hdf5/RELEASE
2000-02-09 16:36:37 -05:00

517 lines
19 KiB
Plaintext

Release information for hdf5-1.2.0beta
------------------------------------
CHANGES SINCE VERSION 1.0.1
Documentation
-------------
* More examples
* Updated user guide, reference manual, and format specification.
* Self-contained documentation for installations isolated from the
Internet.
Configuration
-------------
* Better detection and support for MPI-IO.
* Recognition of compilers with known code generation problems.
* Support for various compilers on a single architecture (e.g., the
native compiler and the GNU compilers).
* Ability to build from read-only media and with different compilers
and/or options concurrently.
* Added a libhdf5.settings file which summarizes the configuration
information and is installed along with the library.
* Builds a shared library on most systems that support it.
* Support for Cray T3E and Windows/NT.
Debugging
---------
* Improved control and redirection of debugging and tracing messages.
Datatypes
---------
* Optimizations to compound datatype conversions and I/O operations.
* Added nearly 100 optimized conversion functions for native datatypes
including support for non-aligned data.
* Added support for bitfield, opaque, and enumeration types.
* Added distinctions between signed and unsigned char types to the
list of predefined native hdf5 datatypes.
* Added HDF5 type definitions for C9x types like int32_t.
* Application-defined type conversion functions can handle non-packed
data.
* Changed the H5Tunregister() function to use wildcards when matching
conversion functions. H5Tregister_hard() and H5Tregister_soft()
were combined into H5Tregister().
* Support for variable-length datatypes (arrays of varying length per
dataset element). Variable length strings currently supported only
as variable length arrays of 1-byte integers.
Data Spaces
-----------
* New query functions for selections.
* I/O operations bypass the stripmining loop and go directly to
storage for certain contiguous selections in the absense of type
conversions. In other cases the stripmining buffers are used more
effectively.
* Reduced the number of I/O requests under certain circumstances,
improving performance on systems with high I/O latency.
Persistent Pointers
-------------------
* Object (serial and parallel) and dataset region (serial only)
references are implemented.
Parallel Support
----------------
* Improved parallel I/O performance.
* Supported new platforms: Cray T3E, Linux, DEC Cluster.
* Use vendor supported version of MPIO on SGI O2K and Cray platforms.
* Improved the algorithm that translates an HDF5 hyperslab selection
into an MPI type for better collective I/O performance.
Tools
-----
* Enhancements to the h5ls tool including the ability to list objects
from more than one file, to display raw hexadecimal data, to
show file addresses for raw data, to format output more reasonably,
to show object attributes, and to perform a recursive listing,
* Enhancements to h5dump: support new data types added since previous
versions.
* h5toh4: An hdf5 to hdf4 converter.
CHANGES SINCE THE Version 1.0.0 RELEASE
* [Improvement]: configure sets up the Makefile in the parallel tests
suit (testpar/) correctly.
* [Bug-Fix]: Configure failed for all IRIX versions other than 6.3.
It now configures correctly for all IRIX 6.x version.
CHANGES SINCE THE BETA RELEASE
* Added fill values for datasets. For contiguous datasets fill value
performance may be quite poor since the fill value is written to the
entire dataset when the dataset is created. This will be remedied
in a future version. Chunked datasets using fill values do not
incur any additional overhead. See H5Pset_fill_value().
* Multiple hdf5 files can be "mounted" on one another to create a
larger virtual file. See H5Fmount().
* Object names can be removed or changed but objects are never
actually removed from the file yet. See H5Gunlink() and H5Gmove().
* Added a tuning mechanism for B-trees to insure that sequential
writes to chunked datasets use less overhead. See H5Pset_btree_ratios().
* Various optimizations and bug fixes.
CHANGES SINCE THE SECOND ALPHA
* Strided hyperslab selections in dataspaces now working.
* The compression API has been replaced with a more general filter
API. See doc/html/Filters.html for details.
* Alpha-quality 2d ragged arrays are implemented as a layer built on
top of other hdf5 objects. The API and storage format will almost
certainly change.
* More debugging support including API tracing. See Debugging.html.
* C and Fortran style 8-bit fixed-length character string types are
supported with space or null padding or null termination and
translations between them.
* Added function H5Fflush() to write all cached data immediately to
the file.
* Datasets maintain a modification time which can be retrieved with
H5Gstat().
