PetscSF: Various fixes and cleanups

* Add missing PetscSFGetType()
* PetscSFDestroy_XXX() now calls PetscSFReset_XXX(), otherwise changing
the SF type with PetscSFSetType() may leak memory.
* Fix

PetscSF: Various fixes and cleanups

* Add missing PetscSFGetType()
* PetscSFDestroy_XXX() now calls PetscSFReset_XXX(), otherwise changing
the SF type with PetscSFSetType() may leak memory.
* Fix init/computation of {min|max}leaf and use maxleaf where required
rather than recomputing the value in a loop
* Add logging event for PetscSFSetUp()
* Add tons of missing error checking
* Enable missing Fortran wrappers that can be generated automatically
* Support for command line options -{info|log}_exclude
* Various doc fixes
* Add tests for SFs with empty graphs

show more ...

Merge remote-tracking branch 'origin/knepley/fix-snes-ex69' into tisaac/feature-dmfield

* origin/knepley/fix-snes-ex69: (311 commits)
PetscFECreateDefault: fix custom Fortran interface to use MPI_

Merge remote-tracking branch 'origin/knepley/fix-snes-ex69' into tisaac/feature-dmfield

* origin/knepley/fix-snes-ex69: (311 commits)
PetscFECreateDefault: fix custom Fortran interface to use MPI_Comm
SNES ex69: Updated test output
SNES ex69: Removed all mantle stuff and change fields to constants - Also use dsymutil for executables
SNES ex69: Removed special partition
SNES ex69: Fixed null space creation - Null space locks its vectors
PetscFE: PetscFECreateDefault() now takes MPI_Comm instead of DM
cleanup very ugly PCGAMG manual page
minor clarifications for VecScatterCreate() manual page
gcov: PETSC_ARCH is required onlyfor -run_gcov - and not for -merge_gcov [where it doesn't exist]
make clearer in help statements what --with-external-packages-dir is for.
Website: Added news entry for PETSc User Meeting 2018 on main page.
CUDA: Further test output fixes for mat_tests-ex2
Turn off elemental test for parmetis running since this does not work on batch systems.
3.8 Changes: KSPCHEBYSHEV
CUDA: Fixed most test errors due to MatMultTranspose*
website: update urls
website: fix Jason's name to match pic. Also add Todd's pic
Dev manual: copyedit
DMPlexDistributeLabels: Propagate the output flag
cuda/cusp: update minimum supported versions of cuda to 7.5 and cusp to 0.5.0
...

show more ...

Merge branch 'haplav/fix-PetscSFView-no-graph-set' [PR #878]

* haplav/fix-PetscSFView-no-graph-set:
PetscSFView: call PetscSFSetUp only if graph is set.

PetscSFView: call PetscSFSetUp only if graph is set

Merge branch 'barry/tao-to-new-test-harness' into barry/all-new-test-harness-branches

Merge branch 'master' into knepley/feature-plex-adjoint-prelim

* master: (251 commits)
yaml: add a backup download URL
Fix documentation of MatSetRandom function
Ignore -malloc-hbw option if m

Merge branch 'master' into knepley/feature-plex-adjoint-prelim

* master: (251 commits)
yaml: add a backup download URL
Fix documentation of MatSetRandom function
Ignore -malloc-hbw option if malloc is already set
Plex: Fixed array initialization
pgi: compile fix
knl: update test to use memkind and remove option that worked arround a bug with intel 17
Check the return value for memkind_posix_memalign
Change the default behavour for -malloc_hbw
Bib: More refs
test: remove runex13f90 as its now converted to new test suite
fix win32 crash. This test was previously run with '-n 4' - so restoring it
Fix petscsfgetgraph_ to work for all Fortran compilers by directly using array of PetscSFNodes
bugfix for MatDuplicate_MPIAIJ_PtAP Reported-by: Samuel Lanthaler <s.lanthaler@gmail.com>
strip out dead sundials usage code; change sundials test to only run 5 timesteps to prevent different number of steps on different systems.
Remove extra lines of output file
Example uses C99 constructs so must mark with requires: c99
fix const missmatch for DMDAVecRestoreArrayRead()
remove conflict between function name in example and new PETSc TS function
builds: attempt to do maint/master/next test builds in separate clones
ex62: it turned out that CHACO partioning is not deterministic. Use a prescribed partition instead
...

Conflicts:
src/dm/interface/dm.c
src/snes/examples/tutorials/ex13.c

show more ...

