Benjamin Uekermann

I would vote for ```c++ class configurationError : preciceError {}; class communicationError : preciceError {}; class numericalError : preciceError {}; // e.g. when RBF does not converge or QN has...

It would further be good if we could also catch warnings in the tests. Some of our tests throw warnings intentionally.

Would this interfere with the localized configuration we have right now? Meaning that each package has its own local configuration? In my opinion this is a very neat feature and...

> We should be able to keep the code in the individual packages. That's good. > This would even allow us to do things like sorting data and meshes by...

> The order is given by the order in the coupling scheme, correct? It's given by the order **of** the coupling scheme**s**.

Historic remark: we had the "Fastest Fortran bindings" at some point, which implemented sth like this already. At least for two instances. https://github.com/precice/precice/blob/v1.6.1/src/precice/bindings/fortran/SolverInterfaceFASTEST.cpp

There is tons of literature on this, e.g. https://arxiv.org/pdf/1801.01288.pdf

@Lianxiali Thanks for opening this issue and starting the discussion. In my opinion this could be a valid usecase.The solver tells preCICE that it reached steady-state. At least for a...

I adjusted the title because this issue is mainly a collection for what is needed to extend the data mapping to volume coupling. There are other themes as well, e.g....

> Hm, I would actually prefer the previous version of this implementation (https://github.com/precice/precice/pull/1342) @DavidSCN That one would have let to really complex decision trees, only due to a very rare...