Benjamin Uekermann

I also like this one: ```xml ``` Making > Adapter should generally support data initialization (i.e. it has to call `requires_initial_data()`) a quality criterion is IMO independent of the question...

This nicely fits the future plans to standardize adapters. Very similar idea (but probably worth a separate issue one day): We could do the same for our logging.

I think being very explicit here is good. I am in favor of ```cpp void startProfilingSection(string_view eventName); void stopLastProfilingSection(); void addProfilingSectionMetadata(::precice::string_view key, int value); ``` Actually, the metadata, we could...

If I remember the situation correctly, the problem was the solver with smaller timestep sizes changes during the simulation. First, it is participant A, afterwards participant B. And we don't...

I am not sure if I get your point. What do you mean exactly by "fluid or structure cannot converge"? Let's say the fluid solver cannot converge. Then, it would...

I think I understand better what you need now. > I wonder if the latest version of precice can do this? Not the full feature set that you ask for....

> Actually, I am using parallel coupling with first participant method. precice seems do not prohibit this coupling method. We are missing an error message here. This should be prohibited....

Is it clear for the user that with `sync-mode="on"`, they could create their own deadlock? In general, I am a bit afraid that we are opening another box for v3.0...

IMO, this is useful, but I would postpone it to 3.x.

@IshaanDesai Would you then see an urgency to get this into v3.2.0? We are still missing C and Fortran bindings, tests, and docs.