On Thu, 21 May 2026 14:13:51 GMT, John Hendrikx <[email protected]> wrote:

>>> I think however we need a more compelling reason to get this merged (like a 
>>> broken test in current FX related to some complex property interaction), 
>>> otherwise I don't see this happening.
>> 
>> I see. Is that possible to create?
>
>> > I think however we need a more compelling reason to get this merged (like 
>> > a broken test in current FX related to some complex property interaction), 
>> > otherwise I don't see this happening.
>> 
>> I see. Is that possible to create?
> 
> What I mean is, if there is a real bug that this PR would solve, then it is I 
> think more likely to be integrated. Currently the bug this PR solves is 
> recursive changes involving a ChangeListener that does re-registration of a 
> listener (removing it from `old` and adding it to `new`) -- the bug would 
> appear if the property is modified again before all its listeners were 
> notified.
> 
> So in theory, it could be adding a change listener (as the user) to one of 
> the more obscure properties that Node has, and having that change listener do 
> something that triggers a nested change. If that property also has a 
> ChangeListener on it (from FX itself) that does the re-registration pattern, 
> then it would likely get the framework in a bad state (old listener not 
> removed, or listener added twice).
> 
> To find one, one could look for places where this re-registration pattern is 
> used (I think I see it often used on scene properties), and then see if 
> adding a user listener on the same property (or related one) could interfere 
> with this process if it triggered a recursive change.

@hjohn can you check GHA failures please?

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

PR Comment: https://git.openjdk.org/jfx/pull/1081#issuecomment-4919845023

Reply via email to