Carefully removed \t from src/mat/. All sources are now free from tabs (exceptions: LaTeX docs, ftn-auto/ folders)

Hg-commit: 41ce18e98b1c9ae2af455e6fa76b9abd371521ee

removed silly abreviations

Hg-commit: 0641f9ef776cf11cc3336b494981da756bf9d14f

removed the silly #if defined(PETSC_USE_COMPLEX) special cases when the complex version gives the exact same results as the real version for real, i.e. PetscConj, PetscRealPart, do nothing and are ha

removed the silly #if defined(PETSC_USE_COMPLEX) special cases when the complex version gives the exact same results as the real version for real, i.e. PetscConj, PetscRealPart, do nothing and are harmless for the real case
the less #if def code duplication we have the better

Hg-commit: 158afdac623e9fb04c1dbe14153c0895e6c4feef

show more ...

Clang MPI type checking now works for user-defined types

Hg-commit: a7385c2b2cce59d13c376cee7ce8117a20c668ac

Add MPIU_BOOL and MPIU_ENUM to control clang-3.2 warnings when using MPI_INT directly

Hg-commit: 659ebb827654e17cd01d6fb5cf1a54155f4e77a7

removed trailing spaces from all files

Hg-commit: 1cfa55059afe46c5145e0e830cd797b97216abfb

removed all incorrect spaces before CHKERR...

Hg-commit: 653d9985c7cfe8fed21a37081a5b64f5e24e6c17

ran clang --analyze over PETSc and fixed many of the complaints. Mostly missing CHKERRQ() and set but not used variables, no bugs found, lots of complaints about null references, not sure what is up

ran clang --analyze over PETSc and fixed many of the complaints. Mostly missing CHKERRQ() and set but not used variables, no bugs found, lots of complaints about null references, not sure what is up with that; gets in the way of real useful messages.

Hg-commit: c183c49f249a22d7ef32be2c51749bad41445e53

show more ...

merge from 33

Hg-commit: 3a826b3e296df68ee2aa8682fc345424bf183de6

Bug fix: Use MPIULong_Send/Recv in MatLoad_MPIBAIJ.

Hg-commit: f7068d3f22659cd4edebaaec74cd24979836d12e

moved options checked in Vec/MatAssemblyEnd() into Vec/MatSetFromOptions()
eventually we need to do the same for KSPSolve,SNESSolve,TSSolve and anyother options checking in the main computational str

moved options checked in Vec/MatAssemblyEnd() into Vec/MatSetFromOptions()
eventually we need to do the same for KSPSolve,SNESSolve,TSSolve and anyother options checking in the main computational streams

Hg-commit: 76fd8399373a0d6da8c686fbc4c3e4d0b8d88a97

show more ...

XXXiew_Private() -> XXXViewFromOptions()

Hg-commit: 956ff26f22f2a1ac0320a5a195731edd8cf3234a

updated MatView_Private() and -mat_view to new paradigm for accessing -xxx_view options via PetscOptionsGetViewer()

Hg-commit: 48311e8a6a065367ee181abfeb345fa704b2d6ee

Backout 1bdcc2952450, diagonals must be removed from MPIAdj for ParMetis

Hg-commit: c1b2d1fd6943f8f9ad643a6fd3212168e6dd7d11

Fix MatPermute_MPIBAIJ, including support for non-square matrices, with tests

This patch includes contributions from Rémi Lacroix <remi.lacroix@lri.fr>.

The BAIJ implementation is still not scalabl

Fix MatPermute_MPIBAIJ, including support for non-square matrices, with tests

This patch includes contributions from Rémi Lacroix <remi.lacroix@lri.fr>.

The BAIJ implementation is still not scalable. It could be made scalable
using the approach in MatPermute_MPIAIJ.

Hg-commit: 233a0b40516716b61a3ac63fe8963cde6416dd42

show more ...

Fix MatGetSubMatrix_MPIBAIJ_Private to use the compressed index sets [petsc-maint #141575]

Recovers correct behavior for MatPermute

Hg-commit: 5145186cfb2264f95d2e31937ff79b0a026f043d

MatMPIXXXSetPreallocation() manual pages should list , PetscSplitOwnership()

Hg-commit: a62438cd7cec9d4eb5fc793b2de6c47367e5545a

reemove OBJSC and OBJSF from invididual makefiles and auto-define in conf/variables

Hg-commit: 2f01c17f3c19dc4b89aaf306cfd60a12312c270d

Const-correctness for MatGetRowIJ()/MatGetColumnIJ()

The arrays have always been read-only according to the
documentation. This patch should cause no behavioral change.

This const-correctness casca

Const-correctness for MatGetRowIJ()/MatGetColumnIJ()

The arrays have always been read-only according to the
documentation. This patch should cause no behavioral change.

This const-correctness cascaded into the private ordering and coloring
routines.

Hg-commit: 7efa5f739222584a785675705479b4e17c346e6e

show more ...

const-correctness for MatMPIAIJGetSeqAIJ

Hg-commit: 599655901578e54a05acc8cb5c8afb8688387ec8