Improve configure's large-file support control.
Description:
Modified configure to now attempt to add defines necessary for
supporting largefiles on all systems, instead of solely on linux. This
is in response to user requests to enable largefile support on Solaris
by default, as well as to give extra control on AIX (instead of just
jamming the necessary flag into the config files).
The old --enable-linux-lfs flag is being deprecated in favor of the
--enable-largefile flag (enabled by default), which can be used on all
platforms. --disable-linux-lfs can still be used to disable largefile
support (on linux) when the --enable-largefile flag is not specified.
On systems where large files cannot be supported in this manner,
configure will report as such.
Tested:
h5committest
AIX (NCSA's blue_print machine)
duty, liberty, and linew.
Autotool Upgrade / Libtool Bug Fix
Description:
Updated autotools referenced in bin/reconfigure as follows:
Autoconf 2.64 --> Autoconf 2.65
Automake 1.11 --> Automake 1.11.1
Libtool 2.2.6a --> Libtool 2.2.6b-mcg
The referenced libtool version is a custom version of 2.2.6b. It
has been tweaked to fix a bug in libtool that occurs
when using PGI 10.0 compilers. A check incorrectly categorizes
the C++ compiler as version 1.0 instead of 10.0, and the link
line is subsequently set up incorrectly and fails to compile.
A patch has been made available and will be included in the next
release of libtool, but in the meantime I've applied the patch to a custom
installation as indicated above. This bin/reconfigure now references
the custom installation, and the resulting configure script will correctly
categorize the PGI 10.0 C++ compiler.
Ran bin/reconfigure to regenerate configure and makefiles.
Tested:
jam and amani with PGI 10.0 compilers.
h5committest
The mixed use of RUNTEST (original) and RUNTESTS (new) caused
confusion. E.g., the timings in test/ was still using the old
$RUNTEST. It made more sense to use $RUNTEST which is used
by the dejagnu feature of automake. So, I changed all
$RUNTEST or $RUNTESTS to $RUNEXEC.
config/commence.am & config/conclude.am are the two files
that got changes. Also fixed an error in test/Makefile.am.
The rest are changed by bin/reconfigure.
Tested: h5committested.
Bring r17924 from trunk to 1.8 branch:
Add detection for POSIX lstat() routine to configure script (mostly for
non-UNIX/Linux machines) and add macro wrapper for it.
Alphabetatize the system/library calls we test for, to make them easier
to read.
Removed the sigaction() detection & macro wrappers, since it's not
used by the distribution currently.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
(h5committested on trunk)
Merge from Trunk
Description:
Merging all of my CFLAGS-related changes from trunk to 1.8
This includes revision #s: 17616, 17625, 17627, 17639, and 17643.
Tested:
h5committest and some additional checks on our FreeBSD as well
as NCSA's machines.
Merge from trunk
Description:
Merged revisions 17228 and 17440 from trunk to 1.8
Specifically, this brings support of --enable-static-exec flag into 1.8
Tested:
Manually on jam, linew, smirom, liberty, plus h5committest.
Currently, there is no automatic regression test that exists due to
portability issues. Behavior is both different and undefined on certain
systems (and while 'nm' command seems to exist on all machines, behavior
is confirmed to be different on Mac, possibly others). Solution will be
to set up some sort of framework in daily tests to build statically,
remove shared paths, and verify executables can function.
Updating autotools
Description:
Bring revisions 17345 and 17357 from trunk to 1.8, specifically:
- Automake upgraded to 1.11
- Autoconf upgraded to 2.64
- bin/reconfigure script edited to use new versions (on jam), and
run to generate new configure script and Makefile.in's.
- configure.in script edited to add "_cv_" to all AC_CACHE_VAL strings
(in order to comply with new autoconf standard).
- bin/install-sh script replaced with new version as provided
by automake.
Tested:
- All issues on trunk were resolved, so only tested on jam and linew. Any
other outliers should be caught by daily tests.
Clean up various compiler warnings from IRIX64 build
Tested on:
Mac OS X/32 (amazon) debug & production
(yes, I know it's not an IRIX64 system :-)
Too minor to require h5committest
Description:
Previously, there was no versioning for H5Z_class_t. This prevented applications
written for 1.6 using custom filters from being able to use the 1.8 library.
