search

ESCOMP / CMEPS

NUOPC Community Mediator for Earth Prediction Systems
https://escomp.github.io/CMEPS/
24 stars 79 forks source link

For T compsets, only run GLC if med_to_glc is true #425

Closed billsacks closed 10 months ago

billsacks commented 11 months ago

Description of changes

This makes the logic for running GLC consistent with runseq_general.py. This change is relevant for the unusual case where a T compset (CISM forced by dlnd) is being run with CISM in noevolve mode: in this case, we don't want to run CISM, and trying to run CISM leads to incorrect behavior in a restart case (which is how this issue was detected).

With this change, ERS_D_Ly3.f09_g17_gris4.T1850Gg.derecho_intel.cism-noevolve passes.

Specific notes

Contributors other than yourself, if any: @Katetc

CMEPS Issues Fixed (include github issue #): none

Are changes expected to change answers? No - bfb (except for the unusual case of a T compset with noevolve CISM)

Any User Interface Changes (namelist or namelist defaults changes)? No

Testing performed

ERS_D_Ly3.f09_g17_gris4.T1850Gg.derecho_intel.cism-noevolve

Using escomp/cism-wrapper@bf0b382

billsacks commented 11 months ago

@Katetc can you please confirm that this solves your issue with ERS_D_Ly3.f09_g17_gris4.T1850Gg.derecho_intel.cism-noevolve? And can you also confirm that a normal (not noevolve) T compset still passes and is bit-for-bit with this change?

@jedwards4b I'm assigning you in order to have an official CMEPS gatekeeper as the official assignee.

Katetc commented 10 months ago

Hi everybody, this code update fixes the test in my sandbox. Thanks for your help @billsacks and @jedwards4b !

billsacks commented 10 months ago

Thanks a lot @Katetc !

@jedwards4b I think this is ready to merge whenever you feel is a good time.