MNT Update doc lock files to get Sphinx-Gallery 0.19 minigallery bugfix

[lucyleeow]

Reference Issues/PRs

Bump min Update doc conda lock file for newest sphinx-gallery version which fixes minigallery duplication bug: sphinx-gallery/sphinx-gallery#1430 sphinx-gallery/sphinx-gallery#1435

What does this implement/fix? Explain your changes.

Any other comments?

cc @ogrisel @glemaitre

[github-actions]

✔️ Linting Passed

All linting checks passed. Your pull request is in excellent shape! ☀️

_{Generated for commit: d4e9d5b. Link to the linter CI: here}

[ogrisel]

Please push an empty commit with the [doc build] commit flag.

Otherwise LGTM. Thanks @lucyleeow!

[glemaitre]

There is not a lock file associated with the yaml one? In other words, we should need to regenerate the lock files isn't it?

[lesteve]

There is not a lock file associated with the yaml one? In other words, we should need to regenerate the lock files isn't it?

Yes we would need to regenerate the lock-file and push a [doc build]. Something like this should do it (I thought there was some quick doc for it but I can't find it right now):

python build_tools/update_environments_and_lock_files.py --select-build doc

[lucyleeow]

Yes but I was not 100% on the change so didn't update the lock files.

I checked and 0.17 had the mini gallery duplication bug so it's probably best to bump the min version.

I'll update the lock files

[lesteve]

To be honest what we care about is that the doc build has sphinx-gallery 0.19 so that we don't have duplicated examples in the API page on the scikit-learn.org website. We don't really care about bumping the min dependencies of sphinx-gallery but we can also bump it I don't have a strong opinion on this. The doc-min-depencies is mostly used to make sure that the examples run fine with old numpy, scipy, matplotlib, etc ... versions.

[lucyleeow]

The doc-min-depencies is mostly used to make sure that the examples run fine with old numpy, scipy, matplotlib, etc ... versions.

Yes this was why I was hesitant about updating the min dep. Happy to close ?

[lesteve]

Yes this was why I was hesitant about updating the min dep. Happy to close ?

It would still be useful to update the doc build lock-file and make sure that the issue has been fixed so something like this + a [doc build] commit:

python build_tools/update_environments_and_lock_files.py --select-build 'doc$'

For further reference, PartialDependencyDisplay is one of the page that currently shows a duplication

[lesteve]

No urgency at all, but since we install sphinx-gallery from conda-forge, we would also need that sphinx-gallery 0.19 is on conda-forge which doesn't seem to be the case yet.

Probably we need to wait a bit before the conda-forge automation creates a PR and then the PR needs to be merged, and then wait a bit more for it to be actually available.

[lucyleeow]

No urgency at all, but since we install sphinx-gallery from conda-forge, we would also need that sphinx-gallery 0.19

oh yeah! Some other kind soul created that

Looking at it more, I think it makes sense to bump the version in pyproject.toml and _min_dependencies.py as these may used by people to install dependencies/determine what version is required to build docs. BUT if we update _min_dependencies.py we should probably update build_tools/circle/doc_min_dependencies_environment.yml to match? WDYT?

[lesteve]

In general, no strong opinion on bumping the min dependency of sphinx-gallery. I would slightly lean towards keeping it to whatever we have since this is not a huge bug and it went unnoticed in scikit-learn for quite some time. You can still build fine with older versions and it's not like sphinx-gallery min dependency is forced on anyone, it's more which range of versions we decide to test in our CI.

BUT if we update _min_dependencies.py we should probably update build_tools/circle/doc_min_dependencies_environment.yml to match? WDYT?

This is done by the update_environment_and_lock_files.py script 😉. Basically you udapte _min_dependencies.py and update_environment_and_lock_files.py and then you update environment + lock-files by running update_environment_and_lock_files.py. The best doc is probably the docstring in the script itself actually.

IIRC pyproject.toml update is manual but there is a test that fails somewhere if it does not match _min_dependencies.py. Also I don't think many people actually rely on pip install scikit-learn[docs] or all the other variations for tests, examples, benchmarks but I could well be wrong about that ...

[lucyleeow]

Okay I will update this PR to just update the lock files and not bump min deps.

This is done by the update_environment_and_lock_files.py script 😉.

🤦 ah I knew that but it just didn't click that was a .yml file. I will blame working too late at night, trying to catch you Europeans in your morning 😅

[lucyleeow]

Doc build failing due to not being able to download dataset in example:

    ../examples/miscellaneous/plot_display_object_visualization.py failed leaving traceback:

    Traceback (most recent call last):
      File "/home/circleci/project/examples/miscellaneous/plot_display_object_visualization.py", line 35, in <module>
        X, y = fetch_openml(data_id=1464, return_X_y=True)
      File "/home/circleci/project/sklearn/utils/_param_validation.py", line 218, in wrapper
        return func(*args, **kwargs)
      File "/home/circleci/project/sklearn/datasets/_openml.py", line 1028, in fetch_openml
        data_description = _get_data_description_by_id(data_id, data_home)
      File "/home/circleci/project/sklearn/datasets/_openml.py", line 363, in _get_data_description_by_id
        json_data = _get_json_content_from_openml_api(
      File "/home/circleci/project/sklearn/datasets/_openml.py", line 254, in _get_json_content_from_openml_api
        raise OpenMLError(error_message)
    sklearn.datasets._openml.OpenMLError: Dataset with data_id 1464 not found.

trying again (in case this is transient) ...

[lesteve]

The duplication seems indeed to have been fixed, see the PartialDependencyDisplay rendered page.

I will blame working too late at night, trying to catch you Europeans in your morning 😅

😴 😉