* The h5ls tool can display much more information, including all the
values of a dataset.
CHANGES SINCE THE FIRST ALPHA
* Two of the packages have been renamed. The data space API has been
renamed from `H5P' to `H5S' and the property list (template) API has
been renamed from `H5C' to `H5P'.
* The new attribute API `H5A' has been added. An attribute is a small
dataset which can be attached to some other object (for instance, a
4x4 transformation matrix attached to a 3-dimensional dataset, or an
English abstract attached to a group).
* The error handling API `H5E' has been completed. By default, when an
API function returns failure an error stack is displayed on the
standard error stream. The H5Eset_auto() controls the automatic
printing and H5E_BEGIN_TRY/H5E_END_TRY macros can temporarily
disable the automatic error printing.
* Support for large files and datasets (>2GB) has been added. There
is an html document that describes how it works. Some of the types
for function arguments have changed to support this: all arguments
pertaining to sizes of memory objects are `size_t' and all arguments
pertaining to file sizes are `hsize_t'.
* More data type conversions have been added although none of them are
fine tuned for performance. There are new converters from integer
to integer and float to float, but not between integers and floating
points. A bug has been fixed in the converter between compound
types.
* The numbered types have been removed from the API: int8, uint8,
int16, uint16, int32, uint32, int64, uint64, float32, and float64.
Use standard C types instead. Similarly, the numbered types were
removed from the H5T_NATIVE_* architecture; use unnumbered types
which correspond to the standard C types like H5T_NATIVE_INT.
* More debugging support was added. If tracing is enabled at
configuration time (the default) and the HDF5_TRACE environment
variable is set to a file descriptor then all API calls will emit
the function name, argument names and values, and return value on
that file number. There is an html document that describes this.
If appropriate debugging options are enabled at configuration time,
some packages will display performance information on stderr.
* Data types can be stored in the file as independent objects and
multiple datasets can share a data type.
* The raw data I/O stream has been implemented and the application can
control meta and raw data caches, so I/O performance should be
improved from the first alpha release.
* Group and attribute query functions have been implemented so it is
now possible to find out the contents of a file with no prior
knowledge.
* External raw data storage allows datasets to be written by other
applications or I/O libraries and described and accessed through
HDF5.
* Hard and soft (symbolic) links are implemented which allow groups to
share objects. Dangling and recursive symbolic links are supported.
* User-defined data compression is implemented although we may
generalize the interface to allow arbitrary user-defined filters
which can be used for compression, checksums, encryption,
performance monitoring, etc. The publicly-available `deflate'
method is predefined if the GNU libz.a can be found at configuration
time.
* The configuration scripts have been modified to make it easier to
build debugging vs. production versions of the library.
* The library automatically checks that the application was compiled
with the correct version of header files.
Parallel HDF5 Changes
* Parallel support for fixed dimension datasets with contiguous or
chunked storages. Also, support unlimited dimension datasets which
must use chunk storage. No parallel support for compressed datasets.
* Collective data transfer for H5Dread/H5Dwrite. Collective access
support for datasets with contiguous storage only, thus only fixed
dimension datasets for now.
* H5Pset_mpi and H5Pget_mpi no longer have the access_mode
argument. It is taken over by the data-transfer property list
of H5Dread/H5Dwrite.
* New functions H5Pset_xfer and H5Pget_xfer to handle the
specification of independent or collective data transfer_mode
in the dataset transfer properties list. The properties
list can be used to specify data transfer mode in the H5Dwrite
and H5Dread function calls.
* Added parallel support for datasets with chunked storage layout.
When a dataset is extend in a PHDF5 file, all processes that open
the file must collectively call H5Dextend with identical new dimension
sizes.
LIST OF API FUNCTIONS
The following functions are implemented. Errors are returned if an
attempt is made to use some feature which is not implemented and
printing the error stack will show `not implemented yet'.
