Closed cponder closed 1 year ago
This occurred due to ccs_configs PR #3 as part of the 'cesm2_3_beta11' update in EarthWorks PR #12.
When we started EarthWorksOrg/ccs_configs_cesm, it was based on tag 'ccs_config_cesm0.0.38' from ESMCI/ccs_configs_cesm which had no support for NCAR's test machine Gust. ccs_configs PR #1 tried to address this by merging a branch that contained the current porting effort to Gust by Jim Edwards as of Feb 2023. When EarthWorksOrg/ccs_configs_cesm was updated, it advanced from being based on tag 'ccs_config_cesm0.0.38' to tag 'ccs_config_cesm0.0.49' which was made in Nov 2022. This caused the files to "time-travel" and is what caused the apparent reversion from paths that used sensible environment variables to hard-coded paths.
Proposed fix: You can try merging the same branch as was done in EarthWorksOrg/ccs_configs_cesm PR #1 as a one-off fix for Gust. You can find that here (or maybe here in case the branch name disappears).
If this were another system (i.e. not a proxy/test system), I'd be more motivated to fix it myself. We will eventually get these changes since they are in the 'cesm2_3_beta13' externals. That would require a 'cesm2_3_beta12' update first. I was made aware of a science results change in the atmosphere of 'cesm2_3_beta13' based runs, so I'm not in a hurry to catch up to that version until I learn more about that.
I think this has been resolved with PR #14, so I'm closing this. If that's not the case, someone please comment or re-open this.
I cloned using this command:
Running the
./case.setup
gave this message:It looks like the file
ccs_config/machines/config_machines.xml
is hard-coding the pathsinstead of parameterizing them like in the earlier versions: