Fix pnetcdf issues in recent iHAMOCC updates about blom HOT 5 CLOSED

Hi Tomas, since the PNETCDF issue on betzy NorESMhub/cime#34 is older than the commit with the explicit use statements: can you confirm that the problem was present before already (on betzy)? If so, potentially a two step approach would help: first going to fix issues connected to the explicit use statements (I assume some use statements could be missing) for the machine, Dirk is working on (maybe he could share the run-recipe there), and afterward we could head to betzy.

from blom.

Hi Jöran, my understanding when talking with Alok was that the PNETCDF issue would not be solved on betzy (or that no one was working to solve it) - @monsieuralok please correct me if I'm wrong. So we don't have an option to test for this bug on betzy.

I think the pnetcdf option should still work on Fram. Just now I was trying to do a meson build of BLOM/iHAMOCC on my own PC, but there seems to be too many things missing to test pnetcdf, so I was just about to start on Fram.

I can start by pulling a version before the explicit use statements were introduced and compare with one afterwards. This should give us a indication. Will post on this issue once I have some progress.

from blom.

Ok, I see. After re-reading the compilation error of Dirk and looking deeper into the code, we miss to import i0,ii,jj,j0,jj (and potentially some others) from mod_xc in read_netcdf_var.F90. I need to check the other files, where PNETCDF is used, though. Shall we branch from the milestone branch to fix this?

from blom.

master and release-1.3 currently point to the same commit (98ade59), so it doesn't matter which one to branch from. It would be good to make a separate bugfix branch from 98ade59 and merge into master and release-1.3 respectivesly, when we have fixed the bug.

from blom.

By the way, I have reproduced the bug on Fram. The machine settings on Fram use pnetcdf by default.

from blom.