fixed some non-standard PETSc code: source file with no source, #include without full path from PETSc root to include fileupdated iphone builderHg-commit: b815c2c0cd32f2d30547b5032e995df93748071f
64-bit fixes for ESSL, thanks Inge Gutheil [petsc-maint #125579]Hg-commit: ff52f8d5f71ebebf1ea1ffa4d33817471809a166
Fix (1) 64-bit indices and txpetscgpu and (2) Fortran GPU interfacesHg-commit: 536240654e207580e21bde8411a5a50eadca67eb
Fixed bug in MatMult_SeqAIJ when using threadcomm.Hg-commit: 2b02b76d260a23e98c2222ff32005e5321e427b4
bugfix for MatTransposeColoringCreate_SeqAIJ() as reported by [petsc-users] MatRARtHg-commit: 5559d72dc28a2ff7cfd2d6d094d62a33b06918ef
merge from 3.3Hg-commit: f1a35fbfdd53c4748d0eb4f835894e7b46682288
Started incorporating threadcomm in Mat_SeqAIJHg-commit: 7b6cd942f0e79061cffe5479787ccc32fa9d0fea
Removed old pthread code.Hg-commit: adce8f56f5e22f1cc42f23405c6d3e111fb0272c
slight changes to make sure all command line options work for cusp and cusparse matrices.Hg-commit: 1e1f03951f76fc9dbbf490d9786ae69a26a4f54f
bugfix of MatRARtSymbolic_SeqAIJ_SeqAIJ()Hg-commit: 98a30f005c74fb9b638f6d5bd6f356d5f6012c88
Refactoring how the options are set for cusp/cusparse classes.Hg-commit: 8930801acb6b12d18a9eca7a7e8a988ad381a43a
merge from 3.3Hg-commit: 8fcc41e2e7803ce76a43d9f041947cd269608feb
backout all chmod stuff in bf971ba58125[but keep diffs to src/mat/impls/aij/mpi/mpimatmatmult.c ]Hg-commit: 4a812fc1f35496ce2717c7da9ef12ffabbd13258
removed a spurious call to MatSetBlockSize.Hg-commit: bf971ba58125bd8cf7ceb2a9d0f5cdf27c27cc6a
Fixing the MatSetFromOptions code. make test passes on two architectures (different machines) with these changes)Hg-commit: d05f0662b2acd1fb65c5c0073085fef7200a4a51
Recent pulls from petsc-dev plus some small changes.Hg-commit: 0df21714398a0c8070bf50e118c22b5f2e03a9a6
Committing so that I can pull from petsc-dev.Hg-commit: e93aeb36e62fd5e6181a21a3d56e8d0b339491b5
pull of patch to MatDuplicate_SeqAIJ_Inode() from petsc 3.3Hg-commit: 530d238bbff078efb0645e1bf0b931aa9198b7b8
MatDuplicate_SeqAIJ_Inode() should reset function pointers for C output matrix not the original A input matrixHg-commit: 728c9c542cf2e9ecfda50bc25b4c704dc737b739
committing after previous merge.Hg-commit: 62ea3b3249ac52e1d599b51313301ac9e25ed491
Some local changes.Hg-commit: 554d9907a87259cc15d90b3fa0540f3e10379a06
fixed some divide by zero errors.Hg-commit: d78479ce1a4d1ad8dbd87606acc49e45de7e39d1
Make MatInvertBlockDiagonal return const because the user is not intended to modify this copyHg-commit: a86c8fe2e1845e1ac77307240b83c5ed8b8aca08
fix [uni] testexamplesHg-commit: c940a13bc08bed5b8a874f1870fb9a448eb25e8c
Pulled in changes from petsc-dev. make tests passing. Everything looks good.Hg-commit: 94eb5d50210da3db6d654576e30f2a31ec2adf25
1...<<71727374757677787980>>...168