[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
- From: Jonathan Borden <email@example.com>
- To: "Bullard, Claude L (Len)" <firstname.lastname@example.org>,Tim Bray <email@example.com>, firstname.lastname@example.org
- Date: Wed, 11 Jul 2001 16:53:02 -0400
Bullard, Claude L (Len) wrote:
> OTOH, I stand by the comment below. For the
> long haul, SGML is a safer better bet. Safety and convenience
> are sometimes uncomfortable bedfellows as anyone who keeps
> secure data on a Palm unit they leave at the airport finds out.
Len, you certainly are entitled to your opinion. At the end of the day, the
most significant advantage(IMHO) that XML has _over_ SGML, is that XML is
defined in EBNF. Let me say it again: XML is defined in EBNF. The importance
of this must be understood.
In terms of longevity: how many working copies of SGML parsers do you expect
to be available 10 years from now? Who do you expect is willing to write new
SGML parsers for new platforms such as .NET, i.e. in languages such as C#,
or Python (or Java for that matter)? What makes SGML so safe a bet ... that
it is an ISO standard? There is a long long long long way to go from reading
the ISO standards to producing software that is capable of processing SGML.
Suppose the XML specification were etched in stone and suppose the entire
inhabitants of the planet earth were to move a million light years away at
some distant time in the future. If an extraterrestrial with sufficient
facility in logic and languages landed on earth and were able to decipher
the specification (presumably the EBNF would be the only part that made any
sense :-), an XML parser could be written in ... minutes (modulo those
nastly little WFC and VCs)
Or a more down to earth example, suppose we give the most brilliant (but
non-English speaking) young Khmer developers the XML specification with
instructions only on how to differentiate a production from its text
description, again modulo WFC and VCs, I'd wager a parser could be written.
Now if we could only teach a cockroach to parse XML we'd have real
> -----Original Message-----
> From: Tim Bray [mailto:email@example.com]
> At 08:17 AM 11/07/01 -0500, Bullard, Claude L (Len) wrote:
> >XML is designed for a short haul timespan.
> >When you want to design for the long haul,
> >go back to the SGML parent and work from
> Len's a person who usually combines reasonable comments
> with amusing, well-crafted, rhetoric. I'm not sure why
> he's veering into this kind of flamebait these days. -Tim