There is now an H5Z_class1_t and H5Z_class2_t to enable compatibility. H5Zregister is
*not* versioned, it determines which version of the struct has been passed in by the
value of the first field (id or version, both are ints).
Tested: jam, linew, smirom (h5committest), jam (--with-default-api-version=v16)
- Updated bin/reconfigure to use latest version of automake (1.10.2).
Regenerated Makefile.in's by running bin/reconfigure.
- Added libtool version numbers to c++, fortran, hl, hl c++, and hl fortran
libraries.
Tested:
jam, liberty, smirom
Bug Fix
Description:
Fixing BZ #1381. The --includedir=DIR configure option, which is used to
spceify installation location of C header files, did not work correctly as
the path was hard-coded into config/commence.am. I'm presuming this is
because an older version of automake didn't know where to put C header
files. In any case, removing this line now defaults the includedir to the
same directory that it is currently hard-coded to, and also fixes the
configure flag which allows for customization of this value.
Tested:
jam, liberty
have 2 expected outputs for 2 h5ls runs depending if run on a big or little endian machine. Configure.in was modified to export a variable carrying endianess information to testh5ls.sh. This script then compares the current run with 2 expected outputs, one for a big-endian machine (linew was used to generate the output), other for little endian (jam was used to generate the output)
the way h5ls prints types, it starts searching for NATIVE types first. One solution would be h5ls not to detect these native types, using for example the same print datatype function that h5dump does, that would make the output look the same on all platforms ("32-bit little-endian integer" would be printed instead). Drawback, this "native" information would not be available. Other solution is to have not one but 2 expected outputs and make the shell script detect the endianess and compare with one output or other
tested: jam, linew
Description:
One open() call with O_CREAT flag did not have the third required argument.
Solution:
Added the right third argument. Also changed all open() calls to HDopen
macros for more portable coding.
Tested:
h5committested.
Description: Libtool wasn't working on linew because the script was
not working with the Solaris Bourne shell. Libtool has built-
in detection to ensure that it uses an appropriate shell,
but our config/commence.am was hard coding this shell to be
/bin/sh. Removing this line allows the shell to be picked up
by configure, thus allowing libtool to use the correct shell
on linew (bash). This was initially added to correct for a problem
on an old machine (janus), so shoudn't be needed for machines
we currently support.
Tested: full make check install on kagiso and linew, compile only
on smirom, duty, and liberty.
Description:
1) configure now sets LD_LIBRARY_PATH before checking for presence of
SZIP encoder, so user does not need to set this. The path is then
saved and sent to Makefiles, and used when "make check-install" is
invoked, so user doesn't need to set this manually.
2) Upgraded libtool to version 2.2.6a
3) Rearranged tools build order as h5dump depends on existence of h5diff
and h5import. Since h5dump is a sister directory as opposed to a parent
of either, it doesn't know about the build rules of these tools, so
setting any sort of explicit dependency confuses automake as it won't
know how to build the tools. Instead, setting CONFIG = ordered forces
an in-order traversal of each tools subdirectory. (without it will also
traverse in order by default, but this should prevent gmake's -j option
from jumping ahead as well). 'make check install' should now be able to
be invoked after 'configure' without causing a dependency failure.
4) Removed H5_HAVE_LARGE_HSIZET macro from vms/src/h5pubconf.h, as this macro
has now been out of the code for some time. (vms pubconf needs to be
updated manually, which is why it was still hanging around).
Tested: kagiso, smirom, liberty
Note: h5diff looks to be causing failures in h5copy and h5dump tools tests,
though these were present before any of my changes. My changes only
affect build order and configuration setup, and shouldn't prevent
fixes for these failures coming in after this checkin. Other than these,
tests pass fine.
Description:
Program would crash, complaining MPI calls were invoked after
MPI_Finalize() has occurred. Previously, H5close() before
MPI_Finalize() would remove the crash.
Solution:
It turned out that two H5 property objects (mpio_pl and acc_tpl)
were not closed before MPI_Finalize(). In the at_exit code,
HDF5 library attempted to close them by releasing the MPI Communicators
in them too. That was the error. Adding code to close them properly
before MPI_Finalize() took care things.
Tested:
Only in kagiso parallel. Did not run h5committest since
kagiso would have been the one running parallel test. This
part of code would not be compiled at all in non-phdf5 mode.
