Hendrik Ranocha

Thanks for your input, @stillyslalom! Sure, that's part of the process. The first step will be to fix everything such that we can run simulations using plain `Float32` everywhere without...

Of course, it would be a great PR to - convert methods to use generic types - check that the performance is not impacted - add tests to verify desired...

Right now, I hesitate a bit to use rational constants. Unless (something like) https://github.com/JuliaLang/julia/pull/41258 is merged, there can and will be surprising problems. For example, the case of one half...

Now that https://github.com/JuliaLang/julia/pull/41258 is merged, we might be able to use this form of providing constants in Julia v1.8 (or whenever the change is included in a new release). Edit:...

If we used the standard `derivative_matrix` for nonconservative terms in the `split_form_kernel!`, we would also need to add additional terms at interfaces, wouldn't we? Thus, the way we handle this...

> What do you mean by "we are also in control of interpreting the data (via Trixi2Vtk)"? Do we specify elsewhere how to plot from the saved data? The save...

Since this came up again in our meeting today: I added a few references to related array types to the first post. Right now, I expect we need the following...

> @ranocha also mentioned there are a few places where `cache` is passed in last, but is silently modified. This should also eventually be fixed to be consistent with "input...

It will not fix all of them since there is stuff such as https://github.com/trixi-framework/Trixi.jl/blob/6ecb47a75ae82ff15aea60851132e87895350f36/src/solvers/dgsem_tree/dg_2d.jl#L793 where we use some kind of thread cache.

> Octavian.jl - are we OK with adding this as an additional dependency? Yes - it's already an indirect dependency via [StrideArrays.jl](https://github.com/chriselrod/StrideArrays.jl/blob/master/Project.toml)