Merge branch 'maint'

dsymutil: check for it only on OSX. And fix check for shared_darwin target (use check for OSX - not dsymutil check)

Reported-by: Pedro Marcelino <pedronovaesmelo@gmail.com>
Thanks-to: Jed Brown <jed

dsymutil: check for it only on OSX. And fix check for shared_darwin target (use check for OSX - not dsymutil check)

Reported-by: Pedro Marcelino <pedronovaesmelo@gmail.com>
Thanks-to: Jed Brown <jed@jedbrown.org>

show more ...

Merge branch 'maint'

cmake: run configure/bootstrap stage of cmake parallely

Merge remote-tracking branch 'origin/balay/fix-gmake-check/maint'

Merge branch 'maint'

make: configure should error out if 'gnumake' is not found [or provided by user].

mkl: PETSC_MKL_SUPPORTS_BAIJ_ZERO_BASED depends on PETSC_HAVE_MKL_SPARSE_OPTIMIZE
so move the corresponding configure check from mkl_sparse.py to mkl_sparse_optimize.py

Also add 'mkl_sparse_s_create

mkl: PETSC_MKL_SUPPORTS_BAIJ_ZERO_BASED depends on PETSC_HAVE_MKL_SPARSE_OPTIMIZE
so move the corresponding configure check from mkl_sparse.py to mkl_sparse_optimize.py

Also add 'mkl_sparse_s_create_bsr' to self.functions list - so that it is checked
before the test for PETSC_MKL_SUPPORTS_BAIJ_ZERO_BASED

Reported-by: "Schaferkotter, Michael R CTR FNMOC, N5" <michael.schaferkotte.ctr@navy.mil>

show more ...

mpiuni: remove -Iinclude/petsc/mpiuni usage and fix:
- inlude mpif.h
- make info

Merge branch 'maint'

Fortran: detect gfortran-8 and use size_t for string lengths

https://gcc.gnu.org/gcc-8/porting_to.html

configure: add support for getCompiler('FC')

Add configure.Language context manager

Cleaner and than manual pushes and pops; should eventually replace all
current uses of pushLanguage() and popLanguage().

Merge branch 'maint'

Initialize PETSc objects in Fortran to 0 (NULL) instead of 1.

This allows uses to call PetscObjectDestroy() on objects that have never been created without crashing trying to free from the address 1

Initialize PETSc objects in Fortran to 0 (NULL) instead of 1.

This allows uses to call PetscObjectDestroy() on objects that have never been created without crashing trying to free from the address 1. (Since PetscObjectDestroy() can be called on null objects).

I am not sure why I previously used 1 as the initial value instead of 0. I now see no benefit in using 1.

Commit-type: feature
Reported-by: Marco Tiberga <M.Tiberga@tudelft.nl>

show more ...

petsc4py: update to snapshot with TAOTEST change

Merge branch 'maint'

Merge remote-tracking branch 'origin/balay/update-make-l/maint' into maint

Merge remote-tracking branch 'origin/balay/update-cmake-3.11.1/maint' into maint

sundials: fix --download-openmpi to use --with-mpi-root option when building sundials via --download-sundials

Reported-by: Gernot Plank <gernot.plank@gmail.com>

cmake: cmake-3.7.0 build fails on fedora28 - when libcrypto.so is installed - so update to latest version -3.11.1

This versionbuilds fine on opensolaris - so remove cmake-3.0.1 usage

In file includ

cmake: cmake-3.7.0 build fails on fedora28 - when libcrypto.so is installed - so update to latest version -3.11.1

This versionbuilds fine on opensolaris - so remove cmake-3.0.1 usage

In file included from /home/balay/tmp/petsc/arch-linux2-c-debug/externalpackages/cmake-3.7.0/Utilities/cmlibarchive/libarchive/archive_cryptor.c:32:
/home/balay/tmp/petsc/arch-linux2-c-debug/externalpackages/cmake-3.7.0/Utilities/cmlibarchive/libarchive/archive_cryptor_private.h:107:17: error: field ‘ctx’ has incomplete type
EVP_CIPHER_CTX ctx;
^~~

show more ...

Merge branch 'maint'

make: update --download-make to use a patched tarball [from fedora] - that has compile fixes for gcc-8.1

tests: attempt to speedup builds/tests by using MAKE_LOAD [set higher than NPMAX]
- add --with-make-load configure option for user override
- increase make_test_np factor from .35 to .5

Merge remote-tracking branch 'origin/barry/bugfix-petscformatconvert'