Re: Using your own DTD (was Re: %flow and headers and address)

Murray Altheim (
Mon, 30 Sep 1996 20:01:44 -0500

Message-Id: <v02140b01ae7613c7c0e6@[]>
Date: Mon, 30 Sep 1996 20:01:44 -0500
To: Foteos Macrides <MACRIDES@SCI.WFBR.EDU>
From: (Murray Altheim)
Subject: Re: Using your own DTD (was Re: %flow and headers and address)
Cc: Peter Flynn <>,

Foteos Macrides <MACRIDES@SCI.WFBR.EDU> writes:
> (Murray Altheim) wrote:
>>Foteos Macrides <MACRIDES@SCI.WFBR.EDU> writes:
>>>        I was assuming that Peter would offer his composite DTD for
>>>public discussion by all interested persons, and would amend it
>>>toward some consensus of what would be most useful to as many of
>>>those interested persons as possible, as well as making his own
>>>judgements based on his expertise in SGML.
>>>        A lot of what will be "restored" in it was designed to
>>>degrade gracefully for clients which do not support it.  But
>>>using it is presently problematic with respect to validation
>>>if you also use markup developed since, and not in, the (expired)
>>>HTML 3.0 DTD.
>>If you guys
>        What's this "you guys"??!!??  It should be discussed and
>refined by everyone who has in interest in it.  That includes you,
>too, doesn't it?

Well, I meant that a little tongue-in-cheek. There's a certain irony for me
in anyone trying to create a "standard" HTML DTD anymore, given W3C's
gradual movement into the role of "standards body" (as defined commonly in
the press, the industry, etc.). I sometimes wonder if you and I and a few
others around here aren't all sitting on a large mushroom with a hookah,
and just don't realize it.

I feel I'm certainly in the minority in continuing to look at the IETF as
the only relevant standards body for the Internet (well, including ISO,
ANSI, etc. as appropriate.) There's a difference between a standards body
and a vendor consortium that wants to be a standards body, but that seems
to be a rather muddy distinction for some.

OTOH, an "fully-SGML-functional" Web would allow all manner of DTDs, so use
of a non-standard HTML DTD is fine by me. It's just hard to get anyone
outside of one's organization to use a common DTD if there's no weight
behind it (witness the already-successful W3C "HTML 3.2", even though it's
not yet finalized). And we're *still* a little premature in expecting
browsers to actually pay attention to DOCTYPE.

>>                decide to go off and rebuild the world outside of W3C, you
>>should at least consider what you're doing as a/the possible proposal you'd
>>submit to the IETF in reopening an HTML working group. Then the product of
>>what you're working on would actually become (theoretically, at least) a
>>recognized standard. Hey, and you might get participation by a lot of
>>people willing to provide valuable feedback (such as Earl).
>        It might be better to take one step at a time.  What's clearly
>needed is a composite DTD that includes and coordinates what has already
>been formally proposed and/or implemented in at least one client.  That
>includes essentially everything in the (expired) HTML 3.0 DTD, and a lot
>more at this point.

Up until this Spring, Glenn Adams and I here at Cambridge Spyglass NEE
Stonehand maintained an "HTML 2.X" DTD, which was just such a DTD. We
haven't (since purchase by Spyglass) had the priority on maintaining it, so
it is no longer current (no frames, for example). Oh, it never did math,
although I did have an internal version at one point with the math from 3.0
as a module.

I have made the modules of the modular 2.0 DTD (which is functionally
identical to RFC 1866) available, so if Peter wanted to use that as a
basis, I'd be happy to lend what time I have available in assisting his use
of that material. I believe that would be much preferred to creating a
monolithic DTD, as then public submission of DTD fragments would be much
simplified, and given that James Clark's (no, not the Netscape James Clark)
latest SP code allows for URLs in the right side of a catalog file, the DTD
could even be created on the fly from modules on disparate servers.

One of the long-term failings of the initial HTML standards effort may have
been the belief that a single HTML specification could provide all the
functionality needed by various users of the Web (if anyone ever believed
that). I think it was the Balkanization article that best pointed out that
there is more than one audience for this technology, and that the efforts
of the W3C, being closely tied to the vendors, won't necessarily provide
the functionality needed by the academic, arts, scientific, etc.
communities. This is one avenue that I hope a new working group would
attempt to ameliorate.


     Murray Altheim, Program Manager
     Spyglass, Inc., Cambridge, Massachusetts
     email: <>
     http:  <>
            "Give a monkey the tools and he'll eventually build a typewriter."