Always say why something is out of date

[larsoner]

Adds logging in two branches that was missing from the logger.debug. Gives output like:

[build target] outdated 'auto_examples/future/plot_future_imports' from dependency '/home/larsoner/python/sphinx-gallery/sphinx_gallery/tests/tinybuild/doc/auto_examples/future/images/thumb/sphx_glr_plot_future_imports_broken_thumb.png': 2023-08-09 18:36:25.911416 -> 2023-08-09 18:36:34.238679

Perhaps it's a bit verbose but it at least indicates that this file is marked outdated because a dependency of it was created after the file that uses it.

BUT this brings up an issue -- the dependency itself was not changed the two calls to make html. The conditional if depmtime > mtime: around line 501 checks if the mtime of the file during previous build is older than that of the dependency in this build, when it seems like it should check if the mtime of the dependency during previous build is older than that of the dependency in this build. In other words, if I create index.rst, then create whatever.inc, and .. include: whatever.inc, index.rst will always show up as out of date just because of the order in which I created index and whatever.inc.

Should I try to fix that here as well? It seems like a potentially tough thing to fix because it would mean adding all dependency files to self.all_docs (where self is the BuildEnvironment) -- or maybe better adding a self.all_deps that is a str->float mtime mapping for all dependencies.

[larsoner]

BUT this brings up an issue -- the dependency itself was not changed the two calls to make html

Looks like this has actually been fixed on master sometime so updating fixed this part for me!

[larsoner]

https://github.com/sphinx-doc/sphinx/actions/runs/5821558864/job/15784236771?pr=11572#step:9:2434

                        if depmtime > mtime:
>                           depmtime_dt = datetime.utcfromtimestamp(depmtime)
E                           ValueError: year 53608964 is out of range
...
=========================== short test summary info ============================
FAILED tests/test_intl.py::test_gettext_dont_rebuild_mo - ValueError: year 53608965 is out of range
FAILED tests/test_intl.py::test_html_rebuild_mo - ValueError: year 53608964 is out of range
FAILED tests/test_intl.py::test_xml_footnotes - ValueError: year 53608964 is out of range
====== 3 failed, 1957 passed, 22 skipped, 2 warnings in 192.82s (0:03:12) ======

[picnixz]

Wow. Ok, so maybe utcfromtimestamp is not the correct way to do it? I think on line 492 we are using fromtimestamp (we divide the timestamp by 1_000_000 before but I'm not sure that it's ok to do the same here, so you'll definitely need to check whether this should apply or not). I also think it's the preferred way to do it (see https://docs.python.org/3/library/datetime.html#datetime.datetime.fromtimestamp).

[larsoner]

we divide the timestamp by 1_000_000 before

Ahhh okay this is indeed probably the problem -- I'll simplify the code and push soon!

[larsoner]

Okay finally green!

[AA-Turner]

Please may you add an entry to CHANGES?

A

[larsoner]

Done @AA-Turner !

[larsoner]

Windows failure looks probably unrelated... is it a flakey test?

[AA-Turner]

is it a flakey test?

Yes, see #11232; #11500.

A

[AA-Turner]

Thank you @larsoner!

A