Wolfgang Bangerth

Avoid use of reserved names.

2

The page at https://github.com/LLNL/sundials/blob/develop/doc/shared/sundials/SUNContext.rst describes the `Context` structure as declared as follows: ``` struct _SUNContext *SUNContext ``` This is unfortunate: Both C and C++ reserve all names that (i) start...

Internal error in SolutionTransfer

1

This is a shrunk version of a test case provided by M. Baensch on the mailing list a couple of days ago: https://groups.google.com/g/dealii/c/us2AK80OjUU/ ``` #include #include #include #include #include #include...

Use free consensus algorithms functions instead of the class version.

1

Not for 9.4. One of the remaining users of the class-based consensus algorithms functionality is in `mpi_compute_index_owner_internal.(h|cc)`. There, the `DictionaryPayload` class is declared and implemented to wrap a bunch of...

compute_local_to_global_vertex_index_map does not wait for communication to finish

1

This function starts communication via `MPI_Isend` and puts the requests into the `first_requests` object, but then never waits for these sends to complete. This is a potential data race.

There are a number of places where the HDF5 writers will fall on their faces if a process has no cells. One is here: ``` template void DataOutInterface::write_xdmf_file( const std::vector...

Unfinished tutorial programs

5

Let's have a list of tutorials programs with a record for when they were merged, as well as those that are around but not finished yet. **Merged in time for...

Once #13212 is merged, I will do one more change: Currently, the `CellData` structure stores a `receiver` element that is an MPI rank. But this is conceptually not part of...

Inordinately costly setup of linear operators with Trilinos

10

@tamiko @jppelteret: FYI My student @juleoc02 has found that it takes an inordinately long time to set up linear operators when using Trilinos payloads. Here is a test code, of...

Fix wrong default mapping for non-quad/hex meshes

4

#11539 changes a whole bunch of places that we know are only correct for quad/hex meshes. #11520 also leaves a number of TODOs. We need to figure out a way...

Issue with sign_change in FEPolyTensor

11

A student in my department (hi @GrahamBenHarper!) is working on some modifications in `fe_poly_tensor.cc` and asked a question that led me down a rabbit hole at the end of which...