Bring r17148:17149 from trunk into 1.8 branch:
Refactor how chunked dataset information is computed, moving it earlier and
avoiding more recomputation.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
FreeBSD/64 6.3 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (jam) w/PGI compilers, w/C++ & FORTRAN, w/threadsafe,
in debug mode
Linux/64-amd64 2.6 (smirom) w/Intel compilers w/default API=1.6.x,
w/C++ & FORTRAN, in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, in production mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Linux/64-ia64 2.4 (tg-login3) w/parallel, w/FORTRAN, in debug mode
Linux/64-amd64 2.6 (abe) w/parallel, w/FORTRAN, in production mode
Mac OS X/32 10.5.7 (amazon) in debug mode
Mac OS X/32 10.5.7 (amazon) w/C++ & FORTRAN, w/threadsafe,
in production mode
Bring r17146 from trunk to 1.8 branch:
Bring a bunch of misc. improvements & tweaks from the revise_chunks
branch back to the trunk, so that future merges won't be so painful.
Tested on:
FreeBSd/32 6.3 (duty)
(h5committested on trunk)
replaced it with H5Aopen in this round. I also updated some hard-coded numbers with some
macros.
Tested on jam only. I have tested the same change for the trunk.
the attribute structure wan't set correctly. It caused some confusion in H5A_read. This
flag was actually redundant because the library can alwasy check if the data buffer is
present. To fix it, I removed the "initialized" flag in the attribute structure and let
H5A_read check the data buffer.
I tested it on jam only but have tested the same change for the trunk on jam, smirom,
and linew.
Description:
Previously, when H5Z_filter_info was called for a filter that was not present,
it would succeed and return the information from one position past the end of
the pipeline, possibly causing a segfault. This affected at least
H5Pget_fitler_by_id1/2, and possibly other API functions. Fixed to properly
return failure.
Tested: jam, linew, smirom (h5committest)
Bring r17010,17017,17021-17022,17029-17031,17035 back from trunk:
"temporary" file allocation space changes.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
FreeBSD/64 6.3 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (jam) w/PGI compilers, w/C++ & FORTRAN, w/threadsafe,
in debug mode
Linux/64-amd64 2.6 (smirom) w/Intel compilers w/default API=1.6.x,
w/C++ & FORTRAN, in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, in production mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Linux/64-ia64 2.4 (tg-login3) w/parallel, w/FORTRAN, in debug mode
Linux/64-amd64 2.6 (abe) w/parallel, w/FORTRAN, in production mode
Mac OS X/32 10.5.7 (amazon) in debug mode
Mac OS X/32 10.5.7 (amazon) w/C++ & FORTRAN, w/threadsafe,
in production mode
Description:
When using H5T_copy on committed datatypes that are already open, H5T_copy would
properly use the already existing shared struct, but would still deep copy all
of the fields in that struct. This would cause memory leaks, and in the case of
a compound containing a vlen (or reference), the change in size would cause the
size of the resulting type to be set to an incorrect value. Changed H5T_copy to
properly avoid deep copies when using a reopened shared struct.
Tested: jam, linew, smirom (h5committest), purify on jam
Merge these trunk revisions which occurred during the 1.8.3 release code
freeze back to the 1.8 branch:
From Quincey: 16845 16847 16849 16851 16858 16869 16897
From Ray: 16859 16880
From Allen: 16863
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
FreeBSD/64 6.3 (liberty) w/C++ & FORTRAN, in debug mode
Linux/32 2.6 (jam) w/PGI compilers, w/C++ & FORTRAN, w/threadsafe,
in debug mode
Linux/64-amd64 2.6 (smirom) w/Intel compilers w/default API=1.6.x,
w/C++ & FORTRAN, in production mode
Solaris/32 2.10 (linew) w/deprecated symbols disabled, w/C++ & FORTRAN,
w/szip filter, in production mode
Linux/64-ia64 2.6 (cobalt) w/Intel compilers, w/C++ & FORTRAN,
in production mode
Linux/64-ia64 2.4 (tg-login3) w/parallel, w/FORTRAN, in debug mode
Linux/64-amd64 2.6 (abe) w/parallel, w/FORTRAN, in production mode
Mac OS X/32 10.5.6 (amazon) in debug mode
Mac OS X/32 10.5.6 (amazon) w/C++ & FORTRAN, w/threadsafe,
in production mode
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)
Description:
When writing data to a dataset, the data transform was performed after type conversion.
This caused an error if the file type was non-native. This has been changed so data
transforms are always performed on the memory type.
Tested: jam, linew. smirom (h5committest)
- 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
source directory. Since the test opens the files with write mode, it's safer to make copies of
the original files. I added the copying step in test_family_compat in vfd.c. I changed the
Makefile.am to clean up the copied files. I created a new family_v16_00000.h5 with v1.6 library.
Tested on jam - supposed to be simple change.
Description:
The H5L interface was not marked as uninitialized when H5L_term_interface was
called. This caused the interface to not initialize itself later on. This
prevented external links from working after calling H5close, because they were
never re-registered.
Tested: jam, smirom, linew (h5committest)
Description:
Versions of the library between 1.3.0 and 1.6.3 have a bug which prevents them
from opening any file that does not have the root group's symbol table
information cached in the root group's entry in the superblock. Prior to 1.8
this was not an issue as this information was always cached. However, 1.8.0
stopped writing this information (which is not required by the file format
specification), and these older versions can therefore not read files created or
last written by versions 1.8.0 to 1.8.2. This fix modifies the library to once
again add this information to the superblock (when using the old file format).
