Re: A note on <bind /> (similar to <group />)

> group has uses other than setting context node, which means that it does not
> need to have ref, model, or bind.

Agreed it does not need to have an context node. I never said they must 
have one. All I am saying is having a context node changes the behavior

> 
> You may find it more explicit to say <group ref="." /> and it is a good idea
> for authors to write in some cases.
> But explicitness is in some cases verbosity, and in this case the recursive
> composition rules already say what to do.

my point has been
   <group> and  <group ref=".'>  is **NOT** the same
(and that is why I like the explicitly)

In the case where group is bound to a instance node (explicitly) the 
model item properties on the node affect the state of the control bound 
control.

Here is an example, hope this clarifies what I am trying to say

Assuming you have the following CSS rules

    group:readonly {color: red}
    group > group:readonly {color: blue}

and a document fragment that is
    <group ref="a">
        <div/>
        <group/>
    </group>

and another document fragment that is

    <group ref="a">
        <div/>
        <group ref="."/>
    </group>

Also supposing node "a" is has model item property readonly.

I think the user should see two different behaviors.
In the first case the user should see the entire fragment displayed in 
red, but in second case the inner group should be displayed in blue.

> 
> The issue of <label>, <output>, <message>, <help>, <hint>, <alert> is, I
> think, separate from <group> which is already decided and <bind>, which I
> point out is underspecified.

Bind may be underspecified, but I am just questioning the 
interpretation. (As I said not specifying a nodeset may not be the same 
as specifying nodeset=".")

PV

Received on Thursday, 5 February 2004 16:08:55 UTC