Re: URL better than FPI

Arjun Ray wrote

> Well, under the -ahem- rules, that would be the SGML system's fault.

Yes, as you probably saw by now, Russell pointed out the FORMAL NO as
well.

> Yes, but there is no need to put the system address in a *document
> instance* if the public identifier is there already.

But you mean no need in an ideal world. Since the XML spec mandates that
you do this, you can't mean no need in XML.

> When we're
> talking about XML and the Web, I can't imagine that it woudn't or
> couldn't be normal to assume that the canonical name *will* have a
> system address *necessarily* associated with it.

I can easily imagine conforming XML parsers that ignore the public
identifier if it is there and just use the system identifier, and will
fail if the system identifier is an HTTP url and I'm using my laptop on
the train. This is what I was refering to that you asked me to clarify

> But I don't see why a validating parser would necessarily
> fail just because a http: URL had to be dereferenced.  Could you
> clarify?


It will fail to validate if it can't find the dtd, and there is no
reason why it should be able to find the DTD if the SYSTEM identifier is
a URL which can't be reached if my machine is not connected.

To give a real example, I don't think xp supports either catalogues or
caching of previously fetched files, so if I want to apply xsl
stylesheets to xhtml documents using xp/xt They had better have a system
id that points to a file on my local system. Under the suggested change
to the XHTML draft such a file would no longer be conforming xhtml.


David

Received on Tuesday, 22 February 2000 04:47:16 UTC