notes on conference call of 5 august

Notes on HTML-Math ERB Conference Call
5 August 96

In attendance:

Robert Miner	The Geometry Center
Bruce Smith	Wolfram Research
Neil Soiffer	Wolfram Research
Ron Whitney     American Math Society
Ralph Youngen   American Math Society

[Notes prepared by RW.  Corrections welcome.]

Robert and Neil had sent a new display-list specification to Bruce
after their confab in Champaign a couple of weeks ago.  Bruce opened
the telephone meeting by saying he thought the specification looked
good, and that he was willing to throw his quibbles in with the rest
of the Board members' after Robert posts the spec to the full
discussion list.  Robert said he would do so soon (and has already).

The group then discussed, and returned also to, some details of what
information is carried in the display list.  Bruce felt that some
constructions in his first outline of display-list format were not
really SGML compatible (he mentioned separation of items by <mc>
("math comma") elements).  Neil and Robert had comments on this, but
will have to fill them in since my notes are vague.  There will be a
change in the way attributes are handled, insofar as the rendering
applications will be reading the HTML-Math dictionaries to retrieve
the various default attributes of tokens, so the parser and
expression-tree mapper will not place attributes into the display-list
unless such attributes are overrides of the attributes in the

Ron became concerned about how much "expression" information would be
assumed or required by renderers in order to do a good job with
rendering data imported directly to display-list format and not
mediated by the HTML-Math parser and operator precedence dictionaries.
Neil said the information would be readable, but that, without
expression information, one couldn't expect top-quality rendering.
If, for example, the string "x+2a" is imported to display-list format,
the implicit multiplication operator between "2" and "a" and its
higher precedence (higher than that of "+") will not be present, so
spacing and line-breaking may be compromised.  We are probably in need
of more extensive examples to tell whether or not the quality of
the rendering for data imported directly to display-list format is
lower than will be tolerated by an "average" reader of mathematics.
Much will become apparent when we have a renderer in hand and have,
say, imported a fair quantity of AMS TeX data to it.

We discussed resource-use and capabilities of browsers.  Robert
mentioned that he felt the current Java code he's seeing in connection
with HTML-Math will be a large amount of data to download and will
probably be painfully slow.  Neil said that browsers have many
difficulties which we'll have to wait on them to fix (e.g. font
problems and CSS).  Neil commented that Dave Raggett and Steve Hunt
had made a pitch to the browser vendors regarding access to fonts
(true, Neil?) and that they might be responding in the near term
(e.g. by the end of August).

Bruce said that he would be writing parsers in both Mathematica and
Java.  There was a brief discussion on which environment might be
speedier for development; it sounded that both had their difficulties.
Bruce also has to learn Java.  Ron mentioned that he hoped one parser
would be available a couple of weeks before the ERB meeting
Sep30/Oct1, and Bruce said such a deadline would be tight.

Bruce indicated that he would post his "to do" list in regard the
whole specification of HTML-Math to this list so that we can form the
agenda for the Sep30/Oct1 meeting.

Finally, Robert divulged that he is initiating his marriage on the
wrong foot, actually choosing to tie the knot and take two weeks off
while we all stew here on the Expression Tree.  We wished him well in
spite of this.