Library
H5check - check that lib version matches header version
H5open - initialize library (happens automatically)
H5close - shut down the library (happens automatically)
H5dont_atexit - don't call H5close on exit
H5get_libversion - retrieve library version info
H5check_version - check for specific library version
Property Lists
H5Pclose - release template resources
H5Pcopy - copy a template
H5Pcreate - create a new template
H5Pget_chunk - get chunked storage properties
H5Pset_chunk - set chunked storage properties
H5Pget_class - get template class
H5Pget_istore_k - get chunked storage properties
H5Pset_istore_k - set chunked storage properties
H5Pget_layout - get raw data layout class
H5Pset_layout - set raw data layout class
H5Pget_sizes - get address and size sizes
H5Pset_sizes - set address and size sizes
H5Pget_sym_k - get symbol table storage properties
H5Pset_sym_k - set symbol table storage properties
H5Pget_userblock - get user-block size
H5Pset_userblock - set user-block size
H5Pget_version - get file version numbers
H5Pget_alignment - get data alignment properties
H5Pset_alignment - set data alignment properties
H5Pget_external_count- get count of external data files
H5Pget_external - get information about an external data file
H5Pset_external - add a new external data file to the list
H5Pget_driver - get low-level file driver class
H5Pget_stdio - get properties for stdio low-level driver
H5Pset_stdio - set properties for stdio low-level driver
H5Pget_sec2 - get properties for sec2 low-level driver
H5Pset_sec2 - set properties for sec2 low-level driver
H5Pget_core - get properties for core low-level driver
H5Pset_core - set properties for core low-level driver
H5Pget_split - get properties for split low-level driver
H5Pset_split - set properties for split low-level driver
H5P_get_family - get properties for family low-level driver
H5P_set_family - set properties for family low-level driver
H5Pget_cache - get meta- and raw-data caching properties
H5Pset_cache - set meta- and raw-data caching properties
H5Pget_buffer - get raw-data I/O pipe buffer properties
H5Pset_buffer - set raw-data I/O pipe buffer properties
H5Pget_preserve - get type conversion preservation properties
H5Pset_preserve - set type conversion preservation properties
H5Pget_nfilters - get number of raw data filters
H5Pget_filter - get raw data filter properties
H5Pset_filter - set raw data filter properties
H5Pset_deflate - set deflate compression filter properties
H5Pget_mpi - get MPI-IO properties
H5Pset_mpi - set MPI-IO properties
H5Pget_xfer - get data transfer properties
H5Pset_xfer - set data transfer properties
Datasets
H5Dclose - release dataset resources
H5Dcreate - create a new dataset
H5Dget_space - get data space
H5Dget_type - get data type
H5Dget_create_plist - get dataset creation properties
H5Dopen - open an existing dataset
H5Dread - read raw data
H5Dwrite - write raw data
H5Dextend - extend a dataset
Attributes
H5Acreate - create a new attribute
H5Aopen_name - open an attribute by name
H5Aopen_idx - open an attribute by number
H5Awrite - write values into an attribute
H5Aread - read values from an attribute
H5Aget_space - get attribute data space
H5Aget_type - get attribute data type
H5Aget_name - get attribute name
H5Anum_attrs - return the number of attributes for an object
H5Aiterate - iterate over an object's attributes
H5Adelete - delete an attribute
H5Aclose - close an attribute
Errors
H5Eclear - clear the error stack
H5Eprint - print an error stack
H5Eget_auto - get automatic error reporting settings
H5Eset_auto - set automatic error reporting
H5Ewalk - iterate over the error stack
H5Ewalk_cb - the default error stack iterator function
H5Eget_major - get the message for the major error number
H5Eget_minor - get the message for the minor error number
Files
H5Fclose - close a file and release resources
H5Fcreate - create a new file
H5Fget_create_plist - get file creation property list
H5Fget_access_plist - get file access property list
H5Fis_hdf5 - determine if a file is an hdf5 file
H5Fopen - open an existing file
Groups
H5Gclose - close a group and release resources
H5Gcreate - create a new group
H5Gopen - open an existing group
H5Gpop - pop a group from the cwg stack
H5Gpush - push a group onto the cwg stack
H5Gset - set the current working group (cwg)
H5Giterate - iterate over the contents of a group
