Re: SAX 2.0 enhancement proposal

[David Brownell <david-b@pacbell.net>]

> I think the open question

Seems to me there are a reasonable number of them by now,
more than just one, scattered across at least two W3C specs.
(Infoset still not a REC?  Hmm...)


>     is whether or not a system identifier
> {can|must|should|may|must not} be made absolute the instant that it's
> returned by the tokenizer. I don't think the spec is clear on that
> point.

That's pretty similar to the issue of becoming clear on the distinction
between an "XML Processor" and the application it's connected to,
which has resisted clarification for several years now.  Is there any
place the XML spec even talks about tokenization in terms of the
boundary between application and processor?  I don't recall it doing
any of that.  There's not even a requirement in the spec to support
augmenting the "fetch data for URI" primitive, or providing functional
requirements for such a processor function.


> There's absolutely nothing wrong with PIs in the internal or external
> subset to provide semantics. They're every bit as valid and reasonable
> (though perhaps not as interoperable) as declarations.

That clause doesn't restrict itself to the DTD; quite the opposite, with
that example using elements.  Though I'd disagree about "absolutely"
nothing wrong when creating interop problems.

- Dave