file does not exist, the code tried to open the file with EXCL CREATE.
ROMIO cannot handle file-open with EXCL Create due to racing problem.
The first process creates the file which then fails all
other processes.
Solution: In the parallel code, turn on TRUNC mode to allow "late" open
calls to succeed too. It is safe to force TRUNC mode since the file is
known non-existing at that point. The MPIO implementation has to
provide the real solution.
Platform tested: O2K with ROMIO.
while he is visiting LLNL. I changed the default creation template
offset and length to 4. Will fix the problem later.
Changes since 19980205
----------------------
./src/H5H.c
./src/H5Hprivate.h
./src/H5O.c
./src/H5Ocont.c
./src/H5Odtype.c
./src/H5Oefl.c
./src/H5Olayout.c
./src/H5Oname.c
./src/H5Onull.c
./src/H5Oprivate.h
./src/H5Odspace.c
./src/H5Ostab.c
./src/debug.c
./html/H5.format.html
Added an extra 4-byte field after the heap magic number for
alignment on the DEC alpha. Changed object header message
alignment to 8-bytes.
./src/H5F.c
./src/H5Farray.c
./src/H5Ffamily.c
./src/H5Fistore.c
./src/H5Flow.c
./src/H5Fprivate.h
./src/H5Fsec2.c
./src/H5Fstdio.c
./src/H5Gnode.c
./src/H5O.c
./src/H5Odtype.c
./src/H5P.c
./src/H5Pprivate.h
./src/H5T.c
./src/H5Tconv.c
./src/H5Tpkg.h
./src/H5Tpublic.h
./src/H5V.c
./src/H5detect.c
./test/cmpd_dset.c
./test/dsets.c
./test/dtypes.c
./test/extend.c
./test/hyperslab.c
./test/istore.c
./test/th5p.c
./test/theap.c
Fixed a few irix64 warnings regarding size_t vs. int,
variables set but not used, printf formats
./config/irix64
Added `-woff 1196' to get rid of errors about __vfork() being
implicitly defined in a system header file.
./src/H5B.c
Fixed a stack alignment problem.
h5_write.c - stores INT array as INT32 little endian dataset in the HDF5 file.
h5_read.c - gets info about the dataset;
reads hyperslab from the dataset in the file into
2-dim slice of the 3 dimensional array.
h5_compund.c - creates compound dataset and writes it to the file;
reads subsets of the dataset.
h5_extend_write.c - writes extendable dataset.
h5_extend_read.c - gets info about the dataset;
reads dataset and one of the chunks.
failures. Row back the offset and length sizes to 4 (instead of
sizeof(size_t)) for now to pass all tests. Will fix it later.
(Also changed the default create template in H5F.c).
not assign the decoded value back to n.
Solution: Removed temporary variable _n (don't see why it is needed.)
Use the variable n directly.
Platform tested: IRIX64 -64
----------------------
./src/H5H.c
Changed a temporary buffer from 20 bytes to 52 bytes.
./test/tfile.c
Default sizes for file addresses and sizes are the same as
sizeof(size_t) on the machine that creates the file.
The default file addr and size sizes for new files are the
same as sizeof(size_t) on the machine that creates the
file. They used to always be 4.
./src/H5Pprivate.h
./src/H5Psimp.c
Fixed `start' and `nelmts' arguments to be `size_t'.
----------------------
./src/H5C.c
./src/H5Cprivate.h
./src/H5D.c
./src/H5Dpublic.h
Added H5Dget_create_parms(), trying to stay one step ahead of
Elena ;-)
----------------------
./MANIFEST
Added H5Fmpio.c. One way to check that you've properly
included new files is to run `./bin/release none', then unpack
the resulting tar file somewhere and try to compile it. The
tar file will be ./releases/hdf-5.0.0a.tar.
./config/freebsd2.2.1
./config/linux
Added `-ansi'
./acconfig.h
Added definitions for PHDF5 and HAVE_PARALLEL.
./configure.in
Added minimal support for parallel build. Kim and Albert will
have to flesh this out or I can do it if they're more specific
about what they need.
