On Sat, 10 Mar 2007 14:43:44 +0100, Elliotte Harold
<[EMAIL PROTECTED]> wrote:
What are those of us who wish to use XML tools on our documents supposed
to use?
Set the 'xml_tool.ignore_doctype' property to 'true'. I'm sure most XML
libraries have this (or a similar) option. At least most of the ones I've
worked with has had them, and not having a DTD does not mean that you
can't be strict about what you consume. There will be HTML5 XSD and RNG
schemas available (I would actually think they exist already) for you to
use in your application once that day comes. DTD is dead. Deal with it.
We will need a real DTD at some point, to declare the entities if
nothing else.
Then define your own internal subset. You don't need a full DTD for the
entities. Just use XHTML 1.1's if you must:
<http://www.w3.org/TR/xhtml-modularization/dtd_module_defs.html#a_xhtml_character_entities>
We will not be able to use <!DOCTYPE html>.
If you can get your parser to ignore it, sure you can.
Possibly this could be two-fold. E.g there could be both
<!DOCTYPE html>
and something like
<!DOCTYPE html PUBLIC "-//WhatWG//DTD XHTML 5.0//EN"
"/DTD/xhtml5.dtd">
That's an option too, but it would be confusing for authors.
I know some browser-centric folks here just hate DTDs and schemas of any
kind; but we will need them, even if the browsers don't. We will create
and use them, even if there's no normative DTD in the spec.
I don't hate schemas. I work with them almost every day. I also consume
XHTML in various forms without ever having to touch a DTD. I'll say it
again; DTD is dead.
One thing that's struck me in working with the spec over the last few
days is just how hard it is to follow the various content models, and
how much simpler most of them would be to read if they were described in
a RELAX NG schema or a DTD.
That I can agree with. A compact Relax NG schema would definately be
useful to codify (and thus clarify for many of us) the lengthy text that
describes the conformance requirements. The schema should only be
informative, of course, but it could even prove to be useful in the
normative part of the spec, since having to think in RNG terms often makes
you find bugs in the langauge you wouldn't see otherwise.
The Atom Syndication Format specification uses RNG in this style and I
think it works amazingly well:
<http://atompub.org/rfc4287.html#schema>
--
Asbjørn Ulsberg -=|=- http://virtuelvis.com/quark/
«He's a loathsome offensive brute, yet I can't look away»
