DIV/ID/CLASS [was: Constructive Review Comments]

Murray Altheim (murray@spyglass.com)
Thu, 16 May 1996 14:58:50 -0500

Message-Id: <v02110103adc12c9ecb4c@[]>
Date: Thu, 16 May 1996 14:58:50 -0500
To: "Daniel W. Connolly" <connolly@beach.w3.org>
From: murray@spyglass.com (Murray Altheim)
Subject: DIV/ID/CLASS [was: Constructive Review Comments]
Cc: dsr@w3.org, www-html@w3.org

Daniel W. Connolly <connolly@beach.w3.org> writes:
>Note that the HTML 3.2 is descriptive, and hence "enhancement
>requests" are out of scope. There is room for discussion about how to
>describe the current state of affairs (e.g. ways to describe things
>like FONT, CENTER, CLASS, ...) but there is no room (in the HTML 3.2
>review process) for enhancement requests (e.g. special entities for

HTML 3.2 is only descriptive of the products produced by the members of
W3C, and reflects pressure from only one segment of the Web community; the
"current state of affairs" is much broader than that. While I agree with
the spirit of trying to keep feature requests in 3.2 down to a dull roar,
stating that there is "no room" for features beyond W3C's "descriptive" DTD
only closes the door to valid features that are simply outside the purview
of W3C. ID and CLASS have been added to various IETF drafts (i18n), non-W3C
and non-IETF DTDs, and have been used effectively to mark up important
meta-information. Pretending that there are unbreakable rules regarding
what must or cannot be in HTML 3.2 is simply stonewalling. Wherever has DIV
been implemented that it was included? You've already broken the rule with
DIV, why not include ID and CLASS?

I agree with comments made in this forum that incorporating both ID and
CLASS into 3.2 will promote document markup for future use with
stylesheets, as well as legitimize current usage in a mainstream DTD. Since
IDREF is not part of HTML syntax, the argument that ID cannot be allowed
because hyperlinks in common browsers don't operate on them is simply a
matter of continuing to proscribe (ie., not specify) IDREFs in the DTD or
application conventions, ie., leaving it as it is now.

HTML reinvented the existing SGML IDREF-->ID mechanism with HREF-->NAME,
and I don't see any reason to change that by adding a second linking
mechanism. Neither ID nor CLASS cause any known conflicts in current tools.
I suspect one reason that ID may not be advocated is that few current UAs
can determine ID uniqueness. Other than that, I cannot see any good reason
why ID and CLASS _cannot_ be included in HTML 3.2.

On the subject of DIV, I also cannot understand the promotion [DSR] of its
use as a presentation-oriented containment element (<DIV ALIGN=CENTER>?)
rather than a structural element (<DIV CLASS=CHAPTER>). Something like SPAN
(which has no content structure connotations) seems much more appropriate
than DIV for marking up arbitrary sections of text. If we're going to be
adding a DIV element, I'd rather leave CENTER and promote DIV for structure
than corrupt DIV _as_ CENTER.

>And note that while I'm monitoring this forum to some extent, the
>editor of the specification is not (yet). So you might want to hold
>on to some of your comments until you see a working draft released.

You have released a working draft of HTML 3.2 -- I seem to recall a press
release to that effect dated May 7th. What further release should we wait


[DSR] "Re: DIV/CLASS: Mike Wexler: Re: HTML 3.2 ",
    Dave Raggett in www-html, Wed, 15 May 1996 17:44:00 -0400

     Murray Altheim, Program Manager
     Spyglass, Inc., Cambridge, Massachusetts
     email: <mailto:murray@spyglass.com>
     http:  <http://www.stonehand.com/murray/murray.html>
            "Give a monkey the tools and he'll eventually build a typewriter."