cc @DragaDoncila @Czaki

Codecov Report

Attention: Patch coverage is 92.91339% with 9 lines in your changes are missing coverage. Please review.

❗ No coverage uploaded for pull request base (main@c5846e8). Click here to learn what that means.
Report is 24 commits behind head on main.

@@           Coverage Diff           @@
##             main    #6743   +/-   ##
=======================================
  Coverage        ?   92.38%           
=======================================
  Files           ?      611           
  Lines           ?    54837           
  Branches        ?        0           
=======================================
  Hits            ?    50663           
  Misses          ?     4174           
  Partials        ?        0           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

So the _provide_active_layer and _provide_active_layer_list do not current have tests and AFAICT they are not being used (I don't think any of our actions have a Layer/LayerList annotations). I feel like Talley may have written these for some future plans. Regardless I have added a simple test.

We've narrowed down why the test_run_outside_ipython test fails when make_napari_viewer is used in a prior test. TLDR: we should be using make_napari_viewer for this test, as we do for all other tests using the Viewer, because this ensures proper clean up and set up for tests - in this case it's the cache clearing that we're missing.

This is what happens when you add a dummy test that uses make_napari_viewer to the top of the test_app file:

• The first (dummy) test runs, mock_app fixture creates an app during test setup and registers all non-qt napari built-in app-model actions, providers and processors
• The test starts, builds a make_napari_viewerwhich clears all caches, runs initialize_plugins and registers qt plugin actions with the app (samples, widgets)
• qt_main_window.__init__ runs, registering napari built-in qt-dependent actions, qt providers and qt processors with the app
  • potential problem some of the plugin actions previously probably request the viewer, which we can't provide until this point and they also may need qprocessors to work at all e.g. widgets
• dummy test concludes

Then, it's time for the real test to run - the test_run_outside_ipython test.

• During test setup, mock_app builds a new app, registers non qt stuff including the viewer toggle action
  • potential problem: this action implicitly depends on qt because its "check toggled" function requires a Viewer - this really is a qt action and should be moved
  • potential problem the Viewer CAN be non-qt, but the viewer PROVIDER needs qt, is there a world where someone validly wants the headless viewer via injection
• The test doesn't use make_napari_viewer; this means the cache isn't cleared (for plugins, or for qactions)
• The test tries to instantiate a "normal" viewer
  • "normal" viewer doesn't run initialize_plugins because it's cached, so no qt or nonqt plugins stuff is registered
• qt_main_window runs, doesn't init_qactions, because it's cached, so no qt processors, providers or actions are registered for built-in napari stuff
• somewhere down the line we rebuild menus, triggering a check for whether the view action is toggled
• the toggled check function requires a viewer
• but qt providers were never registered, because init_qactions was never run
• the test fails because a Viewer can't be provided

Note the test failures are because some non-qt view actions actually req Qt (and the viewer provider I have just moved). This can be fixed by using make_napari_viewer, (which forces init_qactions to be run) - this makes sense in some tests but not others.

E.g., this test fails:

I think this needs to stay as is, and so this PR is probably dependent on #6767.

CI failure is real.
I will need to move the processors that use the newly migrated providers into _qt/ as well. The processor tests will also need to be moved.

Okay @DragaDoncila this is finally green!

@Czaki @DragaDoncila could this go in?

Yes. I just merged main to validate if latest updates (like mypy requirements) do not broke something.