> Terry Allen writes:
> > Dave Morris writes:
> > >FINALLY ... a small change to the HTML 2.0 draft ... it seems to
> > me that a critical element missing is a formal specification of
> > exactly where the DTD should be obtained from when it IS specified
> > with an HTML document. This requires a network oriented solution
> > such that an SGML user agent which wants to use the DTD to control
> > parsing of the HTML document can find the thing without magic.
[...]
> This looks like a prima facie case for <link rev="dtd" href="...">
That won't work -- an SGML parser can't resolve the <LINK>
until it parses the <HEAD>, which it can't do until it
has the DTD.
One approach is to use a URI for the DTD's system identifier:
<!DOCTYPE HTML SYSTEM "http://www.w3.org/pubtext/html-4.dtd">
It's better to use a PUBLIC identifier, but there is (so far)
no way to automatically resolve them. I vaguely recall
a discussion of an "fpi:" scheme for URNs, which would use
the usual URN resolution mechanism. So a browser that saw:
<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML Level 4//EN">
would query the URN resolution service for the URN
fpi:-//IETF//DTD HTML Level 4//EN
and it would return one or more URLs by which the DTD
could be retrieved.
SoftQuad Panorama has surely solved this problem;
can anyone comment? (I don't have a Windows box handy
so I've been unable to read the Panorama docs.)
--Joe English
joe@art.com