Yeah, Dan made the valiant effort here but I don't think it's practical. I
don't feel bad asking users to change import statements to pick up bug
fixes.

It would be great if we can get this into the 2.3 rc.
On Mar 24, 2011 7:05 AM, "Daniel Rice (דניאל רייס)" <[email protected]> wrote:
> I spent a few days attempting this -- it's not so simple. For example:
>
> 1) Enums can't extend Enums. Several public types contain nested Enums.
> 2) If a type oldA extends newA, and a method in it returns an instance of
> oldA, you cant simply delegate to the implementation in newA since that
will
> return a newA, which is not an instanceof oldA. So you need to write
> wrappers for everything.
> 3) It's unclear to me how inheritance works on annotations
> 4) Generators will have to be modified to work off the different
annotation
> types
>
> I think there were other problems as well that I can't recall at the
moment.
> My belief is that continuing with this approach would have polluted the
new
> code base with special cases to handle parallel enums and annotations.
>
> I spoke to Ray about it and he agreed to the plan of leaving a snapshot in
> the old location and trying to remove it in the next release.
>
> On Thu, Mar 24, 2011 at 8:57 AM, <[email protected]> wrote:
>
>> Hollow out the to-be-deleted classes to avoid the massive duplication of
>> code going on in this patch. The old types should extend the new types
>> and where that isn't possible, the old type should be rewritten as a
>> delegate.
>>
>> Anything in the rebind or **/impl packages can be a straight move.
>> They're not public types.
>>
>>
>> http://gwt-code-reviews.appspot.com/1383808/
>>
>
> --
> http://groups.google.com/group/Google-Web-Toolkit-Contributors

-- 
http://groups.google.com/group/Google-Web-Toolkit-Contributors

Reply via email to