Lists Home |
Date Index |
Because ID is not the only one of these and at
some point most of the common DTD defaults will
be declared in the roots of documents and so
are in effect, just the value pair aspects of
the DTD inside the pointies. Say,
xml:href, xml:src, etc. Is there a limit here?
Yes, I understand Chris Lilley's post to the
TAG list. I am asking the same question asked
when the reserved XML namespace was proposed:
where does it stop being a convenience and
become THE way to declare the architecture?
The exceptions made to enable consistent
processsing but avoid DTDs in the name
of well-formedness seem to invalidate the
concept of well-formedness. Why not roll
over and use the existing mechanisms? Given
as you say that nearly all vocabularies need
this is an admission that well-formedness is a
limited notion based on a flawed understanding
of how markup systems interoperate reliably. Admit
that when one has to send or share semantically
loaded information, one sends the DTD or schema.
These exceptions are not "low hanging fruit" or
"easy victories"; they are band aids.
From: Tim Bray [mailto:firstname.lastname@example.org]
Simon St.Laurent wrote:
> While xml:id strikes me as a necessary evil, at this point I strongly
> emphasize the necessary. As I move back into hypertext (what I came to
> XML for in the first place), the need for a reliable ID mechanism seems
> to override the protests that the internal DTD subset is adequate.
I can sympathize with the xml:id idea, except for that every existing
deployed XML vocabulary I've ever seen without exception has an ID
attribute, and the name of the ID attribute is "id". Why not just roll
over, give in, and say that absent other information, if "foo" is in XML
then foo#bar means <any-namespace:any-tag id="bar">? -Tim