Add a GUI test for Enum.create_editor

[midhun-pm]

Related to #965

Adds a GUI test for the create_editor method for the Enum trait.

Checklist

• Tests
  - [ ] Update API reference (docs/source/traits_api_reference)
  - [ ] Update User manual (docs/source/traits_user_manual)
  - [ ] Update type annotation hints in traits-stubs

[midhun-pm]

Found another issue with the Enum editor when values is a collection, that this test fails to catch as it merely iterates through all the items in the dropdrown, changes them and ensures that the value of the trait also changes. Will need to change this test, after fixing the UI issue.

UI Bug seen in the Numbers field in the image.
Scratch that, it was caused due to some inconsistency in my local environment, PR is good for review.

[midhun-pm]

There is one known issue which this test case conveniently avoids. enthought/traitsui#781

[kitchoi]

Thank you for looking into this and identify the traitsui issue. It is certainly worth doing this exercise.

[codecov-io]

Codecov Report

Merging #988 into master will increase coverage by 1.71%.
The diff coverage is 96.19%.

@@            Coverage Diff             @@
##           master     #988      +/-   ##
==========================================
+ Coverage   73.05%   74.77%   +1.71%     
==========================================
  Files          51       51              
  Lines        6514     6457      -57     
  Branches     1309     1279      -30     
==========================================
+ Hits         4759     4828      +69     
+ Misses       1363     1258     -105     
+ Partials      392      371      -21     

Impacted Files	Coverage Δ
traits/ctrait.py	71.07% <ø> (+12.51%)	⬆️
traits/has_traits.py	72.17% <ø> (+0.92%)	⬆️
traits/trait_base.py	68.83% <ø> (+5.25%)	⬆️
traits/trait_types.py	72.31% <92.59%> (+0.66%)	⬆️
traits/trait_list_object.py	97.51% <97.05%> (+23.22%)	⬆️
traits/testing/optional_dependencies.py	95.00% <100.00%> (+0.55%)	⬆️
traits/trait_handlers.py	61.11% <100.00%> (ø)
traits/editor_factories.py	79.59% <0.00%> (-9.19%)	⬇️
traits/trait_type.py	77.21% <0.00%> (-1.10%)	⬇️
traits/trait_dict_object.py	75.44% <0.00%> (ø)
... and 9 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update bac9f7b...386ede7. Read the comment docs.

[kitchoi]

I am not sure if this is ready for review again...
If we are aiming at test coverage by launching the GUI without getting too involved with the details, this is all we need?:

    def test_create_editor(self):
        obj = EnumCollectionGUIExample()

        # Create a UI window
        ui = obj.edit_traits()
        try:
            self.gui.process_events()
        finally:
            with self.delete_widget(ui.control):
                ui.dispose()

[midhun-pm]

I am not sure if this is ready for review again...
If we are aiming at test coverage by launching the GUI without getting too involved with the details, this is all we need?:

    def test_create_editor(self):
        obj = EnumCollectionGUIExample()

        # Create a UI window
        ui = obj.edit_traits()
        try:
            self.gui.process_events()
        finally:
            with self.delete_widget(ui.control):
                ui.dispose()

Oh, I thought that we were going to keep it the way it is, but I came up with the current code after watching a lunch and learn by Corran and don't have much experience testing GUIs that use traits. Should I change it to the test you suggested ?

[kitchoi]

The situation is a bit unique here for traits, because supposedly traits should not depend on traitsui, so we'd want a more lightweight tests. For applications using traits and traitsui, it would make sense to have some integration tests that exercises GUI logic.

[mdickinson]

@kitchoi Yes, I think that's all we need; just enough to exercise the code in the create_editor methods, and no more. Any of the details of the UI should be tested in TraitsUI, not Traits.

[kitchoi]

LGTM!