[svn-r24612] fix bugs in parallel tests exposed in corner cases when running with 1
or 2 processes.
First bug is in testpar/t_mdset.c, where the test reports an error in
addition to skipping the test if there are less than three procs. Fix
to just skip the test.
Second bug is in testpar/t_dset.c in actual_io_mode tests, where
incorrect expected value for IO mode was set if the number of procs
running the test is 1.
tested with h5committest.
[svn-r24611] This is a partial fix for HDFFV-8634. It fixes the problem without --enable-fortran2003, where the optional parameter should not be given as an option for an interface to a C function. --enable-fortran2003 does not work due to a compiler bug in intel 14.0
Tested: kite (intel 14.0 with --enable-fortran)
[svn-r24604] Feature: HDFFV-8336
Changed to use the quick steps described in INSTALL_CMake.txt.
Previous version is available by --old.
Added a help page to explain how to use.
Added a fix to figure out the current HDF5 version and use that to find the
correct file for install. (Still need to figure out the platform name.)
Tested: Jam and Koala
[svn-r24598] allocate and use the H5AC_aux_t struct even if num_procs == 1. This
was a bug exposed by running a single process with a program that
generates lots of dirty metadata that needs to be flushed.
h5committest done!
[svn-r24480] Description:
Bring r24479 from 1.8 branch to trunk:
Fix ph5diff worker command handling so it can correctly respond to
situations where there's no work to perform. Also, a few minor code cleanups.
Tested on:
Mac OSX/64 10.9.0 (amazon) w/parallel
(too minor to require h5committest)
[svn-r24476] Description:
Remove duplicate "intent" flags from top & shared file structures, leaving
them only in the shared file structure.
Tested on:
Mac OSX/64 10.9.0 (amazon) w/gcc 4.8, both parallel & serial
(too minor to require h5committest)
[svn-r24471] JAVA-1798: Add the testfiles for hdf-java and the h5dump tests including the test to read filter data.
Tested: cmake and autotools - local linux
[svn-r24467] tfilters.h5 was created and checked in on 21 Feb 05, rev 10060, and never regenerated. The code for scaleoffset continued to change and the number of params for scaleoffset changed from 7 to 20 on the checkin of 23 Mar 05, rev 10384.
The h5dump tests only looked at headers and never the data. Data reads would fail with invalid number of params (7 expected 20).
Regenerated file and replacement ddls updated.
New h5dump test will be added to read the data.
[svn-r24463] Issue 8484 - Segfault during H5Fopen of corrupt file. I put a condition check to make sure a pointer isn't NULL.
Tested on jam - very simple change.
[svn-r24416] Fixed some compiler warnings in the dsets test.
Tested on:
32-bit LE linux (jam)
(These are just minor casting changes in the tests, hence the minimal
testing.)
[svn-r24400] Maintenance: When the libraries are built and tested in place, several files are left
behind by "make distclean" (see below); modified appropriate Makefile.am files and ran
bin/reconfigure to create Makefile.in to fix the problem.
Here is the list of the files "make disclean" doesn't remove:
tools/h5dump/testfiles
tools/h5repack/testfiles/h5diff_attr1.h5
tools/h5repack/testfiles/tfamily00000.h5
tools/h5repack/testfiles/tfami...
[svn-r24392] Cleanup testfiles from scripts. Update lists of files to be cleaned. CMake update to clean testfiles.
Tested: local cmake, autools, and h5committest
[svn-r24380] The parallel test code now uses HDfree, HDmalloc, and HDcalloc
exclusively.
Part of the preparation for a fix for HDFFV-8551.
Tested on:
32-bit LE linux (jam) w/ parallel and Fortran.
There are no behavior changes, so testing was minimal.
[svn-r24378] Fixed some commenting typos in configure.ac and reconfigured.
Tested on:
32-bit LE linux (jam)
There are no code changes. A build/test was done just to confirm.
[svn-r24376] Updated all code to use HDmalloc, HDcalloc, and HDfree.
Prep work for adding a memory free function, part of HDFFV-8551.
Tested on:
32-bit LE linux (jam)
* with C++, Fortran
* with thread safety on
These are minor changes (almost entirely renaming foo functions
to HDfoo) that don't change functionality. This should not
require a full commit test.
[svn-r24368] Issue 8147 - clang's -fcatch-undefined-behavior exposed undefined behavior in the alignment detection algorithm for integers in H5detect.c. Quincey decided to solve it in two steps. First, catch illegal instruction signal in the original algorithm. Second, use POSIX __alignof function to detect alignment if it's available. This checkin is the first step.
Tested on wren, jam, ostrich, platypas, and koala.
[svn-r24360] Jira issue 8528: H5Fget_obj_ids overfilled the list of object IDs by one. This is the second round of checkin after
receiving review comments from people. I put the safeguard in both H5F_get_objects and H5F_get_objects_cb to prevent
overfill the list.
tested with h5committest.
[svn-r24358] Changed the big test so that it runs using the h5_fileaccess() fapl.
The previous behavior was to execute a sec2, stdio, and family
test everytime it ran, which was very inefficient under check-vfd.
Tested on:
32- and 64-bit LE linux (jam and koala)
64-bit BE linux (ostrich)
64-bit OS-X 10.6 (fred)
[svn-r24333] Added a few volatile keywords to H5detect.c to quiet Visual Studio warnings. Also added a void to a Windows-specific function.
These are both trivial changes that don't affect behavior.
Tested on: Visual Studio 2010 on 64-bit Windows 7
