Benjamin Uekermann

> @uekerman could a participant write data to a mesh solely for the sake of providing a custom convergence measure or primary data for acceleration? Yes, or for a watchpoint,...

Can give ugly problems at destruction when one participant uses a `master` and the other one not and trace logging for `com` is switched on. ``` #11 0x7f0e97587eb7 in _ZN7precice7logging6Logger5traceENS0_11LogLocationERKNSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE...

1. is not very wrong, but we could add the participant name. Only makes a difference if there are multiple m2ns. 2. is indeed a bug.

You mean that you use two different `precice-config.xml`s for both coupled solvers? That is not allowed and could indeed be a good idea to implement a check for this.

> The parallel/TestQNXemptyPartition tests are still (mostly) with only primary data. True :+1: but we could still easily increase test coverage with the boolean switch explained above.

Sounds good to me :+1: Without having looked at all details yet, the following plan could make sense: 1. Increase test coverage of the existing IQN-ILS secondary data (and potentially...

Closed by #2006 and #2010

> I will first check for tutorials reproducing this. Not strictly necessary in my opinion. Reproducing with an integration test should be easier.

Hi! We currently have a student project running that should extend the adapter to (mechanical) FSI, but if you already have some previous work this, for sure, could help. We...

> In theory it should be doable to do some multiprocessing on the solve loop, since the microsimulations are idenpendent of each other in the timestep itself. I think, there...