Robert Pincus

@SamuelTrahanNOAA Note that the `main` branch does not yet run in single precision either. This is the next major task but likely won't happen until August.

@charleskawczynski Thanks for the suggestion. In fact I was just putting together a list of things I'd like to include in coverage testing. This will likely be a separate repo.

@charleskawczynski This issue motivated the new regression tests in commit e68ad73f71ceace653311a2ed6b4a4402074b154. See tests/ directory. I hope to figure out a way to integrate code coverage into the continuous integration automatically.

An initial response by email: ``` It should be possible to remove C bindings from any routines that aren’t called by the Fortran front-ends. I’m skeptical, though, that using assumed-shape...

Note that the current tests leave many routines and blocks uncovered (see #62).

@Chiil Handling different-sized sets of g-points should come for free. Yes to being able to vary the number of gases used, at least between two extremes, but I'd put this...

`mo_rcemip_profiles.F90` now provides the ability to make profiles at arbitrary vertical resolution, so a program based on that could easily vary points 1, 2, and 3.

Cheil, thanks for this suggestion. As I said I'm going to think about this - it's empirically ok but it doesn't converge properly, nor is the threshold set with respect...

Thanks for the excellent contribution @peterukk. I'm a bit buried with teaching but will merge some version of this in a few weeks.

Thanks, @MennoVeerman This is a good catch.