I'm working on https://github.com/ESCOMP/CTSM/issues/1135 and plan to make a CTSM tag soon with that change (likely ctsm5.1.dev011). Once I make that change, this test in the aux_glc test suite will no longer do what it says it's doing: SMS_Lm13.f10_f10_musgs.I2000Clm50Sp.cheyenne_intel; this test has the following comment:
Include an I compset test with CISM in NOEVOLVE mode. The purpose of this test is: Often there are answer changes in CISM, but we expect that these should not affect NOEVOLVE runs (which is the typical CESM configuration). This test allows us to do a baseline comparison to confirm that expectation. 13-months rather than 1-year in order to allow time for CISM to affect CLM (and CLM to produce the next month's history file).
(This is a test that I added a while ago.)
However, after my change, I2000Clm50Sp will not be running CISM in NOEVOLVE mode; instead it will be using SGLC. In fact, we won't have any I compsets with CISM in NOEVOLVE mode.
I think the best thing to do in order to keep the functionality of this test is to change it to use a compset with an evolving CISM, like I1850Clm50SpG, but have it point to a testmods directory that turns off ice evolution (via a shell_commands file that contains ./xmlchange CISM_EVOLVE=FALSE). An alternative would be to define one I compset with CISM2%NOEVOLVE just for the sake of this test; that feels more confusing to me, but I'll leave this decision up to @Katetc .
@Katetc can I leave this to you to take care of? I think it will only be important once you update the CISM-wrapper externals to pull in a new version of CTSM, but you could safely do as I suggested before then. I should point out that the existing test will pass (but change answers), but won't actually be testing what it claims to be testing once you are pointing to a newer version of CTSM.
I'm working on https://github.com/ESCOMP/CTSM/issues/1135 and plan to make a CTSM tag soon with that change (likely ctsm5.1.dev011). Once I make that change, this test in the aux_glc test suite will no longer do what it says it's doing:
SMS_Lm13.f10_f10_musgs.I2000Clm50Sp.cheyenne_intel
; this test has the following comment:(This is a test that I added a while ago.)
However, after my change, I2000Clm50Sp will not be running CISM in NOEVOLVE mode; instead it will be using SGLC. In fact, we won't have any I compsets with CISM in NOEVOLVE mode.
I think the best thing to do in order to keep the functionality of this test is to change it to use a compset with an evolving CISM, like
I1850Clm50SpG
, but have it point to a testmods directory that turns off ice evolution (via ashell_commands
file that contains./xmlchange CISM_EVOLVE=FALSE
). An alternative would be to define one I compset withCISM2%NOEVOLVE
just for the sake of this test; that feels more confusing to me, but I'll leave this decision up to @Katetc .@Katetc can I leave this to you to take care of? I think it will only be important once you update the CISM-wrapper externals to pull in a new version of CTSM, but you could safely do as I suggested before then. I should point out that the existing test will pass (but change answers), but won't actually be testing what it claims to be testing once you are pointing to a newer version of CTSM.