Bug summary

If a check_figures_equal test fails in test_pickle.py (which has no baseline image tests at all), then running triage_tests.py fails with "ValueError: Can't find baseline dir for test_pickle".

Code for reproduction

Make test_pickle::test_complete fail, e.g. by commenting out the `fig_test.figimage(loaded.canvas.renderer.buffer_rgba())` line; run that test (which should fail), and run tools/triage_tests.py.

Actual outcome

ValueError: Can't find baseline dir for test_pickle

Expected outcome

The test triager succeeds.

Additional information

Fixing involves either creating an empty test_pickle directory in the baseline_images directory, or allowing missing directories in triage_tests (which already interprets missing baselines as arising from check_figures_equal tests):

diff --git i/tools/triage_tests.py w/tools/triage_tests.py
index 5153b1c712..6df720f29d 100644
--- i/tools/triage_tests.py
+++ w/tools/triage_tests.py
@@ -263,7 +263,7 @@ class Entry:
             ]
         self.thumbnails = [self.dir / x for x in self.thumbnails]

-        if not Path(self.destdir, self.generated).exists():
+        if self.destdir is None or not Path(self.destdir, self.generated).exists():
             # This case arises from a check_figures_equal test.
             self.status = 'autogen'
         elif ((self.dir / self.generated).read_bytes()
@@ -281,7 +281,6 @@ class Entry:
             path = self.source / baseline_dir / reldir
             if path.is_dir():
                 return path
-        raise ValueError(f"Can't find baseline dir for {reldir}")

     @property
     def display(self):

Opening as an issue because I don't really mind about which solution to pick, so I'll let whoever cares pick one.

Operating system

No response

Matplotlib Version

3.11.0.dev525+g9f7b3dd205

Matplotlib Backend

No response

Python version

3.13

Jupyter version

No response

Installation

git checkout