1. petsc
  2. PETSc
  3. buildsystem-3.3

Commits

Show all
Author Commit Message Date Builds
Satish Balay
fix f2003 test to be stricter [fix issue with opensolaris f2003]
Satish Balay
add Fortran 2003 check [if so - set flag PETSC_USING_F2003]
Satish Balay
#118137 checkCompression() should also look for zlib.lib on window Also hdf5.py should use libraries.compression for extralib
Satish Balay
#117582 complex c/zdot blas calls are handrolled in petsc - so COMPLEX_DOT_RESULT_ARGUMENT is no longer needed
Matt Knepley
Fix problem import
Jed Brown
Try to fix clang++ to use -x c++, but disable because of Clang bug
Jed Brown
Add --with-visibility
Satish Balay
error out if --with-cc etc is specified with --with-mpi-dir and mpi compiler wrappers exist.
BarryFSmith
don't check for extended_float since we don't use and fails on building openmpi
Satish Balay
fix typo
Satish Balay
#116216 give an error if PETSC_DIR/externalpackages is used for --with-package-dir
Satish Balay
#115668 add needsCompression=1 to hdf5.py [and more verbose related error messages]
Satish Balay
1. Do not weed out -L options from cxxlib. For one we use ${CXXLIB} ${CLIB}, so -L should be in CXXLIB [and not necessarily in CLIB]. And secondly we remove duplciate -L with self.libraries.toStringNoDupes() at the end of configure anyway. 2. However do try to remove duplicate -l from FLIB [similar to CXXLIB]
Satish Balay
#115674 self.setCompilers.LIBS was set to clibs only when FC was enabled. [This should also be set for FC=0 CXX=yes]
Jed Brown
Check for MPI_Init_thread
Jed Brown
Look for CUSP headers in $cuspdir/include
Satish Balay
remove Intel PEPCF90 check. It causes grief with cmake - and is no longer present with intel-11,12 compilers anyway.
Satish Balay
#114794 -lpepcf90 check creates a wrong define - and adds extra space, as in: Defined "HAVE_LIB _LPEPCF90" to "1"
Satish Balay
increase make_np to ncpu/3. Also stash this value in self.
Satish Balay
#114992 fix for archlinux where umfpack can come from either 'umfpack' or 'suitesparse'
Matt Knepley
Added libz check, and package support
Satish Balay
#113240 fix for CFLAGS='' - revert earlier fix e9bf3e78320e and initialize CFLAGS etc to 'Null' [instead of '']
Matt Knepley
Added getFullDefineName()
Satish Balay
update to latest stable openmpi
Satish Balay
upgrade to latest hdf5 tarball
Satish Balay
revert a913e4d88b2a d6ed175b07bc 23628f1ee9b3 [inline patch for hdf5]
Matt Knepley
Fix for complex dot check
Satish Balay
add -known-64-bit-blas-indices option. Also check for blas in matlab location aswell [for this option]
Satish Balay
default to using -g when using cuda [instead of -g3] as "nvcc-4.1 -g3 -arch=sm_20" breaks.
Jed Brown
Add HAVE_MPI_REPLACE define
  1. Prev
  2. Next