Marc Fehling

Do not compare output of `std::cout` for `tests/examples`.

7

Some tests from `tests/examples` consistently give false negative results on my machine. Upon launch of a deal.II application, I noticed that my system sometimes writes a warning to `std::cout`. Although...

Prefer relative tolerances for iterative solvers in all tutorial steps.

12

I've noticed that we are setting absolute tolerances for iterative solvers in many tutorial steps. We should use relative tolerances from the outset as a 'best practice', so that we...

FSI cylinder with flag: Newton iteration did not converge.

1

I was playing with your examples on fluid structure interactions, specifically the Turek-Hron benchmark. With the [input.json](https://github.com/exadg/exadg/blob/master/applications/fluid_structure_interaction/cylinder_with_flag/input.json) and [input_aitken.json](https://github.com/exadg/exadg/blob/master/applications/fluid_structure_interaction/cylinder_with_flag/input_aitken.json) files, the application crashes. This **only** happens when using a large...

Cleanup some github workflows.

1

- Removed a stage in `linux-debug-parallel-tpetra` that did nothing. - `linux-debug-cuda-11-clang` archived detailed.log, whereas we only print it in all other workflows. - Use separate build directory for `osx`. -...

Related to #17787. We can now finally do it this way, since https://github.com/actions/upload-artifact/issues/485 has been fixed. This will get rid of the awkward zipfile-in-a-zipfile situation.

Currently, we run Gzip to create a `*.tar.gz` file, and then Zlib creates a `*.zip` file containing the tarball through `actions/upload-artifact`. Let's skip the first step and let `actions/upload-artifact` solely...

'integer divide by zero' in `ComputeIndexOwner`

16

Running additional tests in the hp-framework surfaced a new error. I'm running a 3D Poisson scenario on a L-shaped domain, very similar to step-75. I will try if we can...

Our github workers run on ubuntu-22.04 and ubuntu-20.04. We should update them to the latest release ubuntu-24.04 soon. Some require a backport, and we should update those once we have...

Can't compile deal.II with Sundials=ON and MPI=OFF

3

This was probably introduced with the SUNComm type in SUNDIALS 7? https://github.com/LLNL/sundials/blob/main/CHANGELOG.md#suncomm-type-added ``` In file included from /usr/include/sundials/sundials_types.h:57, from /usr/include/arkode/arkode_butcher.h:21, from /usr/include/arkode/arkode.h:31, from dealii/include/deal.II/sundials/arkode.h:36, from dealii/source/sundials/arkode.cc:19: /usr/include/mpi.h:442:37: error: conflicting declaration...