| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
possibly all) daily testing. When the behavior is acceptable, the changes should be merged into the regular release script as a --revision option.
|
|
|
|
| |
Merge from trunk
|
| |
|
| |
|
|
|
|
|
|
| |
- Moving truncate to H5F_flush() where it was done a long time ago.
- Add a second H5AC_flush() call after the truncate is done to make sure we write the correct EOA.
- Add a second flush call after releasing file space since that could result in an EOA change
|
|
|
|
|
|
|
| |
resulting dataset.
They have been updated to reflect the actual results of running the code.
(See HDFFV-637)
|
|
|
|
|
|
| |
- fix more instances where both mpi is_initialized and is_finalized need to be checked before making any MPI calls.
tested with h5committest.
|
|
|
|
|
|
|
| |
Chane addressed HDFFV-9010: configure issue with gcc 4.9.2; libtool produced "-l " with no lib name.
Also Scot noted that the -i option for sed does not work on Solaris or Mac, so the sed output is written to a temp file and then moved to the libtool file
Tested h5committest.
|
|
|
|
|
|
|
|
| |
fix for HDFFV-9412:
- Check for mpi is_initialized and is_finalized before making MPI calls in init/terminate of HDF5.
- Add a test to make sure we can make serial HDF5 calls from rank 0 after MPI_Finalize is called.
tested with h5committest
|
| |
|
|
|
|
|
|
|
|
|
| |
- Add a new attribute function characterstic for format:
* H5_ATTR_FORMAT(X,Y,Z) __attribute__((format(X, Y, Z)))
- Rename UNUSED attribute characterstic to H5_ATTR_UNUSED.
- Rename NORETURN attribute characterstic to H5_ATTR_NORETURN
tested with h5committest.
|
| |
|
| |
|
|
|
|
| |
move undef of __attribute__ back to private and remove its usage from public routines.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Bring r26213 from trunk to 1.8 branch:
Minor code cleanups, to align with revise_chunks branch.
Tested on:
Mac OSX/64 10.10.2 (amazon) w/serial & parallel
(h5committested on trunk for 2+ weeks)
|
|
|
|
|
|
|
|
|
| |
Renamed H5_ASSIGN_OVERFLOW() to H5_CHECKED_ASSIGN() and re-ordered
the arguments to be in a more logical order.
Fixes HDFFV-9322
Tested on: h5committest
|
|
|
|
|
|
|
|
|
| |
Fixed a typo in test/th5o.c that caused the wrong property list
ID to be checked.
Fixes HDFFV-730
Tested on: h5committest
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Bring r26195 from trunk to 1.8 branch (minus public API changes):
Switch from enum H5D_CHUNK_BTREE to H5D_CHUNK_IDX_BTREE, along with a couple
of minor whitespace touchups.
Tested on:
MacOSX/64 10.10.3 (amazon) w/serial & parallel
(tested on trunk for 2+ weeks)
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bring r26193 from trunk to 1.8 branch:
Fix locking error when splitting root node of v1 B-tree, and simplify
the iteration over entries in a v1 B-tree (avoiding using the sibling pointer
also).
Tested on:
MacOSX/64 10.10.3 (amazon) w/serial & parallel
(tested on trunk for 2+ weeks)
|
|
|
|
|
|
|
|
|
|
|
| |
Bring r26191 from trunk to 1.8 branch:
Track the min & max keys for a v2 B-tree, so it can more efficiently
determine if a key is present in the B-tree.
Tested on:
MacOSX/64 10.10.3 (amazon) w/serial & parallel
(h5committested on trunk)
|
|
|
|
| |
in the 1.8 branch with the HDF5 1.8.15 release information.
|
| |
|
|
|
|
| |
Tested on: h5committest
|
|
|
|
| |
Do not pass warnings options to non-gnu compilers when compiling with CMake
|
|
|
|
|
|
| |
Fix bad comment in H5detect, this was preventing PGCC from compiling correctly.
Tested on: bluewaters
|
| |
|
|
|
|
| |
Fix mismatching prototypes in H5FD/H5B2
|
|
|
|
| |
Includes r26848/r26868/r26907
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Removes H5F_ACC_DEBUG and H5FD_DEBUG functionality.
The H5F_ACC_DEBUG symbol remains but has been defined to zero.
Fixes: HDFFV-1074
Tested on:
h5committest
32-bit Linux w/ C++ and Fortran and multi VFD
|
| |
|
|
|
|
|
|
|
|
|
| |
lines
The error is that $EXAMPLETOPDIR is used without setting a value first.
Solution is to set them with the proper values.
Tested: jam only to confirm the error no longer exists.
|
|
|
|
| |
Move TOOLS option up before usage.
|
| |
|
| |
|
|
|
|
| |
Tested: linux (jam), Mac (osx1010dev)
|
|
|
|
|
|
|
| |
Added method to install the Mac OSX DMG file by line commands.
Group all install method into a function call for easier maintenance.
Tested: Linux (jam), Mac (osx1010dev)
|
|
|
|
|
|
| |
Retired IBM SP system which was old.
Removed --enable-parallel=mpich information. The option is invalid now (remove
date unknown.)
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Bugs:
The configure summary (libhdf5.settings) was missing from the configure output.
Solution:
Append the configure summary to the configure logfile if configure succeeds.
Tested:
hand ran cmakehdf5 in jam and inspected the logfile.
|
| |
|
|
|
|
|
|
| |
Fix internal H5AC dxpl usage in the library.
tested with h5committest.
|
|
|
|
|
|
|
| |
--enable-parallel has
to be used explicitly to build parallel HDF5 regardless of the compiler type
being used. (MSC - 2015/02/19 HDFFV-9068)"
|
|
|
|
| |
Tested: local linux
|
|
|
|
| |
Tested: trival
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
o Removed 4.1.2. Compressed tar archive (*.tar.Z) since we don't release them any more.
o Add "some versions of tar can auto-uncompress".
o Removed reference to INSTALL_MinGW which was removed in couple releases ago.
o Removed "2.2. Red Storm (Cray XT3)" since its entry in INSTALL_parallel has been removed.
o Added "3.1. Make" to list Gnu like make is required.
o Updated the removal of the feature of auto-configure --enable-parallel when MPI compilers are used.
o Updated "4.3.3. Configuring for 64 or 32 bit support" to show how to configure both 32 or 64-bit binary.
|