I'm using this issue to collect various possible gotchas that might arise with a configuration where one ice sheet is evolving and the other isn't, which we should look into to make sure things are working right. There are likely other things worth checking; I'm just adding things here as they pop to mind. (Update (2021-10-07) I feel like I have updated this list with the likely gotchas.)
[x] Do things work in a hybrid case where one ice sheet is evolving and the other isn't? I'm especially wondering if things work right given this logic:
With the changes I'm making in #52 , this will depend on the ice sheet-specific cism_evolve. With this logic, the evolving ice sheet will start up from a restart file, but the non-evolving ice sheet will not. If this causes problems, we could change this to consider the overall (non-icesheet-specific) cism_evolve.
Update (2021-12-26) it seems better to change this to the overall cism_evolve, so I'm doing that. I don't think this piece of code therefore needs any additional checking; I'm checking it off as done even though I haven't done any real checking of it.
[x] How do things work in nuopc, given this logic:
This depends on the system-wide cism_evolve, which is true if any ice sheet is evolving. So in the case with one evolving and one non-evolving ice sheet, run_type will stay at hybrid or branch. Will this cause any issues for the non-evolving ice sheet in nuopc? i.e., are there issues in nuopc that arise when a non-evolving icesheet tries to start up in hybrid or branch mode? (Or are there issues with the appropriate restart file not being available for that ice sheet?)
[x] (Added 2021-10-07) There is some logic in CMEPS's scripts that depends on CISM_EVOLVE: this will be true if any ice sheet is evolving; we should check the logic in CMEPS to make sure this makes sense (i.e., if whatever is done in CMEPS's scripts when the overall CISM_EVOLVE is true should indeed be done when just one ice sheet is evolving, and vice versa)
[x] (Added 2021-10-07) The cism and ctsm nuopc caps access the overall cism_evolve attribute; look at what they do based on this and make sure it's right for them to look at this attribute that is true if any ice sheet is evolving.
[x] (Added 2021-10-07) And check for other uses of cism_evolve within cism
I'm using this issue to collect various possible gotchas that might arise with a configuration where one ice sheet is evolving and the other isn't, which we should look into to make sure things are working right. There are likely other things worth checking; I'm just adding things here as they pop to mind. (Update (2021-10-07) I feel like I have updated this list with the likely gotchas.)
[x] Do things work in a hybrid case where one ice sheet is evolving and the other isn't? I'm especially wondering if things work right given this logic:
https://github.com/ESCOMP/CISM-wrapper/blob/a36227130cabc21194caf45485aa979526e63940/cime_config/buildnml#L385-L402
With the changes I'm making in #52 , this will depend on the ice sheet-specific cism_evolve. With this logic, the evolving ice sheet will start up from a restart file, but the non-evolving ice sheet will not. If this causes problems, we could change this to consider the overall (non-icesheet-specific) cism_evolve.
Update (2021-12-26) it seems better to change this to the overall cism_evolve, so I'm doing that. I don't think this piece of code therefore needs any additional checking; I'm checking it off as done even though I haven't done any real checking of it.
[x] How do things work in nuopc, given this logic:
https://github.com/ESCOMP/CISM-wrapper/blob/a36227130cabc21194caf45485aa979526e63940/cime_config/buildnml#L355-L360
This depends on the system-wide cism_evolve, which is true if any ice sheet is evolving. So in the case with one evolving and one non-evolving ice sheet, run_type will stay at hybrid or branch. Will this cause any issues for the non-evolving ice sheet in nuopc? i.e., are there issues in nuopc that arise when a non-evolving icesheet tries to start up in hybrid or branch mode? (Or are there issues with the appropriate restart file not being available for that ice sheet?)
[x] (Added 2021-10-07) There is some logic in CMEPS's scripts that depends on CISM_EVOLVE: this will be true if any ice sheet is evolving; we should check the logic in CMEPS to make sure this makes sense (i.e., if whatever is done in CMEPS's scripts when the overall CISM_EVOLVE is true should indeed be done when just one ice sheet is evolving, and vice versa)
[x] (Added 2021-10-07) The cism and ctsm nuopc caps access the overall cism_evolve attribute; look at what they do based on this and make sure it's right for them to look at this attribute that is true if any ice sheet is evolving.
[x] (Added 2021-10-07) And check for other uses of cism_evolve within cism