Merge branch 'jed/is-color-value-type-f'* jed/is-color-value-type-f: config: define PETSC_IS_COLOR_VALUE_TYPE_F instead of by size (fix NAG compile)
config: define PETSC_IS_COLOR_VALUE_TYPE_F instead of by size (fix NAG compile)This eliminates the preprocessor logic of convertingPETSC_IS_COLOR_VALUE_TYPE_SIZE to a Fortran type and leaves no "e
config: define PETSC_IS_COLOR_VALUE_TYPE_F instead of by size (fix NAG compile)This eliminates the preprocessor logic of convertingPETSC_IS_COLOR_VALUE_TYPE_SIZE to a Fortran type and leaves no "else"clause where we previously tried to use #error from Fortran; thatdirective is not supported by the NAG compiler (even inside an #if 0).Reported-by: William F Mitchell <william.mitchell@nist.gov>
show more ...
Configure: Remove spurious dependence of scalarTypes on MPI
configure: default to static libraries if sharedlib test fails. However error out if user specifies --with-shared-libraries=1
Configure should error withif self.precision == '__float128' and self.scalartype == 'complex' and self.languages.clanguage == 'Cxx':because std complex doesn't support __float128Reported-by: Lisa
Configure should error withif self.precision == '__float128' and self.scalartype == 'complex' and self.languages.clanguage == 'Cxx':because std complex doesn't support __float128Reported-by: Lisandro Dalcin <dalcinl@gmail.com>
Added os.path.expanduser() for uses of ~ in --prefixThanks-to: Alicia Marie Klinvex <amklinv@sandia.gov>
Stop configure if user builds with threadsafety and debugging onReported-by: E. Tadeu <e.tadeu@gmail.com>
Configure: The --prefix argument should always be converted to an absolute path
Merge branch 'knepley/fix-configure-HOME'* knepley/fix-configure-HOME: Configure: Give proper default if HOME is missing from the environment Configure: I think RDict should use threads by defa
Merge branch 'knepley/fix-configure-HOME'* knepley/fix-configure-HOME: Configure: Give proper default if HOME is missing from the environment Configure: I think RDict should use threads by default, even if the script does not
Configure: Give proper default if HOME is missing from the environment
Merge branch 'maint'
DATAFILESPATH: always check in $HOME
do package tests (e.g. superlu_dist) that require DATAFILESPATH only when data files are available and PETSc is built with 32 bit integer doubleFortran example did not free created communicator
win32: _isnan() requires float.h - not math.h
makefiles: use FC_FLAGS, FFLAGS, FPPFLAGS in order - in fortran compile targets[similar to corresponding c targets]Reported-by: Lisandro Dalcin <dalcinl@gmail.com>
Configure: When calling an embedded module, you must save and restore the log
Configure: Do not log directly to the framework
Configure: Changed is*() static functions to take a log argument
Configure: executeShellCommand() should always take the log argument
updated PETSc directory layout to match standard packaging strategiesinclude/petsc finclude,private,mpiunilib/petsc confbin/petsc*
Fix configure logging of integer size.If we don't log the integer size, someone could, oh, I don't know, spendhalf a week not realizing that they misspelled "--with-64-bit-indices"as "--with-64bi
Fix configure logging of integer size.If we don't log the integer size, someone could, oh, I don't know, spendhalf a week not realizing that they misspelled "--with-64-bit-indices"as "--with-64bit-indices", chasing down phantom bugs that only appear onlarge problem sizes.
Merge branch 'knepley/feature-parallel-configure'* knepley/feature-parallel-configure: Revert bad ML fix in ba4f728f6c00f645cbd73e74de34ed390600afde Configure: Fix ML for 64-bit integers Conf
Merge branch 'knepley/feature-parallel-configure'* knepley/feature-parallel-configure: Revert bad ML fix in ba4f728f6c00f645cbd73e74de34ed390600afde Configure: Fix ML for 64-bit integers Configure: Trying to clean up use of installer, decoupled mpi4py and petsc4py from PETSc.Configure - One problem is that this changes meaning. In the pkg-config, the install dir is $PETSC_DIR instead of $PETSC_ADIR/$PETSC_ARCH Configure: Must capture output when rerunning configure on a required object Configure+MPI: Remove redundant function Configure: Framework should print out regular errors as well Configure: Replace self.log.write with self.logWrite Configure: Check for Queue support - Python 2.4 should fail this Configure: If --useThreads=0, use serial configuration Configure: Improve error logging in parallel Configure: Can set the number of threads using --useThreads=<num> Configure: functions module should depend on libraries Configure: Objects should not use log or argDB from framework Configure: Enable concurrent execution - Should fallback to serial evaluation, but we might want a flag for this Configure: Removed redundant code Configure: Removing dependence cycles - Cannot depend on oneself (a depressing realization) - Make cannot depend on MPI - MPI implementations cannot depend on MPI - debugging cannot depend on setCompilers - PETSc cannot depend on mpi4py and petsc4py Configure: Allow thread local loggingConflicts: config/BuildSystem/config/packages/SuperLU_DIST.py config/BuildSystem/config/packages/sowing.py
Configure: Trying to clean up use of installer, decoupled mpi4py and petsc4py from PETSc.Configure- One problem is that this changes meaning. In the pkg-config, the install dir is $PETSC_DIR instead
Configure: Trying to clean up use of installer, decoupled mpi4py and petsc4py from PETSc.Configure- One problem is that this changes meaning. In the pkg-config, the install dir is $PETSC_DIR instead of $PETSC_ADIR/$PETSC_ARCH
1234567