Merge branch 'jrwrigh/vecclosureatdepth' into 'main'

feat(plex): Move DMPlexVecGetClosureAtDepth to public

See merge request petsc/petsc!8824

Add the logging of gpu energy

- Remove unneeded PETSC_HAVE_DEVICE macro
- -log_view_gpu_energy requires CUDA version >= 12.2
- Use PetscDefined instead of macro

feat(plex): Move DMPlexVecGetClosureAtDepth to public

Change the `destroy()` function argument of `TSTrajectorySetTransform()` to type `PetscCtxDestroyFn *`. These means the destroy function must deference the argument before operating on it

Change the `destroy()` function argument of `SNESSetConvergenceTest()` to type `PetscCtxDestroyFn *`

Development Tools: Vim, Emacs, Eclipse

Change names of Mat_XXX product contexts to MatProductCtx_XXX for code maintainability

Update destroy callback of all MatProductCtx and MatShellSetMatProductOperation() to use PetscCtxDestroyFn

The goal of all the commits in this MR is to unify the destruction of "application" contexts.

This will help with supporting language bindings. For example, for Fortran the application context
would

The goal of all the commits in this MR is to unify the destruction of "application" contexts.

This will help with supporting language bindings. For example, for Fortran the application context
would be a user derived type

Fix prototype of DMShellSetDestroyCtx function argument to use PETSc standard way of doing it with PetscCtxDestroyF.

Also fix PetscViewerGLVisSetFields prototype of destroy function

These two locations were missed when PetscCtxDestroyFn was introduced as the universal function prototype for context destruction in 3.23

Reported-by: Victor Eijkhout

show more ...

Merge branch 'jolivet/no-device-no-petscdevice' into 'main'

Add `--with-devicelanguage` to make PETSc a C library again

Closes #997

See merge request petsc/petsc!8783

configure: add --with-devicelanguage=(C|C++)

configure: make SYCL a devicePackage

Merge branch 'jczhang/2025-09-19/precision-mumps' into 'main'

Support different precisions with MUMPS

See merge request petsc/petsc!8715

KSPHPDDM: deprecate KSPHPDDMPrecision in favor of PetscPrecision

MUMPS: add multi-precision support

revert sources to dev
- reset PETSC_VERSION_RELEASE
- reset doc/changes/dev.md

Merge branch 'barry/2025-09-25/skip-python-exec-hardwire' into 'main'

configure: add -with-python-exec-from-env option to skip hardwiring of the python executable path

See merge request petsc/pet

Merge branch 'barry/2025-09-25/skip-python-exec-hardwire' into 'main'

configure: add -with-python-exec-from-env option to skip hardwiring of the python executable path

See merge request petsc/petsc!8733

show more ...

configure: add -with-python-exec-from-env option to skip hardwiring of the python executable path

Merge branch 'barry/2025-09-27/cleanout-pcsetpresolve' into 'main'

Remove left over parts of PCSetPreSolve() that were mistakenly left in !8419

See merge request petsc/petsc!8744

Remove left over parts of PCSetPreSolve() that were mistakenly left in !8419

Merge branch 'stefanozampini/multistage-partitioner' into 'main'

Multistage partitioner for DMPLEX

See merge request petsc/petsc!8730

PetscPartitioner: add multistage partitioner

Merge branch 'jrwrigh/isoperiodic_plex_mat_preallocate' into 'main'

fix(Plex): Fix Mat preallocation when isoperiodic is used

See merge request petsc/petsc!8712

Merge branch 'barry/2025-07-30/fortran-rm-common-block-example' into 'main'

Remove support for PETSc objects in Fortran common block

See merge request petsc/petsc!8616

Remove support for having PETSc objects in common blocks

This involves removing --with-fortran-type-initialize ./configure option

Also make name values for uninitialized objects in Fortran with FOR

Remove support for having PETSc objects in common blocks

This involves removing --with-fortran-type-initialize ./configure option

Also make name values for uninitialized objects in Fortran with FORTRAN_TYPE_INITIALIZE' and PETSC_FORTRAN_TYPE_NULL_RETURN instead of numerical values -2 and -3

show more ...

plex: Correct Mat preallocation for isoperiodic grids

Requires:
- Using isoperiodic SF instead of the (purely topological) pointSF
- Force the use of the isoperiodic SF even if in serial
- Handle th

plex: Correct Mat preallocation for isoperiodic grids

Requires:
- Using isoperiodic SF instead of the (purely topological) pointSF
- Force the use of the isoperiodic SF even if in serial
- Handle the case of an isoperiodic SF leaf-root pair a) being on the
same rank and b) both being present in the same adjancency

show more ...

section: Implement ArrayView function, use for PetscSectionVecView