Pre release v2.5.0rc4

[schlunma]

Checklist

• Used a personal fork of the feedstock to propose changes
• Bumped the build number (if the version is unchanged)
• Reset the build number to 0 (if the version changed)
• Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• Ensured the license file is being packaged.

Closes #30

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

[schlunma]

@conda-forge-admin, please rerender

[schlunma]

One question on this: I cannot find the dependency pillow in our environment.yml, setup.py or meta.yml file. Do we need it?

It is present in doc/requirements.txt, but in this files there are many other dependencies which are not listed here (e.g., autodocsum, docutils, etc.).

EDIT: Similarly, cartopy is also not present in our ESMValCore dependencies (not even in doc/requirements.txt).

[schlunma]

Does anyone no why this test is failing? It ran fine on the nightly builds on CircleCI and GitHub actions, and also runs fine on my local machine and mistral...

[zklaus]

I do not. Let's see if it is a fluke. If not, I'll have a look with tmate.

@conda-forge-admin, please restart ci

[zklaus]

So... I still don't know what's going wrong. But I can say that in the ESMValTool build we simply ignore the corresponding test, see here. However, in that case, we don't even ship the documentation by excluding it from distribution in the MANIFEST.in file. I don't recall the reason for that right now; it may simply be the size with all the pictures in it. For the core, that wouldn't be an argument since all the doc folder comes in at just over 800KB, which is not significant compared to the total size of 20MB, or the 11MB of the NE masks.

I think you should add the --ignore="tests/unit/documentation/test_duplications_in_changelog.py" switch, perhaps even for the whole tests/unit/documentation folder; when the build lands here, this has been sufficiently checked already.

[schlunma]

Sounds good, thanks Klaus!! What do you think about the pillow and cartopy dependencies?

[valeriupredoi]

just took a look too - I am fairly sure conda build is ignoring the doc dir, it finds all Python stuff that need building and puts them in export SRC_DIR=/home/conda/feedstock_root/build_artifacts/esmvalcore_1646296297036/test_tmp for testing, but omitting doc

[valeriupredoi]

we got rid of pillow in Core no? So I'm guessing that was the only place we needed it, but no more, what's the problem with cartopy?

[schlunma]

Not a problem with cartopy, but it's not explicitly listed in any of our dependencies for ESMValCore, so I'm wondering if we need it here.

[valeriupredoi]

I think we can remove both - @zklaus please confirm 🍺

[zklaus]

The silver search (ag, you may prefer to use e/f/grep or ack) shows that we use PIL aka pillow in esmvalcore._provenance and cartopy in esmvalcore.preprocessor._mask. Both are direct dependencies and should be added as such.

[schlunma]

Makes sense, but then it would be good to add them to our environment.yml and setup.py, too, right?

[valeriupredoi]

OK on the doc failed test - it's now clear that conda build does not touch the docs and does not include them in the package ie see the contents of esmvalcore-2.5.0rc3-pyh39db41b_1.tar.bz2 - it packages just the esmvalcore and during the build process it grabs only the tests dir from outside it to run the tests, so we sould not include any unit/integration tests that need files from doc

[valeriupredoi]

cheers, Klaus! - about pillow and cartopy - they are both direct dependency requirements for iris, see their requirements - they also pin them (alas, pillow is needed only for the test env in iris) so I think we should include them too if we import them directly, but we should see how to get around not importing them directly, hence getting us rid of them as direct deps

[schlunma]

Great, thanks for your input. Can we merge this?

[zklaus]

cheers, Klaus! - about pillow and cartopy - they are both direct dependency requirements for iris, see their requirements - they also pin them (alas, pillow is needed only for the test env in iris)

Of course they are.

so I think we should include them too if we import them directly, but we should see how to get around not importing them directly, hence getting us rid of them as direct deps

That doesn't make sense to me. You are saying we shouldn't use libraries ourselves that our dependencies are also using? By that token we shouldn't use Dask, Numpy, Scipy, or Python itself for that matter.

[valeriupredoi]

That doesn't make sense to me. You are saying we shouldn't use libraries ourselves that our dependencies are also using? By that token we shouldn't use Dask, Numpy, Scipy, or Python itself for that matter.

hahah, no, of course not - I'm saying we should see how(if) we can get around not using these two since they're each used in one single place and that is pretty specialized too. But that's for another time, no need to start looking into this now 👍