http://cwiki.apache.org/confluence/display/WICKET/FAQs#FAQs-WhatisthefutureofonAttachandonDetachmethods%3F



On 3/9/07, Igor Vaynberg <[EMAIL PROTECTED]> wrote:

the power of the wiki is that anyone can contribute :)

-igor


On 3/9/07, Bruno Borges <[EMAIL PROTECTED]> wrote:
>
> Cool Igor... Now I see the problem and what's coming next :)
>
> PS: this explanation should be wiked !! =)
>
> []'s
>
> On 3/9/07, Igor Vaynberg <[EMAIL PROTECTED]> wrote:
> >
> > we are trying to consolidate the methods. we have a bunch of
> > internalOnAttach/internalAttach/attach/onattach methods. its a big
mess.
> > what this refactor does is give you one method you can override -
> > onattach()
> > but forces the call to super.
> >
> > doing it like it has been done doesnt work. users assume onattach() is
a
> > template method, they can override and not have to call super - but
this
> > fails if you go more then one method deep!
> >
> > if i create a custom component and do something in onattach(), then
the
> > user
> > subclasses it and they do something in onattach() and dont bother to
> call
> > super() they will break my functionality. my only choice of action is
to
> > make onattach() final in my custom component and provide yet another
> > template for the user, onattach2() ? this just doesnt scale. better to
> > have
> > a simple and clear contract - onattach and ondetach always require the
> > call
> > to super.
> >
> > unfortunately the only way to do that at this point _and_ keep the
same
> > method names is to do what i did.
> >
> > OT there is a jsr for software defect annotations that includes
> something
> > like @MustCallSuper (forget what its called), that combined with an
apt
> > builder in an ide will make these kinds of contracts very easy to
> enforce
> > at
> > compile time. we are just not there just yet.
> >
> > -igor
> >
> >
> > On 3/9/07, Bruno Borges <[EMAIL PROTECTED]> wrote:
> > >
> > > I want to make a comment about the onAttach/Detach refactor:
> > >
> > > This really can't be done without breaking current overridden codes?
> > >
> > > Isn't possible to convert onAttach to final onBeforeAttach, and let
> > > onAttach
> > > empty so users can override? The same for detach...
> > >
> > > []'s
> > >
> > > On 3/9/07, Igor Vaynberg <[EMAIL PROTECTED]> wrote:
> > > >
> > > > he forgot the onattach/detach refactor which will break a lot of
> > clients
> > > > in
> > > > runtime if they overrode onattach/detach and didnt call super().
> > > >
> > > > so dont worry.
> > > >
> > > > -igor
> > > >
> > > >
> > > > On 3/9/07, Nick Heudecker <[EMAIL PROTECTED]> wrote:
> > > > >
> > > > > > Right now, the differences between 2.0
> > > > > > and 1.3 are also models, validators and generics. I think
that's
> > it.
> > > > >
> > > > >
> > > > > Oh, that's all?  For a second I was worried. :)
> > > > >
> > > > >
> > > > > --
> > > > > Nick Heudecker
> > > > > Professional Wicket Training & Consulting
> > > > > http://www.systemmobile.com
> > > > >
> > > > > Eventful - Intelligent Event Management
> > > > > http://www.eventfulhq.com
> > > > >
> > > >
> > >
> > >
> > >
> > > --
> > > Bruno Borges
> > > Summa Technologies Inc.
> > > www.summa-tech.com
> > > (48) 8404-1300
> > > (11) 3055-2060
> > >
> >
>
>
>
> --
> Bruno Borges
> Summa Technologies Inc.
> www.summa-tech.com
> (48) 8404-1300
> (11) 3055-2060
>

Reply via email to