Cache the dataset's rank & dimension sizes, instead of querying them
frequently, to speed up various checks & algorithms. Also, a few minor
cleanups.
Tested on:
MacOSX/64 10.10.2 (amazon) w/serial & parallel
Linux/32 2.6.18 (jam) w/serial & parallel
Minor formatting cleanups and remove unused field from callback struct.
Tested on:
Mac OSX/64 10.10.2 (amazon) w/serial & parallel
(Too minor to require h5committest)
configure.ac
hl/src/H5LTparse.c
hl/src/H5LTparse.h
Address HDFFV-9010: configure issue with gcc 4.9.2 "-l ".
Change default to disable_shared on CYGWIN.
release_docs/INSTALL_Cygwin.txt
release_docs/RELEASE.txt
Update for changed default and no CYGWIN szip binary.
c++/src/cpp_doc_config
bin/h5vers
AddressHDFFV-9010: Add cpp_doc_config to h5vers
Tested with h5committest, on CYGWIN.
Separate allocating chunk on disk from inserting the chunk record into the
index. This allows a "SWMR-safe" insert/update of chunks (the chunk is always
allocated -> written -> inserted/updated in the index).
Tested on:
Mac OSX/64 10.10.2 (amazon) w/parallel & serial
Linux/32 2.6.18 (jam) w/serial & parallel
Description:
- In DataType::DataType(const PredType& pred_type), using DataType::copy
will invoke DataType::close() unnecessarily, which will produce undefined
behavior. Changed to call H5Tcopy directly, code reuse is not useful in
this case.
- Also, fixed CommonFG::childObjVersion to return expected value outside of
an if/else block.
Platforms tested:
Linux/ppc64 (ostrich)
Linux/64 (platypus)
Linux/32 2.6 (jam)
That broke the testings as some testfiles have zlib compressed datasets.
Added options control to enable the linking of zlib external libarary by
default and turn off the szip library linking as szip library may not be
avaiable. This matches the established settings.
Tested: run cmakehdf5 by hand in jam and platypus.
Also tested in wren but it failed in the testing stage.
Also tried "cmakehdf5 --script" in jam. It failed.
Description:
The test added failed in some machines because the data file contains infinity values that different machines print them differently as "inf", "INF", "Inf", ...
Solution:
Added a "ignorecase" option to TOOLTEST() to do caseless matching between generated output vs expected output. This solved most machines problem for now.
Tested: h5committest, emu by hand for both development and production modes.
But cmake built h5dump failed to read the data file. Using the same source to build h5dump by autotools produced a h5dump that can read the test data file. Don't know why cmake could not produce a correct binary.
Description:
- Put back the UNUSED parameters in dsets test because the change to remove
the warning last time caused failure in setting filter, in turn, caused
failure in the test with such obscure/unrelated errors!
- Added incRefCount() to other constructors that missed from last time.
Platforms tested:
Linux/64 (platypus)
Linux/32 2.6 (jam)
SunOS 5.11 (emu)
When the selection is set to all, H5Sextent_copy did not update the number of
elements in the selection in the destination space. Fixed H5Sextent_copy to do
this. Added tests for this functionality.
Tested: jam, koala, ostrich (h5committest)
