<Swing Dev> [PATCH] 6463712: JSpinner forwards events from old model

Oleg Sukhodolsky Oleg.Sukhodolsky at Sun.COM
Thu Jul 12 20:03:17 UTC 2007


Hi Roman,

Roman Kennke wrote:
> Hi,
> 
>> So just to let you know what is going on, when the bugfix has been  
>> applied to a local workspace, we generate a "webrev" which shows the  
>> revisions between the old and new files (a kind of diff, but with a  
>> lot more ways to view the changes). This then gets sent to two or  
>> more reviewers, even for trivial bugs like this one. Sometimes these  
>> reviews can sit in a queue for a bit, if the reviewers are hurting  
>> for time.
>>
>> After both reviewers give it the ok, I can put the bug back. It will  
>> then take several weeks before the swing workspace is integrated back  
>> into the main workspace, goes through testing, and pops out in a build.
> 
> Eh. You really want to scare off voluntary contributors, would you? I
> mean, requiring this procedure from paid developers is one thing, but
> voluntary developers often do develop for fun (or some other kind of  -
> let's call it 'social reward' -). And a procedure like this really
> useful for bringing motivation down to zero. I mean, this patch is
> perfect, it's obvious, it even comes with a testcase. And I really
> really doubt that such a procedure adds much to the overall quality of

I agree with you, but this is a bureaucracy we have to live with :(  It 
will be great if we could process contributions faster, but we can not 
:( at least for now)

> the code. Or how would you explain, for example, the crazy code in
> UIManager.initialize() ;-) (I've seen a couple of places in OpenJDK,
> where a non-visible method of another package is called via reflection.
> This is not only evil, it's dangerous too. What if such a method is
> renamed or removed? And after all, it's not even necessary to write such
> code in the first place. A little refactoring of things should help
> too).

I do not know what places you are referring to, but  in most cases I 
know this is because we can not (or do not want to) provide public 
method, because this would be an (unnecessary) API change.  As contrast, 
Swing number of public methods which are only public because they should 
be accessible from laf, but at the same time regular developer should 
not use them.  And this is also bad :(

> Just my 2 cents.

+2c from me :)

Oleg.

> 
> /Roman
> 



More information about the swing-dev mailing list