release_docs/RELEASE for bug fixes, new features, etc.
Update applicable document files too.
Purpose:
Bug fix(kinda)
Description:
Windows handle "fopen, fseek, fread, fclose" C functions slightly different.
Solution:
change to HDopen, HDlseek, HDread, HDclose to cover both Unix and Windows.
Platforms tested:
arabica
Purpose:
New feature to H5Dget_offset
Description:
If user block is set, H5Dget_offset should be able to return the absolute
offset from the beginning of file.
Platforms tested:
eirene, arabica
Code Cleanup & New Tests
Description:
tmisc.c:
Switched from using H5_HAVE_COMPRESSION flag in favor of
H5_HAVE_FILTER_DEFLATE.
dsets.c:
Switched from using H5_HAVE_COMPRESSION flag in favor of
H5_HAVE_FILTER_DEFLATE.
Refactored I/O filter tests to allow new filters to be [more] easily
added.
Added tests for shuffle & deflate+shuffle I/O filters (if the filter(s)
are enabled).
Added test for creating a new dataset with a filter that is not
available.
Added test for attempting to read a dataset created with a filter that
is not available.
Platforms tested:
Tested h5committest {arabica (fortran), eirene (fortran, C++)
modi4 (parallel, fortran)}
FreeBSD 4.7 (sleipnir)
Purpose:
Add a testing case for shuffling algorithm.
Description:
A control test:
Since shuffling 8-bit array should not change the original result; so we use a
8-bit array to test whether after shuffling the array keeps the same values.
Solution:
Platforms tested:
eirene,arabica,modi4
Misc. update:
Update MANIFEST if you add or remove any file.
Update release_docs/RELEASE for bug fixes, new features, etc.
Update applicable document files too.
Purpose:
New function.
Description:
H5Dget_offset returns the offset of a dataset's data relative to the
beginning of a file.
Platforms tested:
arabica(simple function, one test should be enough.)
Misc. update:
RELEASE.txt
Code cleanup
Description:
Clean up some compile warnings and have these tests create their data files
with the correct file drivers and directories, etc.
Platforms tested:
FreeBSD 4.6 (sleipnir) w and w/o parallel
Linux 2.2.x (eirene) w/FORTRAN & C++
Solaris 2.7 (arabica) w/FORTRAN
IRIX64 6.5 (modi4) w/FORTRAN & parallel