Tested: jam, linew, smirom (h5committest)
Description:
Changed Skip list package to use a deterministic skip list. This allows the
skip list package to avoid calling rand() and srand(), even on machines without
rand_r(). There is no longer a p-value or maximum level for skip lists.
Tested: jam, smirom, linew (h5committest)
Description:
Previously, H5Lcopy and H5Lmove would (through H5L_move) improperly apply the
"create intermediate groups" property to the source path traversal, and not the
destination. Fixed it to apply the property to the destination and not the
source. Also fixed H5Lcreate_ud to reject internal link classes without
throwing an assertion.
Tested: linew, jam, smirom (h5committtest)
some argument check in the internal function H5I_find_id and took out the assertion check.
I also removed the argument check in H5Iis_valid because it's in H5I_find_id now.
Tested on jam and linew - simple change.
Description:
Due to a bug in H5F_super_read, every time a file with a user block was opened
it would grow by the size of the user block. The bug has been fixed, and
comments have been added to clarify when an eoa address should be relative and
when it should be absolute.
Tested:
jam, linew (h5committest)
v1.8 library, v1.8 library writes the driver info block in the superblock. But v1.6
doesn't write it. This caused the data after the superblock to be overwritten. The
solution is to use a flag to indicate when the original file doesn't have the driver
info, v1.8 library doesn't write it either.
Tested on jam. I have some trouble to access smirom and linew using h5committest.
Description:
Some files apparently exist in the wild which have corrupt symbol table messages
on the root group. These files can be opened by 1.6 (which uses the cached
information in the superblock) but not by 1.8. This patch fixes 1.8 and 1.9 so
they can now open these files, and will correct them if necessary. Also fix
some potential (rare) problems with array datatype versions.
Tested: jam (parallel; h5committest not working on linew or smirom)
Bring r16574 back from trunk:
Clean up many compiler warnings and make test return non-zero exit code
when a failure is detected.
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
(other platforms tested on trunk)
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
test the correctness of the data when the fill value is defined or not. The
library should let the chunks bypass the cache depending on the size of the
chunks and whether to write fill value to the chunks.
Tested on jam - simple change.
Description:
In some situations it was possible for the fill value to not be written to parts
of a chunked dataset, particularly when extending and/or shrinking. Prior to
the fix for the chunk cache (1015) these bugs would have been exceedingly rare.
Tested: jam, smirom, linew (h5committest)
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
bug fix: TESTING2 macro introduced recently for the test suite (it adds an extra space at the beginning of output, for indentation) was already available for HL and caused compilation problems on AIX. Replaced the name with TESTING_2
tested: h5committest
Bring r16488 back from trunk:
Clean up code and eliminate resource leaks. Also avoid "null" I/O when
chunk doesn't exist and we can skip it.
Tested on:
Mac OS X/32 10.5.6 (amazon)
(too minor to require h5committest)
Bring r16483 back from trunk:
Clean up (i.e. remove) more internal calls to H5E_clear_stack(), along with
some other minor code cleanups.
Tested on:
Mac OS X/32 10.5.6 (amazon)
(too minor to require h5committest)
Description:
The meaning of the "nbytes" field in H5D_rdcc_t was not clear, and some places
assumed it was the maximum size of the chunk cache, while some assumed it was
the current size of the chunk cache. The end result was that only 1 chunk could
be held in cache at a time. This field has been replaced by "nbytes_max" and
"nbytes_used". Performance of cached I/O should improve greatly.
Tested: jam, smirom (h5committest)
file handles.
Description:
An attribute's "oloc" field which specifies the file it resides in was located
in the attribute's "shared" structure. So when an attribute was opened multiple
times all of the handles for that attribute pointed to the same file id, even if
different file id's were used to open the different handles for the attribute.
The "oloc" has been moved to the top level H5A_t struct.
Tested: jam, smirom (h5committest)
Description:
Since the new object header format, it has been possible for a situation to be
created where none of the messages are large enough to hold a continuation
message and there are no null messages to merge with. This makes it impossible
to add a new object header chunk. This case will now be handled by moving every
message in the last chunk to the newly allocated one, except for null messages
which are deleted.
Tested: jam, smirom (h5committest)
Description:
When an attribute was created with a datatype or dataspace that was shared in
the same object header that the attribute was in, the attribute could not be
deleted. Changes made to ensure that the attribute can be deleted both when the
attribute is in the object header and when it is shared in the heap. Object
header message decode routines now take an "open_oh" parameter to enable them to
avoid opening the same object header twice.
Tested: jam, smirom (h5committest)
on disk, the library still loaded it in the cache, which is redundant. I changed it to bypass the
cache and added a test in dsets.c.
Tested on jam and smirom.
Bring r16435 from revise_chunks branch back to 1.8 branch:
Expand object copy tests for chunked datasets to include 1-D datasets
with an unlimited dimension. (Fix typo in comment for test/links.c)
Tested on:
FreeBSD/32 6.3 (duty) in debug mode
(more thoroughly tested already on revise_chunks branch)
#1449 Change set_extent test to use standard standalone test framework
Call h5_fixname (with an array of test filenames) for generating the
filename to create and then call h5_cleanup() when the tests pass, to delete the files
created and close the FAPL from h5_fileaccess().
Defined a macro
#define TESTING2(WHAT) {printf(" Testing %-62s",WHAT); fflush(stdout);}
Similar to TESTING, except that it has a initial indentation space.
The effect is for nested loop tests
Testing with old file format:
Testing with fill value, no compression PASSED
tested: windows, linux
