On Fri, 20 Mar 2026 12:11:07 GMT, John Hendrikx <[email protected]> wrote:

>>> > was this change included? #2007 ?
>>> 
>>> I merge the latest master when reviewing any PR that's been out for too 
>>> long, so yes.
>>> 
>>> My concern is a high risk of regression (even if the regression is 
>>> cosmetic): even if we fix this particular failure, other apps might fail in 
>>> a similar fashion somewhere else.
>> 
>> I tested this change in a number of applications, including ControlsFX and 
>> custom components and there really is no problem. So at least I'm very 
>> confident about this change, especially since the previous check tried to do 
>> the very same thing.
>> 
>> My guess is that `ToolBar` behaves bad for whatever reason and there is 
>> probably a bad implementation in the `Skin`.
>
>> > was this change included? #2007 ?
>> 
>> I merge the latest master when reviewing any PR that's been out for too 
>> long, so yes.
> 
> Thanks, the problem looked similar.  I will see if it is reproducable on 
> Windows, or if this is again a Mac specific issue.
> 
>> My concern is a high risk of regression (even if the regression is 
>> cosmetic): even if we fix this particular failure, other apps might fail in 
>> a similar fashion somewhere else.
> 
> We could introduce a property to switch it back off, so we can get this 
> tested in a wider audience, as I do agree that it will be hard to predict the 
> exact effects given that the old behavior which did a final unnecessary 
> layout each time has part of FX for a long time.

@hjohn I don't understand why you want to commit to a known regression.

Let's find out why the regression happened and then we can try with a system 
property, that sounds like a good idea.  For reference, there is nothing 
special in the monkey tester: the stock `MenuBar` is inside of a `BorderPane` 
which is the root of a `Scene`.

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

PR Comment: https://git.openjdk.org/jfx/pull/1945#issuecomment-4098955342

Reply via email to