(no subject)

Mike Meyer (mwm@contessa.phone.net)
Tue, 19 Mar 1996 23:30:42 PST

Subject:  Re: 
In-Reply-To: <199603200423.UAA21911@server.livingston.com>
From: mwm@contessa.phone.net (Mike Meyer)
Date: Tue, 19 Mar 1996 23:30:42 PST
Message-Id: <19960319.76E35B0.152EC@contessa.phone.net>
To: www-html@w3.org

> Something else that has never been answered in any satisfactory manner -

Well, I already answered it, and nobody complained at the time.

> An author uses "HTML 3", Netscape, M$IE, and Bob's Browser markup tags in
> their documents.  None is a superset of all of the others.
> What is the DTD?

It being a new type of document, it needs a new DTD. Whoever creates
the document type gets to create the DTD.

An intelligent DTD author will use this as a chance to verify that all
documents will be well-behaved, and limit shared features to the
intersection of the sets, make sure that no shared tags with different
semantics creep in, and that all proprietary tags are used in such a
way that they don't lose information if ignored.

In reality, people writing such a tag stew don't validate their
documents against a DTD, so putting a DOCTYPE statement
that lists a DTD is basically a lie, so you're best not putting one

If you really want a DTD that validates anything you feed it that's
syntactically correct, I've seen one. Since it provides no structuring
whatsoever to the document, it's pretty much worthless.

> Good browsers deal with this fine - they just don't use tags that they
> don't understand.  And that is most browsers I've ever used.  No need for
> a DTD.
> A pure SGML browser would choke on this with no DTD to follow.

So? You're not feeding them SGML. Of course they choke. This is no
more interesting than the fact that pretty much every HTML browser -
including all the popular ones - choke if you feed them HTML documents
that includes any number of real SGML constructs.