Update ClimaDiagnostics in compat #4151
Conversation
ph-kev
force-pushed
the
kp/update-diag
branch
from
77a316a to
1f51632
Compare
ph-kev
force-pushed
the
kp/update-diag
branch
6 times, most recently
from
e80086b to
6d747a7
Compare
|
@costachris This will break the gcm_driven_scm experiment (https://github.com/CliMA/ClimaAtmos.jl/tree/main/calibration/experiments/gcm_driven_scm), but I can't tell if this experiment is still in use. Should this experiment be updated to support ClimaDiagnostics v0.3? |
juliasloan25
left a comment
juliasloan25
left a comment
There was a problem hiding this comment.
This looks good to me! I would just want to hear from Costa about the experiment you mentioned before merging.
@szy21 can we tag a patch release so we can update the coupler too, or should we wait since things are broken right now?
We can tag but it will be a breaking release because of this: #4198. Maybe we can tag a release when the coupler surface flux PR is ready? |
Sounds good. Teja and I are also working on updating the Makie compats in ClimaCore/ClimaAtmos/ClimaLand. It will be good to make releases of those packages when those are in too |
Sounds good. We can tag the next minor release after all of these are done. |
nefrathenrici
left a comment
nefrathenrici
left a comment
There was a problem hiding this comment.
This looks good to me. @costachris can chime in, but I think we should add ClimaDiagnostics and a compat "=0.2.15" to the calibration experiment Project.toml. That should help keep the pipeline reproducible.
|
We do still use gcm_driven_scm so pinning "=0.2.15" as Nat suggested sounds good to me! |
ph-kev
force-pushed
the
kp/update-diag
branch
2 times, most recently
from
53a69af to
ac3e89f
Compare
| ClimaAtmos = "=0.27.6" | ||
| ClimaCalibrate = "=0.1.4" | ||
| ClimaDiagnostics = "=0.2.15" |
There was a problem hiding this comment.
I wasn't sure about the versions of the packages. I don't think there was anything breaking going from ClimaCalibrate 0.0.3 to 0.1.4.
cc: @nefrathenrici
ph-kev
force-pushed
the
kp/update-diag
branch
from
ac3e89f to
6ce2094
Compare
This updates ClimaDiagnostics to v0.3. The only breaking change is that the diagnostics are saved at the start of the reduction period instead of at the end.
I don't think this affects anything, but the postprocessing of the NetCDF files. I couldn't find any postprocessing of the diagnostics that depends on the exact time values beside the calibration.
In general, any plots with instantaneous variables should be unchanged, but any plots with time reduction should differ by single time period. For example, a plot with
time = 8dwould change totime = 7dif the time reduction is daily.ClimaDiagnostics v0.2.13 is still supported, because ClimaCoupler does not support v0.3 yet.