Lists Home |
Date Index |
> From: John Cowan [mailto:email@example.com]
> Michael Brennan wrote:
> > This would mean extending the use of labels beyond simply labeling
> > resources and locators within an extended link, but this
> seems to me to
> > be a very logical extension. I don't see why we need IDs
> for this versus
> > a simple label explicitly for the purpose of linking.
> Because IDs are fairly widespread in documents, being used for
> many purposes, but xlink:label attributes are not. That makes it
> annoying to link to existing (unchangeable) documents.
Yes, but xml:id attributes and xml:idattr attributes are not in any existing
documents. My point was to use xlink:label in lieu of these other proposals,
rather than introduce a new special attribute in the xml namespace just for
the purpose of linking. (That's what I meant to say. I shouldn't have used
"ID" in all uppercase in my post.)
And these proposals for a special attribute do not address the issue of
IDness of an attribute any better than using xlink:label -- without a PSVI,
they are just a special name for an attribute, just like xlink:label.
So my question stands (with the clarification that I meant "id" in
lowercase, in specific reference to the proposals for a special attribute
name). The use case of supporting a simple fragment identifier without
relying upon a PSVI seems reasonable to me. But these proposals for a
special xml:id or xml:idattr attributes strike me as completely superfluous
proposals that don't solve any problems that can't be solved by existing
constructs, such as using xlink:label.
(Note to self: brush up on communication skills.)