<Swing Dev> <AWT Dev> Review request #3: 6852592 (revalidate() must be smarter)

[Artem Ananiev]

Anthony Petrov wrote:
> On 07/23/2009 06:54 PM, Artem Ananiev wrote:
>> This looks like a good idea.
>>> And what about the AWT ScrollPane? Shouldn't it return true as well?
>>
>> As well as this one.
> 
> Legacy AWT applications are supposed to invoke the validate() method on 
> the top-level component. After the proposed change the invalidate() will 
> stop invalidation on the scroll pane, making the subsequent validate() 
> call a no-op because the top-level component will remain valid.
> 
> Not only is the change backward-incompatible, but it also tends to leave 
> parts of the component hierarchy invalid. That wasn't important before, 
> but is important now (and that is the aim of this particular fix, by the 
> way.)
> 
> Since:
> 
> a) almost nobody develops pure AWT applications our days, thus making 
> this change unimportant,
> b) this change breaks backward compatibility with existing applications,
> 
> I'm against making the AWT ScrollPane a validate root. Please provide 
> more justification if you strongly feel it is required.

HW/LW mixing has already introduced some backwards incompatibility into 
our code - this is why this discussion appeared. Making SctollPane a 
validate root is at least consistent with the rest of the code (AWT and 
Swing).

Thanks,

Artem

> -- 
> best regards,
> Anthony
> 
>>>> http://cr.openjdk.java.net/~anthony/7-23-invalidate-6852592.2/