Coordinate with the ENSO package update

[lee1043]

ENSO package update was made thanks to @zhangshixuan1987, and its v2.0.0 was released through conda-forge (conda-forge/enso_metrics-feedstock#10). This PR is opened to finalize the needed update for the PMP side to coordinate with the ENSO pacakge.

[zhangshixuan1987]

@lee1043 : I think that based on my local test, you only need to update the YAML file to include the enso metrics package so that pcmdi will compile the enso metrics package.

[lee1043]

@kristinchang3 when you get a chance, could you please validate this PR? It seems it works well on my end and also @zhangshixuan1987's end, but as there is a major change, I was hoping to get it tested one more time in a completely new environment. Running Demo 6 notebook might be a sufficient and important sanity check. Thank you.

[kristinchang3]

@lee1043 I am having trouble installing the enso_metrics package from conda-forge - even though I can see it is merged and shows up in the anaconda library... I tried installing from the dev.yml and also manually using conda install -c conda-forge enso_metrics . But neither was successful.

I am running conda clean --all and will try again to make sure it is not an issue on my local.

[kristinchang3]

@lee1043 I was able to install enso_metrics on nersc which is a linux-64 system. But still am unable to install in my local environment which is osx-arm64.

[lee1043]

@kristinchang3 thanks for testing. conda install conda-forge::enso_metrics worked me for my mac OS. Could you check if any luck by using the same command?

[kristinchang3]

@lee1043 unfortunately, that command did not work for me either...

[kristinchang3]

@lee1043 do you get any results with conda search -c conda-forge enso_metrics ? I just ran this and getting the error that the package is not available

[lee1043]

Hi @kristinchang3, thanks for checking. I am getting the following when I do on my Mac (M3 cpu) but have not had a chance to check with osx-arm64.

$ conda search -c conda-forge enso_metrics
Loading channels: done
# Name                       Version           Build  Channel             
enso_metrics                   1.1.1 py310h2ec42d9_1  conda-forge         
enso_metrics                   1.1.1  py37hf985489_0  conda-forge         
enso_metrics                   1.1.1  py37hf985489_1  conda-forge         
enso_metrics                   1.1.1  py38h50d1736_0  conda-forge         
enso_metrics                   1.1.1  py38h50d1736_1  conda-forge         
enso_metrics                   1.1.1  py39h6e9494a_0  conda-forge         
enso_metrics                   1.1.1  py39h6e9494a_1  conda-forge         
enso_metrics                   1.1.2 py310h2ec42d9_1  conda-forge         
enso_metrics                   1.1.2  py39h6e9494a_1  conda-forge         
enso_metrics                   1.1.3 py310h2ec42d9_0  conda-forge         
enso_metrics                   1.1.3  py39h6e9494a_0  conda-forge         
enso_metrics                   1.1.4 py310h2ec42d9_0  conda-forge         
enso_metrics                   1.1.4  py39h6e9494a_0  conda-forge         
enso_metrics                   1.1.5 py310h2ec42d9_0  conda-forge         
enso_metrics                   1.1.5  py39h6e9494a_0  conda-forge         
enso_metrics                   2.0.0 py310h2ec42d9_0  conda-forge         
enso_metrics                   2.0.0 py311h6eed73b_0  conda-forge         
enso_metrics                   2.0.0 py312hb401068_0  conda-forge         
enso_metrics                   2.0.0 py313habf4b1d_0  conda-forge         

[kristinchang3]

@lee1043 Ah I see now that enso_metrics is not built for osx-arm64 but only linux-64, macOS-64. Is there a reason for this? If not, I think we can update the feedstock recipe to include noarch: python in the build section. This line is included in the pcmdi_metrics feedstock recipe and that is built for noarch.

[lee1043]

@kristinchang3 ah! Great you have figured it out -- thanks! Could you help on updating the recipe?

[kristinchang3]

@lee1043 Yes I will work on that!

[kristinchang3]

Hi @lee1043 apologies for the delay! I updated the feedstock recipe and now successfully installed the enso_metrics package in my environment. The code runs well except for one small issue with Demo 6. The caseid for basic_enso_param.py is "basicTestEnso" but the plots starting in cell 11 are being saved to a hardcoded path with caseid "basicTestEnso_xcdat" so I am getting an error that the path does not exist because the directory names do not match.

I changed these manually to test the code anyway and here are the resulting plots:
ENSO_perf:

ENSO_tel:

ENSO_proc:

Aside from the directory name mismatch everything looks good to me!

[lee1043]

@kristinchang3 thanks, just got it corrected!

[lee1043]

WIll merge this PR once github action tests pass