Christophe Prud'homme

Do we have a test for collision in sequential and parallel? It should not take too much comuting time and we should have a metric to test if computation is...

@VLCeline @lberti where do we stand on this ? We will need to keep a test or two to check fondamental properties (collision...) and move the other one in `feelpp/swimmer`

@VLCeline would it be possible to reduce the computational domain size to reduce computational time ?

We have some timeout for the tests, if they are too long, then we shut them down. We need to reduce the execution time

Do the Model classes depend on the toolboxes? Why should ithey(naive question)? If they don't Why not move the classes in feelpp or another lib outside the toolboxes?

I have some further questions : what are the consequences for feelpp (feelpp+toolboxes) in terms of dependencies ? where do you think should go the lib ? should(must?) it depend...

here are some data about header inclusion cost test | Preprocessing (s) | Compilation & Linking (s) ----|-------------------|-------------------------- feel1 | 115 | 140 feelcore | 9 | 18 feelpoly |...

header inclusion preprocessing time is given by : ``` for i in feel1 feelcore feeldiscr feelpoly feelvf; do make clean; time make $i.cpp.i; done ``` header compilation and linking time...

we should gather here evidence of the various costs at compile time and see where we should put our efforts. I have some ideas on how to do that

here is a nice blog on the subject: https://interrupt.memfault.com/blog/improving-compilation-times-c-cpp-projects#optimizing-header-includes-and-dependencies