Lukas Grützmacher

> Now that the API is approved, I guess you can continue, right? Yes, so far I did not have sufficient time.

Please note that I create my commits always to be merged without squashing. If you disagree with my commits please drop me a note.

> There's a tiny regression in test coverage. Not sure if it's really an issue. I know that there are lines not covered. But they are edge cases. I tried...

It would be great if these settings are not static. When executing tests in parallel they must not influence each other.

> I can't think of a way to make that possible. You? Not yet in detail. I need to investigate. > I'm not sure what you're asking for? I remember...

> Compared to using `Be`, `BeNaN` will make the test more resistant, i.e. doesn't have to be changed, if changing the subject type from `double` to `float`. Interesting question: Is...

In this case we could not check for NaN at all, isn't it?

@daveMueller , @MarcoRossignoli , please note that this fixes EXECUTING of some integration tests.

What means outdated? It was actual when I created the PR. It's a pity that it just stayed there.

Hey @xzxzxc , sorry for strong delay! Today I finally tried to reproduce. I guess, it was already fixed while refactoring in #1967. Please check again with version 6.8.0. At...