Merge remote-tracking branch 'origin/barry/fortran-PetscSFBcastBegin'

Merge commit 'd47bf9aaf1e5266cc3f0ff499f934c85788965a9' into jed/fix-matcreatempibaij/maint

Obtain changed (v3.5) handling of MatMPIBAIJSetPreallocationCSR to match
documentation.

example usage for PetscSFNode in C and Fortran in manual page

Commit-type: documentation
Reported-by: Adrian Croucher <a.croucher@auckland.ac.nz>

Add support for PetscSFBcastBegin/PetscSFBcastEnd() from Fortran including test example

Apparently there was no support for PetscSF previously from Fortran so this required some
additional simple co

Add support for PetscSFBcastBegin/PetscSFBcastEnd() from Fortran including test example

Apparently there was no support for PetscSF previously from Fortran so this required some
additional simple code changes

Commit-type: feature
Reported-by: Adrian Croucher <a.croucher@auckland.ac.nz>
Thanks-to: Jed Brown <jed@jedbrown.org>

show more ...

Merge branch 'master' into knepley/feature-adaptor-plex

* master: (56 commits)
test: Disable TS ex19 with single precision
PETSCVIEWERGLVIS: Support DMView_DA() out of VecView_GLVis()
test: Fi

Merge branch 'master' into knepley/feature-adaptor-plex

* master: (56 commits)
test: Disable TS ex19 with single precision
PETSCVIEWERGLVIS: Support DMView_DA() out of VecView_GLVis()
test: Fix TS ex19 time drift issue with quad precision
testing
Minor edit to 3.8 changes log for AIJMKL/BAIJMKL.
Bib: Added ref
Plex: Change initializers for C89
small update as complained by nightly tests
doc: fix missing from 0e50f9e530a7b78427514d3e384f6941d4a9cc62
small fix for nightly tests
PetscDualSpaceApplyFVM(): Fix for change in dual space semantics
Plex: All tests now passing after DMPlexCreateBoxMesh() update
cleanup MatPtAP_MPIAIJ_MPIAIJ; rm MatPtAPXXX_MPIAIJ_MPIAIJ_ptap()
update MatView_MUMPS()
PFLOW: Bug-fix in parser.
PFLOW: Bug-fix in parser.
Plex: Small fix for box meshing
Plex: Unified box mesh creation to DMPlexCreateBoxMesh() - Makes simplices and tensor cells - Does not work for all combinations, but gives correct error
Plex: Added bounds and faces in each dim to DMPlexCreateBoxMesh()
Plex: Added domain bounds to HexBoxMesh()
...

show more ...

Merge branch 'karlrupp/fix-compiler-warnings-oct2017'

* karlrupp/fix-compiler-warnings-oct2017:
Fixes compiler warnings detected by SLEPc buildbot.

Fixed compiler warnings when using MPIUNI, optimized build.

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

Merge branch 'jed/sf-shared-uni'

* jed/sf-shared-uni:
SF: protect call to MPI_Group_translate_ranks for an empty group

SF: protect call to MPI_Group_translate_ranks for an empty group

This is a workaround for MSMPI which fails when passed a NULL array and
no ranks to translate.

Thanks-to: Satish Balay <balay@mcs.an

SF: protect call to MPI_Group_translate_ranks for an empty group

This is a workaround for MSMPI which fails when passed a NULL array and
no ranks to translate.

Thanks-to: Satish Balay <balay@mcs.anl.gov>

show more ...

Merge branch 'jed/sf-shared-uni' [PR #736]

* jed/sf-shared-uni:
MPIUNI Fortran: fix predefined constants
MPIUNI: define MPI_OP_NULL so that example can compile
DMPlex tests: update output for

Merge branch 'jed/sf-shared-uni' [PR #736]

* jed/sf-shared-uni:
MPIUNI Fortran: fix predefined constants
MPIUNI: define MPI_OP_NULL so that example can compile
DMPlex tests: update output for SF type being set up before viewing
Remove !mpiuni from tests
SFBasic: circumvent send-to-self by marking self as a distinguished group
PetscSFView: dynamically sort by ranks for ASCII_INFO_DETAIL output
SF: add ability for implementations to identify a distinguished group of ranks
PetscSF: add distribution/transposition test
SFBasic: pack into non-contiguous (by rank) root/leaf buffers instead of contiguous buffer
SF: PetscSFGetRanks requires a setup SF, PetscSFView calls setup
MPIUNI: change datatype packing scheme; implement MPI_Type_get_contents
MPIUNI: fix use of MPIUni_Abort, which returns
Add PetscFindMPIInt for searching within sorted array of PetscMPIInt
SF: delay construction of rank mapping until SF type is set
PetscSFReset: call impl reset before corrupting interface data
SFSetGraph: remove support for nroots=PETSC_DETERMINE

show more ...

PetscSFView: dynamically sort by ranks for ASCII_INFO_DETAIL output

SF: add ability for implementations to identify a distinguished group of ranks

This will be used to optimize send-to-self (and work around broken MPI
lacking send-to-self) and to implement fast shar

SF: add ability for implementations to identify a distinguished group of ranks

This will be used to optimize send-to-self (and work around broken MPI
lacking send-to-self) and to implement fast shared memory exchange.

show more ...

