dev@javaserverfaces.java.net

Re: [REVIEW] Restore view optimization

From: Adam Winer <adam.winer_at_oracle.com>
Date: Tue, 19 Sep 2006 14:59:10 -0700

Ryan Lubke wrote:
> Adam Winer wrote:
>> Ryan Lubke wrote:
>>> Adam Winer wrote:
>>>> Ryan,
>>>>
>>>> Doesn't this break existing third-party StateManager
>>>> implementations?
>>> The way I saw it was either way, the default behavior could be
>>> replaced (if the third party provided their own Lifecycle that
>>> didn't decorate the default).
>>
>> I don't understand... if I wrote my own StateManager
>> (and I have), it's not OK for a new rev of JSF to just
>> break "binding" in all my pages. It's not OK to move
>> responsibility from one class to another in the RI
>> if either of those classes are replaceable.
>>
>> The issue isn't "can a third party be made to work",
>> it's "will a third party continue to work", and also
>> whether that code can be written portably (e.g.,
>> against MyFaces and the RI).
> Point taken. Would the check in RestoreViewPhase
> to bypass the doPerComponentActions() if the StateManager
> is StateManagerImpl be sufficient then?

Probably... the only issue left would be what happens
if someone is using StateManagerImpl but a custom
lifecycle, in which case "binding" would get double-evaluated.

-- Adam