Update code

Minor changes:

• !10 (closed)
• !11 (closed)
• !12 (closed)

Bug fix:

• #34 (closed)
• Incorrect handling of caching dictionaries

plus a more extensive set of changes to solve #2 (closed).

Handle PETSc errors with two macros:

• AssertPETSc is meant to be called within OpenFOAM methods that calls PETSc
• PetscCall, instead is intended to be called within callbacks used by PETSc

requested review from @mark

assigned to @szampini

added 1 commit

• 625f7763 - Fix update never case for matrix caching

Compare with previous version

@mark Can you take a look? This improves the performance a little. It also allows one to catch errors and gracefully exit.

I wish OF had an automatic formatter. Feel free to edit for style

mentioned in issue committees/hpc#3

added 1 commit

• 1cdb8be4 - Fix #34 (closed)

Compare with previous version

changed the description

mentioned in issue #34 (closed)

@mark ping

Got back from holidays on Monday - working through various backlogs.

added 1 commit

• 2be8fe19 - BUG: dictionary for caching were recreated each time!

Compare with previous version

Hi @szampini - which petsc version is preferred for builds and the upcoming (OpenFOAM) release?

Hi @szampini - I've gone through the changes (light formatting changes). Anything else to go in before merging?

approved this merge request

assigned to @mark and unassigned @szampini

resolved all threads

changed the description

added 7 commits

• 04916685 - ENH: handle PETSc errors
• b318eba7 - Support newer versions of PETSc having PetscCount as argument
• 8784b887 - ENH: avoid recreating vectors each solve call
• 85bf3f16 - Print PetscInitialize timings and insert OpenFOAM options
• 6ac32df3 - Fix update never case for matrix caching
• a883fa20 - FIX: cuda error with hypre (fixes #34 (closed))
• 84b62fee - BUG: dictionary for caching were recreated each time!

Compare with previous version

resolved all threads

merged

mentioned in commit 5841a4a0

mentioned in issue #2 (closed)

mentioned in issue #31 (closed)