presentation markup for math -- Was: semantic markup for math

William F. Hammond (
Fri, 19 Jul 1996 18:37:17 -0400 (EDT)

From: "William F. Hammond" <>
Date: Fri, 19 Jul 1996 18:37:17 -0400 (EDT)
Message-Id: <>
Subject: presentation markup for math -- Was: semantic markup for math

Mark Fisher <> writes to Dan Connolly
<> (shared with www-html):

: Daniel W. Connolly writes in <>:
: . . .
: >It seems that the people who have actually bitten off the task of Math
: >for HTML are the folks who do symbolic math packages, and they don't
: >intend to stop until they've got interoperability with their packages.
: .  . .
: The symbolic math folks are absolutely correct in pursuing structural math 
: markup, as "more structure" is the direction HTML is heading in.  Structural 
: markup, for both tree structures of text and tree structures of math, is 
: just a lot easier to automatically process. . . .

It's like this:

1.  It is not reasonable to ask general purpose renderer-authors to
    deal with the level of mathematical semantics required for
    symbolic processing.  The needs of symbolic processing can be
    handled on the web by setting a format for symbolic processing
    interchange.  There is no gain for symbolic processing in having
    that be part of HTML.

2.  Slow processing due to high bandwidth pushes away readers.  We've
    had the high bandwidth presentation formats all along (except for
    PDF, which is fairly new).  (You think all those hits on your
    servers represent readers who stayed to look?)

3.  Specialized processing needs push away readers.  (How many can
    handle "application/x-dvi"?)

4.  Learning new markup not required for communication pushes away
    providers of content.

5.  HTML is not really very "structural".  For example, the SENTENCE
    is not something that is recognized in HTML.  Does it make sense
    to make HTML-Math more structural than HTML overall?

6.  There is no end to the layering of context that one undertakes
    in getting formal.  For example, do we want the HTML <head> to
    stipulate the axiom of choice?  It is not imaginable that general
    purpose renderers would accommodate this.

7.  We need a "lowest common denominator" meeting ground for math
    in HTML, and the March '95 W3C draft for HTML-3 was rather close.
    Mimicing LaTeX is the sensible thing to do.

Authors can always provide anchors to "text/symbolic-processing-interchange"
next to their <math> tags.

                                   -- Bill