when i was discussing this with jon i came up with a slightly different idea
there are two issues at work, or two paths
one to get the "root" model, and one to "unwrap" the model.
so we used to have iwrapmodel.getnestermodel() which should have returned
the model that created the wrapper. we used that to unwrap the model.
the assumption has been that unwrapping is always the same as going one step
closer to the root, which with this new componentpropmodel is wrong.
so maybe what we need is a different interface, ICompoundModel.getInnerModel
()
so any imodel can implement icompound and its getinnermodel is the path up
to the root. and this is independent from iwrapmodel.getnestedmodel() which
is a path to unwrapping.
dont know if we need this or if the above makes sense :)
-igor
On 4/6/07, Johan Compagner <[EMAIL PROTECTED]> wrote:
i have thought about this some more.
wat you want is i think is a interface: IIWantParentModel
(setParentModel(IModel))
the problem is when to call it?
For normal models this is easy: getModel() does the checking.
but for ExternalLink that has 2 own models.. should handle that then
internally.
so when we do this:
Object hrefValue = href.getObject();
we should do:
Object hrefValue = testIfParentModelIsSet(href).getObject();
then it works for all models and we don't have anything special.
johan
On 4/5/07, Johan Compagner <[EMAIL PROTECTED]> wrote:
>
> why would the models component have to be null?
> couldn't it be the one with the compound?
> Or be the one with another main model?
>
> But this model (that is still not really a model) is just so that this
> constructor:
>
> public ExternalLink(final String id, final IModel href, final IModel
> label)
>
> can be called with 1 or 2 models that take the value out of a compount
> model??
>
> Just use a property model for that... and do it yourself.
>
> new ExternalLink("id", new PropertyModel("href", XXXX), new
> PropertyModel("label", XXXX))
>
> and XXX can be a compound or another model.
>
> the only thing you avoid here is the XXXX part because you still have to
> do new XxxxModel("label")
> so what is the gain here?
> What we do gain is suddenly models implementations that aren't really
> model implementations.
>
>
> johan
>
>
> On 4/5/07, Jonathan Locke <[EMAIL PROTECTED]> wrote:
> >
> >
> >
> > maybe when a wrap occurs on one of these we can check to make sure
> > the model of the component is null (compound property model is being
> > inherited). we would need to expose Component's check of model ==
null
> > in getModel() through something like isModelNull(). but that would
make
> > it fairly safe.
> >
> >
> > Johan Compagner wrote:
> > >
> > > Then that name of that class is really wrong
> > > So it is a model for something that isn't uses as the real model of
> > the
> > > component??
> > > Now it is just very fuzzy, and if used as normal model you get an
> > > exception.
> > > So thats not something i like to have in the core, that we can have
a
> > > model
> > > that can't be used as a model... (normal model)
> > >
> > > johan
> > >
> > >
> > > On 4/5/07, Jonathan Locke <[EMAIL PROTECTED]> wrote:
> > >>
> > >>
> > >>
> > >>
> > >> Johan Compagner wrote:
> > >> >
> > >> > That one really doesn't work if i understand correctly what it
> > should
> > >> do
> > >> >
> > >> > i guess it should do this:
> > >> >
> > >> > Form form = new Form(new CompoundModel(new Person))l
> > >> > form.add(new TextField(new ComponentPropertyModel("name")))
> > >> >
> > >>
> > >> no. it's for use with something like ExternalLink which takes
> > IModels
> > >> that cannot work with compound models yet. the code above already
> > >> works with just: new TextField("name")
> > >>
> > >>
> > >>
> > >> >
> > >> > AssignmentWrapper(final Component component, final String
> > propertyName)
> > >> >
> > >> > /**
> > >> > * @see wicket.model.INestedModelContainer#getNestedModel
()
> > >> > */
> > >> > public IModel getNestedModel()
> > >> > {
> > >> > return component.getModel();
> > >> > }
> > >> >
> > >> > And this piece of code is really wrong. the getNestedModel should
> > >> return
> > >> > the
> > >> > parent model of the current model and that is not the
> > >> component.getModel
> > >> ()
> > >> > because that is the assigmentwrapper it self!
> > >> >
> > >> >
> > >>
> > >> the component has no model in the case where you use this class
> > >> so getModel will find the compound model via initModel().
> > >>
> > >> use case for this model is ONLY:
> > >>
> > >> final ExternalLink titleLink = new ExternalLink("title",
> > >> new ComponentPropertyModel("link"), new
> > >> ComponentPropertyModel(
> > >> "title"));
> > >>
> > >> where more than one model is being used by a component and that
> > component
> > >> cannot directly support compound propertymodels. the idea is that
> > this
> > >> functionality
> > >> will be folded into PropertyModel so the use case will just be:
> > >>
> > >> final ExternalLink titleLink = new ExternalLink("title",
> > >> new PropertyModel("link"), new PropertyModel(
> > >> "title"));
> > >>
> > >> again, the component has no model. it's referencing the compound
> > >> property
> > >> model
> > >> of the parent.
> > >>
> > >>
> > >> --
> > >> View this message in context:
> > >>
> >
http://www.nabble.com/The-new-ComponentPropertyModel-tf3531176.html#a9859800
> > >> Sent from the Wicket - Dev mailing list archive at Nabble.com.
> > >>
> > >>
> > >
> > >
> >
> > --
> > View this message in context:
> >
http://www.nabble.com/The-new-ComponentPropertyModel-tf3531176.html#a9861817
> > Sent from the Wicket - Dev mailing list archive at Nabble.com.
> >
> >
>
