Run remaining un-run examples - #1421
Merged
larsoner merged 5 commits intoAug 20, 2026
Merged
Conversation
The gallery render has held its examples to warnings-as-errors for a while; `pytest examples` did not, so the thirty it runs could warn freely. Make them fatal there too, from one shared inventory rather than a second copy: `scripts/render_docs.py`'s `WARNING_FILTERS` moves to `mayavi/tests/common.py:EXAMPLE_WARNING_FILTERS`, which both callers now read. It has to be applied in Python -- pytest's own `filterwarnings` covers the process pytest runs in, and every case here is a subprocess. Six of the thirty warned. Four are the pyface workbench PendingDeprecationWarning, unsatisfiable until the import has somewhere to move to, and are filtered. The other two are real: - `tvtk/pyface/ui/qt4/actor_editor.py` tried `traitsui.qt4.editor` before `traitsui.qt.editor`. On traitsui >= 8 the first still resolves, through the shadowed-module alias, and warns while doing it. `scene_editor.py` beside it already had them the right way round; match it. - `examples/mayavi/advanced_visualization/datasets.py` called two things VTK 9.6 deprecated: `CellArray.set_cells(n, cells)`, which 9.7 removes outright, and the `UnstructuredGrid.set_cells` overload taking cell locations. `import_legacy_format` takes the same flat (npts, p0, p1, ...) array the example already builds -- it is what `array_handler` uses internally -- and the two-argument `set_cells` replaces the other. The hand-built `offset` array goes with them: the cell array carries its own offsets now, so building one was teaching something no longer true. `mayavi/tests/datasets.py:mixed_type_ug` is a copy of that same code, reached by nothing that runs with warnings fatal, so it would have gone on quietly until a 9.7 row hit it. Same fix. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Taken from the docs-site artifact of the Docs build on this branch, since the render is only reproducible within one environment and a local macOS render moves all seventy-five of them. The one that matters is example_coil_design_application: its figure has been a picture of two coils and no magnetic field ever since np.NAN went away, because the exception was raised inside a traits notification handler and swallowed. It now has the B field, the axes and the field node in the pipeline tree. The other twenty-five are drift that predates this branch -- the Docs run on main at 3ea411b reports the same twenty-six -- so committing them is what makes the job's diff empty again rather than anything this branch renders differently. example_magnetic_field is not among them: it is in FLAKY_EXAMPLES, so the render now runs it (which is how its divide-by-zero was found) but keeps the committed image. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The new tvtk/tests/test_ivtk.py takes the pyqt6 row down -- not with a failure but with exit 134, so the run dies before pytest can say why. Reproduced in a PyQt6 venv: two independent upstream bugs, both of which every IVTK window (and so ivtk.viewer()) walks into, and neither of which is tvtk's to fix. pyface's _MenuItem calls QMenu.addAction(text, slot, shortcut). PyQt6 has no such overload -- its three-argument forms are (text, slot, type) and (text, shortcut, slot) -- so building the menu bar raises TypeError. That is what failed test_ivtk, which has no splitter to reach the second one. traitsui's _GroupSplitter._resize_items seeds its sizes from Item.width, a Float, and returns early on `if avail <= 0` before coercing anything, so a splitter that is still zero-sized reaches QSplitter.setSizes([-1.0, -1.0]). PyQt6 rejects the floats; the call is inside a showEvent, where the TypeError is unraisable, so Qt aborts the process. That is the exit 134, and no skip inside the test could have caught it -- hence skipping the class. Both reproduce against pyface and traitsui main, so the skip cannot be keyed on a release yet. PySide6 coerces the floats and has the addAction overload, so the rest of the matrix still covers what these tests are for -- the pyface 8 create() regression in tvtk/tools/ivtk.py. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
PyQt6 dropped the unscoped enums, but pyface still names QFont.TypeWriter -- in its console widget, its code editor and its font registry -- so building the Python shell view there raises AttributeError. Verified against pyface 8.0.0 on PyQt6: `PythonShell(parent).create()` fails without the alias and builds with it. mayavi had this fix and lost it. It was written as a helper of `fix_python_shell_view`, and when envisage 8.0.1 made that workaround unnecessary the whole block went, alias included -- but the alias is a pyface bug, nothing to do with envisage, and is still needed. It comes back as a function of its own, applied from `mayavi_workbench_application.run()` beside `fix_view_chooser()`. Nothing had noticed because nothing built a shell: the pyqt6 row's suites do not, and the docs render is PySide6. So `test_workbench_fixes.py` now builds one, which fails with exactly that AttributeError if the alias is taken away again. Also names this as the third of the upstream bugs behind the PyQt6 skip in tvtk/tests/test_ivtk.py -- the previous commit's message and comment had only the two that the abort left visible. The two `WithCrust` windows hit QFont.TypeWriter through the same console widget; nothing applies the alias on a tvtk-only path, and fixing just that one would not make the windows work while the pyface menu and traitsui splitter bugs stand, so the skip stays. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Collaborator
Author
|
Good to run these for example for #1422, so let's get this in! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
... and then fix errors found while running them with warnings-as-errors.