Remove trailing whitespace from C/C++ source files, with the following
script:
foreach f (*.[ch] *.cpp)
sed 's/[[:blank:]]*$//' $f > sed.out && mv sed.out $f
end
Tested on:
Mac OS X/32 10.5.5 (amazon)
No need for h5committest, just whitespace changes...
Description:
In library code, we try not to use system calls directly, but instead use the HD{function} macro instead. This way, we can map special versions of the call on particular systems. Previously, it was all done in H5private.h. However, in an effort to clean up platform-specific definitions, we move all of the Windows macros into a separate file, win32defs.h. This way, we can use the non-Posix versions that Visual Studio sends warnings about.
Some macros are set specifically in the platform-specific header files. Then, any macros left unset will be set by the "default" implementation in H5private.h.
This checkin also cleans up various source files to use the HD* macros when possible.
Tested:
VS2005 on WinXP
VS.NET on WinXP
h5committest (kagiso, linew, smirom)
Description: Changing h5perf installation so only h5perf is installed when
parallel is enabled, while only h5perf_serial is installed when
parallel is disabled.
Tested: kagiso
Description:
perf would crash at the end as it complained it was doing MPI calls after
MPI_Finalize() is called. I suspect the HDF5 library is making MPI calls
in the at_exit area. This should not happened.
Solution:
Added the H5exit() right before MPI_Finalize() fixed the problem.
But it needs to find out why HDF5 library is making MPI calls in
the at_exit() area. It should not.
Also, changed the temporay data file name from /tmp/test.out to
perftest.out, both to avoid multiple instance of this may use the
same /tmp/xxx name and also make the file name less generic.
Tested:
Kagiso PP.
Description:
The fortran Makefile.am used HDF_FORTRAN to indicate it is part of the
Fortran API source so that conclude.am will give fortran api prefix in the
test output. The symbox HDF_FORTRAN is also used in configure for a different
purpose (indicated --enable-fortran). They conflicted.
Similar problem for the symbol HDF_CXX.
Solution:
Changed all the involved Makefile.am to use "FORTRAN_API" instead. It is
a more appropriate name. Same for CXX_API.
Along the way, discovered that the Makefile.am of hl/fortran/test and
hl/cxx/test did not have those symbols at all. Added them in.
Platform tested:
Kagiso serial and h5committested (kagisopp, smirom, linew).
Description: Removing make target 'make check-perform'. The extra target is
unneeded because the tests don't take as long to run as initially
anticipated, and can be lumped in with 'make check'.
Tested: kagiso
Improvement.
Description:
src/libhdf5.settings was the initial configure summary and is installed.
Then configure is changed to dump a summary of the configure settings to
the output and also append it to src/libhdf5.settings. That created
two different output formats and duplicated information. This is the
initial attempt to clean up this confusion and unify the output format.
It is decided to use the src/libhdf5.settings template as the unified means.
This requires more macros symbols be defined. The following symbols are
all related to generating the src/libhdf5.settings file.
AC_SUBST(EXTERNAL_FILTERS)
AC_SUBST(MPE) MPE=no
AC_SUBST(STATIC_EXEC) STATIC_EXEC=no
AC_SUBST(HDF_FORTRAN) HDF_FORTRAN=no
AC_SUBST(FC) HDF_FORTRAN=no
AC_SUBST(HDF_CXX) HDF_CXX=no
AC_SUBST(CXX) HDF_CXX=no
AC_SUBST(HDF5_HL) HDF5_HL=yes
AC_SUBST(GPFS) GPFS=no
AC_SUBST(LINUX_LFS) LINUX_LFS=no
AC_SUBST(INSTRUMENT) INSTRUMENT=no
AC_SUBST(CODESTACK) CODESTACK=no
AC_SUBST(HAVE_DMALLOC) HAVE_DMALLOC=no
AC_SUBST(DIRECT_VFD) DIRECT_VFD=no
AC_SUBST(THREADSAFE) THREADSAFE=no
AC_SUBST(STATIC_SHARED)
AC_SUBST(enable_shared)
AC_SUBST(enable_static)
AC_SUBST(UNAME_INFO) UNAME_INFO=`uname -a`
The src/libhdf5.settings.in has CONDITIONAL's added to it too. The
untrue conditions turned into a "#" and these lines are cleaned by the
post processing script.
Platform tested:
h5committest on kagiso, smirom and linew. (smirom had failures not due to
my changes.)
