Follow-up from "MatHtool: fix stack in OpenMP region"
The following discussion from !5462 (merged) should be addressed:
-
@BarrySmith started a discussion: We could consider adding in the future PetscFunctionBeginOpenMP[User], PetscFunctionReturnOpenMP(), and PetscFunctionReturnVoidOpenMP to make this clearer and start to solidify support of OpenMP in PETSc.
In fact, please convert this thread to an Issue in my name with a time limit of two weeks when this MR is merged.