HTML, SGML and Tag Minimization (was: </P>)

Paul Prescod (papresco@calum.csclub.uwaterloo.ca)
Sun, 18 Aug 1996 12:53:47 -0400


Date: Sun, 18 Aug 1996 12:53:47 -0400
Message-Id: <199608181653.MAA20122@calum.csclub.uwaterloo.ca>
To: Thomas Breuel <tmb@best.com>, www-html@w3.org
From: Paul Prescod <papresco@calum.csclub.uwaterloo.ca>
Subject: HTML, SGML and Tag Minimization (was: </P>)

At 04:37 PM 8/17/96 -0700, you wrote:
>I think it's largely because of the SGML heritage.  SGML had some nifty
>ideas for structural markup, but their realization is less than
>perfect.  

I agree that SGML has big holes, but I don't think that allowing markup
minimization is one of them.

>One big problem is that SGML markup isn't all that convenient
>for manual authoring; 

What meta-markup language do you propose is simpler?

>to counteract that, all sorts of rules for
>letting you abbreviate constructs were introduced, but those make
>automatic processing and error detection/recovery harder and cause
>confusion like "do you need a </P> or not".  

If you always put in all of the tags, there will be no confusion. The
problem is that the HTML community wants quick-and-dirty markup but robust
validation and processing. I think SGML is amazingly accomodating in
allowing this. 

In short:

If the tag minimization rules are too complex, don't use them. 

If the markup is too verbose, use tag minimization rules, but don't complain
that they are "too complex" unless you have a simpler proposal.

>SGML tries to be a kind of
>programming language, and sadly it seems to be lacking many of the
>syntactic features that make programming languages easy to use and
>robust.

SGML does NOT try to be a programming language. SGML is a markup language
(or meta-markup language), and goes further than most markup languages in
allowing concise, natural markup.

SGML also allows some nice tools for reusing markup in order to save typing,
but HTML user agents do not know how to process them.

 Paul Prescod