H5Gmove - change the name of some object
H5Glink - create a hard or soft link to an object
H5Gunlink - break the link between a name and an object
H5Gget_objinfo - get information about a group entry
H5Gget_linkval - get the value of a soft link
H5Gget_comment - get the comment string for an object
H5Gset_comment - set the comment string for an object
Dataspaces
H5Screate - create a new data space
H5Scopy - copy a data space
H5Sclose - release data space
H5Screate_simple - create a new simple data space
H5Sset_space - set simple data space extents
H5Sis_simple - determine if data space is simple
H5Sset_extent_simple - set simple data space dimensionality and size
H5Sget_simple_extent_npoints - get number of points in simple extent
H5Sget_simple_extent_ndims - get simple data space dimensionality
H5Sget_simple_extent_dims - get simple data space size
H5Sget_simple_extent_type - get type of simple extent
H5Sset_extent_none - reset extent to be empty
H5Sextent_copy - copy the extent from one data space to another
H5Sget_select_npoints - get number of points selected for I/O
H5Sselect_hyperslab - set hyperslab dataspace selection
H5Sselect_elements - set element sequence dataspace selection
H5Sselect_all - select entire extent for I/O
H5Sselect_none - deselect all elements of extent
H5Soffset_simple - set selection offset
H5Sselect_valid - determine if selection is valid for extent
Datatypes
H5Tclose - release data type resources
H5Topen - open a named data type
H5Tcommit - name a data type
H5Tcommitted - determine if a type is named
H5Tcopy - copy a data type
H5Tcreate - create a new data type
H5Tequal - compare two data types
H5Tlock - lock type to prevent changes
H5Tfind - find a data type conversion function
H5Tconvert - convert data from one type to another
H5Tregister_hard - register specific type conversion function
H5Tregister_soft - register general type conversion function
H5Tunregister - remove a type conversion function
H5Tget_overflow - get function that handles overflow conv. cases
H5Tset_overflow - set function to handle overflow conversion cases
H5Tget_class - get data type class
H5Tget_cset - get character set
H5Tget_ebias - get exponent bias
H5Tget_fields - get floating point fields
H5Tget_inpad - get inter-field padding
H5Tget_member_dims - get struct member dimensions
H5Tget_member_name - get struct member name
H5Tget_member_offset - get struct member byte offset
H5Tget_member_type - get struct member type
H5Tget_nmembers - get number of struct members
H5Tget_norm - get floating point normalization
H5Tget_offset - get bit offset within type
H5Tget_order - get byte order
H5Tget_pad - get padding type
H5Tget_precision - get precision in bits
H5Tget_sign - get integer sign type
H5Tget_size - get size in bytes
H5Tget_strpad - get string padding
H5Tinsert - insert scalar struct member
H5Tinsert_array - insert array struct member
H5Tpack - pack struct members
H5Tset_cset - set character set
H5Tset_ebias - set exponent bias
H5Tset_fields - set floating point fields
H5Tset_inpad - set inter-field padding
H5Tset_norm - set floating point normalization
H5Tset_offset - set bit offset within type
H5Tset_order - set byte order
H5Tset_pad - set padding type
H5Tset_precision - set precision in bits
H5Tset_sign - set integer sign type
H5Tset_size - set size in bytes
H5Tset_strpad - set string padding
Filters
H5Tregister - register a new filter method
Ragged Arrays (alpha)
H5Rcreate - create a new ragged array
H5Ropen - open an existing array
H5Rclose - close a ragged array
H5Rwrite - write to an array
H5Rread - read from an array
PLATFORMS SUPPORTED
-------------------
Operating systems listed below with compiler information and MPI library, if
applicable, are systems that HDF5 1.2.0beta was tested.
Compiler & libraries
Platform Information Comment
-------- ---------- --------
AIX 4.3.2 (IBM SP)
FreeBSD 3.2 gcc 2.8.1
HP-UX 10.20 cc A.10.32.03
IRIX 6.5 cc 7.2.1
IRIX64 6.5 (64 & n32) cc 7.30
mpt.1.3 (SGI MPI 3.2.0.0)
Linux (SuSE and RedHat) egcs-1.1.2 configured with
--disable-hsizet
OSF1 V4.0 DEC-V5.2-033
SunOS 5.6 cc 4.2 no optimization
gcc 2.8.1
SunOS 5.5.1 gcc 2.7.2 configured with
--disable-hsizet
TFLOPS O/S 1.0.4 cicc (pgcc Rel 1.7-6i)
mpich-1.1.2 with local changes
Unicos 2.0.4.61 (T3E) cc 6.2.1.0
mpt.1.3
Windows/NT 4.0/SP3 MSVC++ 5.0 and 6.0
Cygwin gcc egcs-2.91.57