fix memory leaks and inconsistencies in options database

Hg-commit: 3c916f369892bf9068bcef26c5b0e8b83136c7b2

have consistant INSTALL_DIR [for builds with and without --prefix]
- remove install_src and install_docs targets [they have to be rewriten in python - with uninstall capability]

Hg-commit: 83f42e4ab

have consistant INSTALL_DIR [for builds with and without --prefix]
- remove install_src and install_docs targets [they have to be rewriten in python - with uninstall capability]

Hg-commit: 83f42e4aba314157bf387ad32ef8d121d3b76223

show more ...

more fixes for dynamic libraries, review pathname handling and finalization

Hg-commit: 76e5b2dd2896a7d3e9ab86332cb60c3f57d3c928

added -ksp_monitor_range -ksp_monitor_range_draw -snes_monitor_range -snes_monitor_range_draw to get a
measure of how many of the residual elements are much larger than most of the others. It calc

added -ksp_monitor_range -ksp_monitor_range_draw -snes_monitor_range -snes_monitor_range_draw to get a
measure of how many of the residual elements are much larger than most of the others. It calculates
the percentage of elements that are more then .2*maximum element. (The .2 is a random choice).

Hg-commit: 08ed10caf9cf33c0c34a2477254c014beed4ffe0

show more ...

merged


Hg-commit: b3efa42ec45e04f6248c510bf0ed68f8651278a1

If we assume that by attaching arrays to an object the author
has more or less relinquished control of those arrays,
then it becomes up to the object to free those arrays at some point.
Hence an extr

If we assume that by attaching arrays to an object the author
has more or less relinquished control of those arrays,
then it becomes up to the object to free those arrays at some point.
Hence an extra call in HeaderDestroy, and the freeing (we are assuming
that PetscMalloc was used to create the arrays) of the various attached
arrays.


Hg-commit: 436d9cd8d9df37ad220a7c10489856b85ed4d845

show more ...

Added option -debug_terminal.

This allows using a terminal instead of xterm for the debugger. Also,
the debugger can be started as one window per process in a GNU screen
session.

Hg-commit: 2934c7

Added option -debug_terminal.

This allows using a terminal instead of xterm for the debugger. Also,
the debugger can be started as one window per process in a GNU screen
session.

Hg-commit: 2934c7e4b9d112c31381140f036d6d607d8735e6

show more ...

Added PetscOptionsTruthArray()

Hg-commit: 50979436ba13edb92307992de94fa0ff598fc2d3

bitten once too often by users who put init/solve/finalize in one routine,
then call that routine twice.


Hg-commit: ae733fbe984d31d9d52a784b1d89688c3d87e2e7

forgot to update schur_ and friend to fieldsplit_1 and friend for PCFIELDSPLIT

Hg-commit: bf99aa42addf9a96b0dc662715bd3f928b53a283

fix unused variable warnings

Hg-commit: 3bdf0ed14acbbfc495b1c6fb0a8f09550e1b30f6

added most of the appropriate PetscObjectIncrementTabLevel() calls on KSP objects

Hg-commit: 55ad623694bf2fcd3544ef7f480b24298f32bf4c

dave.mayhem23@gmail.com
Subject: [PETSC #18159] Re: Possibly add PetscOptionsPrintOptionsFile() ?

Changed format of output for PetscOptionsPrint() to print in a format that can
be r

dave.mayhem23@gmail.com
Subject: [PETSC #18159] Re: Possibly add PetscOptionsPrintOptionsFile() ?

Changed format of output for PetscOptionsPrint() to print in a format that can
be read in as an options file

Hg-commit: 30f35cfd85a83b2106d6c8a7a7adf497b4330341

show more ...

merge from release

Hg-commit: 29b1d3527c7e02e50b402d8643da29a34282c6ae

#17837 Get PETSc compiling with MSMPI

Hg-commit: a98da768eb25d5150d27850adafd2646ca877ff6

added PetscOptionsInsertString() for Fortran

Hg-commit: 25c2902ef4184e63576b53a5fee6146d0dfcf8e5

add explicit support for petsc4py in base PetscObject structure

Hg-commit: 7be03bcf3285a7c33e5258c78c03197bce0e9c2c

fix memory leak

Hg-commit: 021d62d9381aa41e8aecb9323fbcf827de8424c1

added VecAXPBYPCZ()
fixed PetscOptionsInsertFile() when optional argument has " " around it
added faq for why is MatSetValues() so slow

Hg-commit: 619014079882b990f403201eb751257f0ca294e0

simplify malloc/free in PetscOptionsInsertFile() - and fix memoryleak

Hg-commit: 38f0cc44a90e2f504a9b6ad7bfb34450f4efbbc4

commit after merge

Hg-commit: f3174971e60c15c9a5d253cd3abe0e5549cff23e

finished adding protection for MPI IO usage in PETSc with PETSC_USE_MPIIO

Hg-commit: 81a22dfeb0164b21d7c287e99b92590bae6c4e5b

doc generation fixes

Hg-commit: 52dd79f890bdfdcf9545223b3ab60a5d7db58c42

commit after merge

Hg-commit: ef52a8078177b8b44731cc9f45cd43f2b6261754

changed PetscOptionsInsertFile() to be scalable, now only process zero in the communicator opens the file
and broadcasts to the other processes. Arguments are changed.

Hg-commit: 045a41f43578938b

changed PetscOptionsInsertFile() to be scalable, now only process zero in the communicator opens the file
and broadcasts to the other processes. Arguments are changed.

Hg-commit: 045a41f43578938b9b7dcef065cf7b8c9a0a8c4f

show more ...