[
Lists Home |
Date Index |
Thread Index
]
John Cowan <jcowan@reutershealth.com> wrote:
| On Wed, Sep 18, 2002 at 08:03:08PM +0000, Arjun Ray wrote:
|
|> IDN public IDs? Nope.
|
| And a good thing too, as they break FPI stability.
Yes, but the SGML Declaration for XML (at least the known one, Annex L)
says FORMAL NO. Public Ids in XML are not FPIs except by *convention*.
People who rely on public ids in XML documents to be FPIs are on thin ice
without knowing it. That convention could eaily have been something else
altogether. At the time ('98), IDN seemed like a decent option, as a
naming convention with URI characteristics.
| There have been four owners of spam.org that I know of, and no guarantee
| that any URI or FPI assigned by one owner will be respected by another.
Who owns FPIs with unregistered owners? Wouldn't the same argument apply?
Ultimately, FPI stability is a matter of convention, enforced through
catalogs, not direct dereferencing.
|> Why no special syntax (like say a reserved name, xmlid) for IDs
| Well, it isn't really a magic ID attribute that's wanted, it's a method of
| declaring the name of a magic ID attribute.
Which would still call for magic syntax, which ultimately boils down to a
magic name :-)
<?xmlid foo?>, anyone?
|
