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

> I'd say that although it's a good candidate for clarification, the spec
> already allows it.

I have to disagree. The spec does not know it. Spec simply does not 
state what should happen in such a case (on bind). The spec is actually 
interpreted the other way in the UI (see below).

> 
> The spec actually doesn't say whether @nodeset is optional or required
> on xf:bind [1]. However, the schema (which is normative) *does* say that
> @nodeset is optional [2]. This therefore leads to a number of possible
> arrangements, many of which I believe are very useful.

While I agree what you state is useful, but this is not a clarification 
it is a new functionality.

You have just stated one aspect of it, now moving to UI layer

   <input appearance="minimal">
   </input>

When input does not have a "ref", that simply means it does not bind to 
any node. This is useful, if the UI wants to provide a field like a 
"scratch pad" for the user (does not necessarily want to bind to any 
item in the instance)

Similarly
    <input ref="..">
      <label>Label</label>
    </input>

In the above case "ref" s not provided by label does *NOT* mean it is 
bound to a node. It simply means binding to none (and hence use the 
value inside). So is output and ...

PV

Received on Thursday, 5 February 2004 12:07:59 UTC