win32: set printf output format to print 2-digit exponent [instead of the default 3-digit]

is: add ISColoringValue to fortran includes

cuda: add "compute_50", "sm_50" to known cuda-arch string list

viennacl: update to 1.5.2

elemental: also update corresponding gitcommit

download elemental with patch for valgrind warnings

Merge branch 'barry/fix-redundant-matrix'

Conflicts:
config/PETSc/utilities/Regression.py

don't create texamples_MAKE

fixed memory leak for MatGetRedundant

Reported-by: Jose Roman <jroman@dsic.upv.es>

Merge branch 'barry/elemental'

Jack's update to elemental release
force elemental tests to get listed by regression.py
fix bugs in elemental interface for reusing matrix in transpose/Hermitian

updates from Jack

Merge branch 'balay/update-sowing-g'

sowing: update to sowing-1.1.16g.tar.gz

bugfix: the current code appears to be run even when tetgen is not enabled.

configure: remove worksonWindows flag - as we don't know if users can
install a given external package by themselves
[eventhough its not supported on windows]

ctetgen: should be used with clanguage=cxx

Merge branch 'balay/fix-install-so'

install: fix rpath encoded in libpetsc.so so that does not refer to .so files in builddir
[esp when using --download-metis etc..]. This is done by changing -rpath PATH from builddir
to installdir in

install: fix rpath encoded in libpetsc.so so that does not refer to .so files in builddir
[esp when using --download-metis etc..]. This is done by changing -rpath PATH from builddir
to installdir in PETSC_EXTERNAL_LIB_BASIC

show more ...

configure: make sure setupDependencies() is invoked for openmp/pthreadclasses

Revert "metis/parmetis: update to metis-5.1.0/parmetis-4.0.3"
There were a couple of reports of breakage of mumps with this update
http://lists.mcs.anl.gov/pipermail/petsc-dev/2013-September/012853.h

Revert "metis/parmetis: update to metis-5.1.0/parmetis-4.0.3"
There were a couple of reports of breakage of mumps with this update
http://lists.mcs.anl.gov/pipermail/petsc-dev/2013-September/012853.html
Subject: [petsc-maint] Issue with Metis in PETSC-DEV?

This reverts commit 67125baed9b71bdad2cf52a801605839abb603c5.

show more ...

configure turns off use of isatty() on batch systems since it isatty() appears not to work correctly on cray and
on batch systems anyways the output always has to go to files since outpus is not inte

configure turns off use of isatty() on batch systems since it isatty() appears not to work correctly on cray and
on batch systems anyways the output always has to go to files since outpus is not interactive

Reported-by: Jed Brown <jed@jedbrown.org>

show more ...

Merge branch 'balay/cleanup-ftn-interface'

ftn: clean some fortran interfaces
- remove usage of PetscToPointer() from ftn-custom routines
- use auto-generated code for MatColoringCreate() and MatColoringDestroy()
- update to latest sowing (f)

ftn: clean some fortran interfaces
- remove usage of PetscToPointer() from ftn-custom routines
- use auto-generated code for MatColoringCreate() and MatColoringDestroy()
- update to latest sowing (f) All PetscObjects are now listed - including MatColoring

show more ...

PetscTime() now defaults to using MPI_Wtime()

- remove previous special code for rs6000
- move nt_time.c and gettimeofday() functionality to MPI_Wtime() in mpiuni
- timper.py is no longer needed