SF: PetscSFGetRanks requires a setup SF, PetscSFView calls setup

This in preparation for different PetscSF implementations requesting
different rank setup.

SF: delay construction of rank mapping until SF type is set

To implement a shared memory optimization for PetscSF without ugly hacks
that scale poorly, we need to be aware of the shared ranks when
c

SF: delay construction of rank mapping until SF type is set

To implement a shared memory optimization for PetscSF without ugly hacks
that scale poorly, we need to be aware of the shared ranks when
constructing the rank mapping. With eager construction of rank mapping,
this means discarding that or doing fairly complicated and
memory-intense conversion. This commit delays that construction, but
has no further behavioral change.

ex1: explicitly call PetscSFSetUp so that PetscSFView shows the full communication graph.

show more ...

PetscSFReset: call impl reset before corrupting interface data

Also remove extra call to impl resets because PetscSFDestroy always
calls PetscSFReset.

SFSetGraph: remove support for nroots=PETSC_DETERMINE

This is disgusting, unused, and undocumented. Evidently Matt wanted it
at some point, though I don't understand the circumstance in which
nroot

SFSetGraph: remove support for nroots=PETSC_DETERMINE

This is disgusting, unused, and undocumented. Evidently Matt wanted it
at some point, though I don't understand the circumstance in which
nroots is not known but the caller has enough information to use the
SF.

show more ...

Merge branch 'master' into knepley/feature-plex-sphere-mesh

* master: (2633 commits)
Bib: Added ref
SNES ex56: Update test output
add a citation to petsc.bib
Plex: Initialize variables
ini

Merge branch 'master' into knepley/feature-plex-sphere-mesh

* master: (2633 commits)
Bib: Added ref
SNES ex56: Update test output
add a citation to petsc.bib
Plex: Initialize variables
initialize ap as complained by the nightly tests
Plex ex3: Update to new point function API
initialize some variables as complained by nightly tests
SNES ex12: Mix misnaming Nf --> Nc
update tests
add tests
add MatView_SeqXAIJ_ASCII_structonly(); enable A->structure_only in MatSetValuesBlocked_SeqBAIJ
typo
fill-in some references
Bib: Fixed ref
Plex: Added PLY to file support
Plex: Adding PLY support
PetscDS: Turn off Fortran bindings for Constants
Plex tests: Updated to new point function API
SNES ex56: Moved tests to new format
DM: Allow field projection functions to take an array of scalars - This will allow us to pass in user data to kernels without void *, more efficiency than aux fields
...

Conflicts:
config/builder.py
include/petscdmplex.h
src/dm/impls/plex/examples/tests/ex1.c
src/dm/impls/plex/plexcreate.c
src/dm/impls/plex/plexgmsh.c
src/snes/examples/tutorials/ex63.c

show more ...

Merge branch 'master' into knepley/feature-plex-pragmatic-parallel

* master: (1379 commits)
Minor fixes for tchem examples due to changes in PETSc infrastructure
update webpage about API changes

Merge branch 'master' into knepley/feature-plex-pragmatic-parallel

* master: (1379 commits)
Minor fixes for tchem examples due to changes in PETSc infrastructure
update webpage about API changes
add help menu for MatCreateSubMatricesMPI()
rename struct Mat_SubMat -> Mat_SubSppt
add a little information to the manual about using tchem
rename MatCreateSubMatrixComposite -> MatCreateSubMatrixVirtual; Mat_SubMatrix -> Mat_SubVirtual
need to protect the \ passed on the makefile on my Mac
Bib: Updated ref
update src/mat/examples/tests/output/ex23_#.out
rename MatGetSubMat -> MatCreateSubMat
Mark MatCreateMPIAIJWithSplitArrays() and MatMPIAIJSetPreallocationCSR() with MAT_NO_OFF_PROC_ENTRIES
superlu: libray is installed in PREFIX/lib64 - fix this to use PREFIX/lib
backport fixes cc105cc3 2b35c90a - required for 3bf428d - that was backported from master - 80c96bb
'echo -n' fails on opensolaris/osx - so avoid using it.
fix diffs in nightlybuilds due to options in ~/.petscrc
add Kelley03 to src/docs/tex/petsc.bib
build: determine moudles, mpi.h used during buildtime and print in 'info'
fix analyzer warnings
fix build failure on opensolaris - where -lm from flibs is removed due to -lm in cxxlibs. But then - only flib gets used for some configure tests [blaslapack, parms]
config: prefer __restrict over C99 restrict keyword
...

show more ...

Merge branch 'barry/remove-funct'