Pull in changes from feature/sdrabenh/r21c_G2G

[sdrabenh]

No description provided.

[gmao-rreichle]

@sdrabenh, I'm afraid there's an issue with CatchCN in your branch. After a very quick look, here's what I think happened. I might be wrong, but what I'm seeing is concerning enough to throw a flag.
In a nutshell, #637 was on a rather old branch of the GCM GC that still had only one GC F90 file for CatchCNCLM40 and CatchCNCLM45. At some time between then and now, CatchCN was restructured, and lot of the code in the CatchCN GC file moved into subdirs and GC F90 files specific to CatchCNCLM40 and CatchCNCLM45. The problem is that when you went from your branch feature/rreichle/r21c_updates (associated with #637) to feature/sdrabenh/r21c_G2G (associated with the present PR), the CN-related changes from #637 got lost.
We'll need to take a closer look and, if I'm not mistaken, figure out the best way to resurrect the changes of #637 that fell through the cracks. For now, I'm holding off on hitting "approve" until I'm more confident of what I'm seeing and how to address the problem.
cc: @biljanaorescanin @gmao-jkolassa

[sdrabenh]

@sdrabenh, I'm afraid there's an issue with CatchCN in your branch. After a very quick look, here's what I think happened. I might be wrong, but what I'm seeing is concerning enough to throw a flag. In a nutshell, #637 was on a rather old branch of the GCM GC that still had only one GC F90 file for CatchCNCLM40 and CatchCNCLM45. At some time between then and now, CatchCN was restructured, and lot of the code in the CatchCN GC file moved into subdirs and GC F90 files specific to CatchCNCLM40 and CatchCNCLM45. The problem is that when you went from your branch feature/rreichle/r21c_updates (associated with #637) to feature/sdrabenh/r21c_G2G (associated with the present PR), the CN-related changes from #637 got lost. We'll need to take a closer look and, if I'm not mistaken, figure out the best way to resurrect the changes of #637 that fell through the cracks. For now, I'm holding off on hitting "approve" until I'm more confident of what I'm seeing and how to address the problem. cc: @biljanaorescanin @gmao-jkolassa

I believe that is exactly what happened. The joys of applying commits after refactoring code! I'm afraid ensuring all the changes in #637 make it into this branch will need to be done by hand. I think all automated tools will fail at this.

[gmao-rreichle]

I believe that is exactly what happened. The joys of applying commits after refactoring code! I'm afraid ensuring all the changes in #637 make it into this branch will need to be done by hand. I think all automated tools will fail at this.

Thanks, @sdrabenh, for confirming my suspicion. Is the target branch release/v1 meant for R21C? If so, we could conceivably forget about the CatchCN updates because presumably R21C won't need CatchCN. The real question is how the R21C History updates will find their way into the develop branch. That's where we need to make sure we got everything, incl the missing CatchCN updates. Are there plans to capture the R21C History changes in develop?

cc: @biljanaorescanin @gmao-jkolassa

[mathomp4]

Thanks, @sdrabenh, for confirming my suspicion. Is the target branch release/v1 meant for R21C? If so, we could conceivably forget about the CatchCN updates because presumably R21C won't need CatchCN. The real question is how the R21C History updates will find their way into the develop branch. That's where we need to make sure we got everything, incl the missing CatchCN updates. Are there plans to capture the R21C History changes in develop?

cc: @biljanaorescanin @gmao-jkolassa

@gmao-rreichle I can partially answer this. release/v1 is a branch created from the last v1 tag of GEOSgcm_GridComp (i.e., before Bill's new physics). So, if there are any needed patches for a GEOS model using v1.17.4, they would go here (and if still applying to current develop on a new v2 tag).

Since R21C is not moving to v2 physics, changes go there.

[sdrabenh]

@gmao-rreichle My latest commit attempts to incorporate all the LONG_NAME changes into the CatchCN components so the changes do not get left behind, even if they are not used in R21C. Please review if these look ok now

[sanAkel]

We should really have an automatic code generator for such changes.

[gmao-rreichle]

@sdrabenh: Many thanks again for fixing up the CatchCN40 and CatchCN45 GridComp files. I added one commit 0179922 that updates the runoff longname, which had somehow fallen through the cracks. Looks good to me know.