Fix flaky simplicial set test

[user202729]

Fixes #38888 .

This test has been failing sporadically recently e.g. https://github.com/sagemath/sage/actions/runs/11733902583/job/32689016794?pr=38938 .

From my understanding of simplicial set, the two sigma_1 are actually representing two different cells (the wedge looks like a "8" shape), but because they both come from S1 (the 1-sphere i.e. circle), they get the same name.

SageMath does some magic during doctest so that dict entries are sorted, but the problem is

        At this point, comparison between v and w is random, based on
        their location in memory. ::

so with some chance it fails.

This change should make it deterministic --- since the two faces are named sigma_1 and sigma_1', the comparison should be by custom name…?

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes. (not applicable)
• I have updated the documentation and checked the documentation preview.

[github-actions]

Documentation preview for this PR (built with commit f070aa3; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

[user202729]

The failed test is

2024-11-08T15:09:09.5768206Z **********************************************************************
2024-11-08T15:09:09.5893109Z File "src/sage/groups/finitely_presented_named.py", line 489, in sage.groups.finitely_presented_named.AlternatingPresentation
2024-11-08T15:09:09.5894783Z Failed example:
2024-11-08T15:09:09.5895452Z     A1.is_isomorphic(A2), A1.order()
2024-11-08T15:09:09.5896032Z Expected:
2024-11-08T15:09:09.5896389Z     (True, 1)
2024-11-08T15:09:09.5896732Z Got:
2024-11-08T15:09:09.5897089Z     #I  Forcing finiteness test
2024-11-08T15:09:09.5897589Z     (True, 1)
2024-11-08T15:09:11.2068744Z **********************************************************************

This is obviously #38889 instead.

By the way, GitHub Actions doesn't seem to be able to report test failure to the GitHub interface?

[orlitzky]

This is obviously #38889 instead.

Potential solution: #38956

[tobiasdiez]

Thanks! The change looks good to me, as far as I understand the problem. But it would be good if someone more familiar with the code has a look at well.

Marking it already as "ci fix" so that other PRs don't keep falling because of this issue.

[miguelmarco]

Since there is no actual code being changed (just doctests), I see no risk in accepting it.

[user202729]

There's always the risk of accidentally telling the user "this is the right way to do things" while it actually relies on dangerous internal implementation detail e.g. are you sure that the 1-cell is not shared between different objects, and mutate it is safe?