>>>>> On Thu, 31 Jan 2013 09:31:52 -0500, Leonardo Uribe <lu4242_at_gmail.com> said:
LU> - At start it says something like this:
LU> "... This renderer is intended to be used with an arbitrary markup
LU> element that has
LU> no decoding. ..."
LU> But later says this:
LU> "... If the component has a manually declared, not auto-generated
LU> clientId, or if the
LU> component has behaviors attached to it, render the clientId as the
LU> value of the "id"
LU> attribute. ..."
LU> In few words, first it suggest that it has no decoding, but later it
LU> says that the
LU> component has behaviors (or client behaviors) attached to it. By definition=
LU> ,
LU> client behaviors has decode behavior, so the first line is
LU> incorrect.
I've changed it to "This renderer is intended to be used with an
arbitrary markup element whose decoding needs can be entirely described
by the section "General notes on decoding" in the RenderKit overview".
LU> How to identify if a property has a related client behavior? In theory
LU> we can define an
LU> specific set of properties (onclick, onmousedown, ....) that are
LU> related to the renderkit,
LU> and maybe that will work for the current spec, but the question that
LU> still is on the air
LU> is if we need a more general way to define such kind of contracts
LU> (ActionSource2,
LU> value change, client behaviors ...). I have a component and I want
LU> that the component
LU> implements ActionSource2 contract, but in a dynamic way, without the
LU> need to write
LU> a component class.
I suggest we can safely leave that vague for JSF 2.2 to allow the widest
possible breadth of implementation choies while still maintaining cross
vendor compatibilily.
Ed
--
| edward.burns_at_oracle.com | office: +1 407 458 0017
| homepage: | http://ridingthecrest.com/