On Fri, 8 Nov 2024 00:35:10 GMT, Martin Fox <m...@openjdk.org> wrote:

> > The failing `SpinnerTest::testEnterEscapeKeysWithDefaultCancelButtons` test 
> > is a bit odd; it expects that Enter in an **editable** Spinner should 
> > result in activating the Default Button. However, a normal TextField 
> > doesn't do this either...
> 
> Maybe I'm misunderstanding this comment but in my testing pressing Enter in 
> an editable TextField can activate the default button as long as the 
> TextField doesn't have an OnAction handler set. It should not be so easy to 
> block Enter from activating the default button.

Ah, perhaps that's the problem, I did have an action handler, although it did't 
consume the event (which normally should mean it's a no-op). Hm...

> It might be helpful to write up a single test that verifies that Enter is 
> handled correctly for TextField, ComboBox, DatePicker, and Spinner. I'm 
> working on a PR to clean up ESC handling and it was useful to write up a 
> single test covering all of these controls. (I want to ensure that the user 
> can always uses ESC to close a modal dialog even if the TextField has a 
> formatter assigned to it. This isn't working currently.)

Yeah, I guess I didn't look deep enough. 
> 
> Why doesn't Spinner work the same as ComboBox? For a ComboBox the client 
> registers the action handler directly on the ComboBox instead of the combo 
> box's editor.

Very good point; when I was looking at the code I was quite confused with the 
way the editor property works -- it being read only seemed quite useless, 
except if that's the only way to register an action handler.  It looks 
suspiciously like a poor design.

-------------

PR Comment: https://git.openjdk.org/jfx/pull/1629#issuecomment-2463704866

Reply via email to