./config/commence.in
Added default value for $(RUNTEST)
./config/conclude.in
`make test' uses value of $(RUNTEST) to run test cases.
./src/Makefile.in
Added PARALLEL_SRC for conditional compilation of H5Fmpio.c
./src/H5D.c
./src/H5Dpublic.h
Added H5Dget_type() for Elena
NOTE: These changes require that configure be rerun. If you're using
GNU make it will happen automatically, otherwise do it by
hand.
----------------------
./MANIFEST
./html/study.html [NEW]
./html/study_1000x1000.gif [NEW]
./html/study_250x250.gif [NEW]
./html/study_499x499.gif [NEW]
./html/study_5000x1000.gif [NEW]
./html/study_500x500.gif [NEW]
./html/study_p1.gif [NEW]
./html/study_p1.obj [NEW]
The conclusions from a study of the HDF5 chunking
layout. Shows overhead and disk seeking.
----------------------
./INSTALL
Added instructions for which C flags to set for debugging.
./src/H5C.c
./src/H5Cpublic.h
H5Cset_chunk() takes const pointer.
./src/H5D.c
./src/H5Dprivate.h
./src/H5Dpublic.h
Added H5Dextend() to extend the dimensions of a dataset.
./src/H5Osdspace.c
./src/H5P.c
./src/H5Pprivate.h
./src/H5Ppublic.h
./test/cmpd_dset.c
./test/dsets.c
./test/th5p.c
Added the optional `maxdims' argument to H5Pcreate_simple()
and defined constant H5P_UNLIMITED which can appear in the
maxdims. Added `const' to arguments.
Implemented H5Pcopy()
Removed the unused file argument from H5P_modify.
Added H5P_extend().
Removed the `flags' field from simple data types and we
determine if the `max' or `perm' arrays are valid by looking
at the pointer. Cleaned up the H5O_sdspace_debug output.
./src/H5T.c
Fixed a printf format.
./MANIFEST
./test/Makefile.in
./test/extend.c [NEW]
Added a test for multi-dimensional unlimited dimensions.
----------------------
./INSTALL
./Makefile.dist
./Makefile.in
./config/conclude.in
Some stupid make's don't know what PHONY means. I added
changed config to _config in the default makefile and added
_test to the other makefiles.
./src/H5Psim.c
A `file_offset' argument got changed to `int' accidently that
causes tests to fail on 64-bit machines. I changed it back to
size_t.
./src/H5T.c
Changed a couple printf formats.
----------------------
./RELEASE
Added Library functions that I missed the first time.
./html/Datasets.html
Added an example for Elena's question about how to read a
single member of a compound data type so it becomes an array
of that member in memory.
./src/H5Pprivate.h
Fixed the prototype for H5P_get_hyperslab() to match the
definition.
./src/H5Psimp.c
Oops, added the kludge back in for the offset argument, which
is still an `intn' instead of a `size_t'.
./src/H5.c
./src/H5public.h
./src/H5F.c
./src/H5T.c
./src/H5Tpublic.h
Changed H5init() to H5open() and added an H5close() to fit our
create/open/close paradigm. The H5open() happens
automatically on the first call to the HDF5 library. The
H5close() happens automatically on exit() (unless the app
turns off that feature). H5close() closes all datasets and
files and releases all resources used by the library.
./test/dsets.c
Added calls to H5open() and H5close() to test them.
./test/dtypes.c
Removed call to H5init() since we no longer need it there.
./src/H5Fstdio.c
Changed the PABLO_MASK to the right value. Thanks Kim.
Problem:
Step 8 failed because H5Pcreate_simple expects size_t dimension
array but the code used an int array. (In IRIX64 -64 mode, size_t
is a 64bit unsigned long but ints are only 32 bits long.) casting
it to size_t* just avoided warning message but did not change the
data type.
Solution:
Throw in a kludge by using a temporary dimension array of size_t.
Can't change the type of h_size since it is also used for the hyperslab
routine which expects an int dimension array. The Hyperslab routine
will be changed. Put in a patch for now.
