On inspection, the XCLIM_TESTDATA_BRANCH environment variable is set to the most recent tag (v2025.1.8).

No further action is required.

It looks as though there are changes to the ways that docstrings are parsed in Python3.13: https://docs.python.org/3.13/whatsnew/3.13.html#other-language-changes

This is breaking a lot of docstring tests. Will see what I can do.

Update: This looks like it's going to be complicated. Here's the PR that added the changes: python/cpython#81283

@Zeitsperre This commit pushed a very simple fix to the indentation issue. I have looked at docstring_parser and it might be a good alternative. The hypothesis is that it would be stronger. My doubt are because its a new dependency. And you know, these tend to break ;) . Also, I haven't taken the time to test it.

I have looked at docstring_parser and it might be a good alternative. The hypothesis is that it would be stronger. My doubt are because its a new dependency. And you know, these tend to break ;) . Also, I haven't taken the time to test it.

I'm not keen to add a new requirement (though maybe that library will find use elsewhere), even if it's a transitory one (until we no longer support Python3.12 and lower, so years, I suppose). If we can extract the parsing logic that we currently have to python version-dependent functions, we'll probably be much better off.

I'll see what I can do with your approach. Thanks!

The new code dedents the docstrings before parsing them, which assumes that all lines of the docstrings have a common indent. This is broken when multi-line docstrings start on the first line : this was the case in sdba.properties, which I fixed in the last commit.

In linting speak, this means we want to enforce rule D213 and not D212:

D212 	Multi-line docstring summary should start at the first line
D213 	Multi-line docstring summary should start at the second line

@aulemahal I think I've managed to clear up a few docstring issues (and some formatting errors that have been present for a long time, too!). The remaining warnings I'm seeing arise from the docstring for robustness_fractions:

We might need to revisit the logic of:

robustness_fractions.__doc__ = robustness_fractions.__doc__.format(
    tests_list="{" + ", ".join(list(SIGNIFICANCE_TESTS.keys()) + ["threshold"]) + "}",
    tests_doc="\n".join(map(_gen_test_entry, SIGNIFICANCE_TESTS.items()))[
        4:  # strip first 4 chars
    ],
)

coverage: 89.967% (+0.03%) from 89.934%
when pulling 96d93d3 on support-python313
into d491487 on main.

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

I'm reluctant to merge this until our documentation builds on Python3.13. The last issue is the parsing of docstrings needed for the robustness_fractions.

If I have time on Monday, I imagine the fix is quite simple (dedenting for all entries before constructing the section).