RFR: 8306121: Scene not rendered initially when changing scenes after fix for JDK-8296621 [v18]

[Thiago Milczarek Sayao]

On Tue, 6 Jun 2023 14:03:35 GMT, Kevin Rushforth <kcr at openjdk.org> wrote:

>>> I ran an automated test job using your latest patch and all of the tests now pass.
>>> 
>>> > Maybe it's possible to do a Java fix where the new Scene without size inherits the old Scene size? This way I think it possible to remove platform specific code such as the one I did and `_updateViewSize`.
>>> 
>>> Hmm. That might be problematic. In any case, I don't want to go down that route to fix this bug.
>>> 
>>> I have two questions on the current version of the fix:
>>> 
>>>     1. Can you explain why the two tests had to be changed? Unless the tests in question really were buggy, and just getting luck before, the changes suggest a change in behavior that applications might notice and be bothered by.
>>> 
>>>     2. Are you sure that on Linux, there are no cases where `_updateViewSize` needs to be implemented? Did you consider just leaving it as-is or would that have failed?
>> 
>> I changed it to use `_updateViewSize` so it's compatible with other platforms.
>> 
>> I'm gathering some information to answer about the test changes.
>
>> I changed it to use `_updateViewSize` so it's compatible with other platforms.
> 
> As it turns out, I was looking at your incremental diff and (mistakenly) thought that `_updateViewSize` was in the current mainline code base and that you had removed it, which is why I asked about it. I now see that what happened is that the initial version of your PR had added it, and a later version had reverted that addition. And your most recent version added it back in. I don't mind either way, so if you think it's cleaner to implement `_updateViewSize` on Linux (even though today it's only implemented on Windows), that's OK. If you want to revert this latest addition of that method, that's OK, too.
> 
>> I'm gathering some information to answer about the test changes.
> 
> Thanks.

@kevinrushforth Got it.

Testing uses `StubToolkit` and not `QuantumToolkit` which before [JDK-8296621](https://bugs.openjdk.org/browse/JDK-8296621) called `requestFocus()` inside `WindowStage` which is part of `QuantumToolkit`. So focusing were never taken into account while testing.

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

PR Comment: https://git.openjdk.org/jfx/pull/1110#issuecomment-1585792601