Revision 1.2 2004/08/02 15:05:39 gosselin
pyhdf-0.5.2
Revision 1.1 2004/08/02 14:45:56 gosselin
Initial revision
List of changes to 'pyhdf' since the initial release.
pyhdf-0.8-3
-----------
October 2, 2008
- Regenerated the SWIG files with SWIG 1.3.33 in order to work correctly on
some 64-bit systems.
- Fixed HDF4Error to properly handle its arguments.
- Use the full HDF4 headers in the extension module so as not to rely on
implicit declarations. This appears to fix some bugs on 64-bit platforms.
pyhdf-0.8-2
-----------
August 18, 2008
- Removed ODL parser including examples and aster examples. These are now
in scikits.rsformats.
pyhdf-0.8-1
-----------
August 4 2008
- Numeric replaced with NumPy as underlying numerical package.
Support for the now phased out 'Numeric' package has been dropped.
- Parser for ODL (Object Description Language) added
(courtesy of Enthought Inc).
- The swig interface file is now part of the distribution.
- Fixed bug in the 'setscale()' method of an SDim instance.
- Users working with a debian-based linux distro reported problems with
the HDF SDgetcompress() / SDsetcompress() compression-related functions.
The cause seems to be related to the use of an HDF library whose version
predates 4.2r1. Debian appears very slow moving to the most
recent HDF release. To avoid depriving debian users of the use of pyhdf,
macro 'NOCOMPRESS' can be set inside 'setup.py'. This will transform
functions SDsetcompress() and SDgetcompress() into no-ops which
will return -1 (HDF error code) if called.
- Examples of the use of the HDF compression functionnalities have been
completely reworked. See directory 'examples/compress' for details.
pyhdf-0.7-3
-----------
July 13 2005
Interim release.
- The main motivation for this release is to benefit from HDF-4.2r1 bug fixes and
optionally the SZIP compression method allowed by HDF4.2 on SDS datasets.
- To use this release, you need to install HDF4.2r1. It will *NOT* work over
HDF4.1. To compress datasets using SZIP, szip 2.0 library must also be installed,
and your HDF installation must have been compiled with the szip option enabled.
- Conversely, pyhdf versions preceeding 0.7-3 will *NOT* work over HDF4.2.
Only pyhdf-0.7-3 will work with HDF4.2.
- Added new compression type SDC.COMP_SZIP, and defined new constants SDC.COMP_SZIP_EC
and SDC.COMP_SZIP_NN to configure the behavior of SZIP.
- Updated methods SDS.sdsetcompress() and SDS.getcompress() to allow compression using
SZIP compression, and inquire about SZIP compression configuration.
- All classes are now of "new-style" type (eg they derive from calss 'object').
- Updated documentation.
pyhdf-0.7-2
-----------
June 9 2004
Bug fix release.
- Do not strip trailing NULL at end of string attributes.
- Single-valued attributes were incorrectly processed in the set()
method of the SDAttr class.
- The following feature was not correctly documented.
When attributes are accessed using the "dot" notation, it is important
to be able to differentiate between true python attributes and
HDF attributes. For ex., if 'sds' is an HDF SD instance and one writes
'sds.x = "abcd" ', should attribute 'x' be considered as an HDF attribute and
its value written in the HDF file, or as a python attribute and stored in the
'sds' object dictionnary ?
To solve that problem, an attribute is considered a python attribute if it
starts with an '_', and as an HDF attribute otherwise. Note that this only
applies when using the "dot" notation (or the getattr()/setattr() equivalent
functions). If one needs to define an HDF attribute whose name starts with a
'_' (eg: '_FillValue'), allocate an SDAttr instance and use its get()/set()
methods.
Ex.:
sds = SD('myfile.hdf').select('mydataset')
sds._FillValue = -999 wrong: _FillValue not stored in the HDF dataset
sds.attr('_FillValue').set(SDC.FLOAT32, -999.0) good!
pyhdf-0.7-1
-----------
December 16 2003
- Implementation of the V (vgroup) API.
pyhdf-0.6-1
-----------
December 3 2003
-Implementation of the VS (vdata) API.
-Added support for unsigned integer types UINT16 and UINT32 inside
the SD API. Version 22 of Numeric must however be installed
(see comments inside `setup.py' for a way to remove support for
unsigned integers and compile with a lower version number of Numeric).
-Improved documentation.
-Bug fixes (thanks to Richard.Andrewsesands.com)
pyhdf-0.5-2
-----------
August 3 2003
-After starting the implementation of another HDF API, it
was found that the idea of having on C extension module
per HDF API was not a good one. It made difficult the sharing
of HDF data structures between the C extension modules.
It was thus decided to have just one C extension module
shared between all the python modules.
-The new C extension module is named "_hdfext". The SD python
module has been modified to import "_hdfext" instead of "_sdext".
-A new "error" module has been added.
pyhdf-0.5-1
-----------
July 31 2003
-Initial release.