notes on teleconference of 26 august

Notes on HTML-Math ERB Conference Call
26 August 96

In attendance:

Stephen Glimm   Mathsoft
Patrick Ion	Mathematical Reviews
Robert Miner    The Geometry Center
T.V. Raman      Adobe Systems
Bruce Smith	Wolfram Research, Inc.
Stephen Watt    Safir, INRIA
Ron Whitney     American Mathematical Society
Ralph Youngen   American Mathematical Society

[Notes prepared by RW.  Corrections welcome.]

The conversation was robust.  As in all of these posted
regurgitations, memory lapses and facile observations should be
corrected; 18-minute elisions should be brought to light.

We opened with a small update on implementation progress.  Robert's
been out for a couple of weeks and is just back to work.  Bruce
has made good progress on the Mathematica parser, but it's not
there yet.  He expects it to be available soon.

The proposed agenda for this teleconference listed (1st) a discussion of
the meeting agenda for Sep30/Oct1, and (2nd) a continuation of the email
correspondence on "semantics".  Since the semantical issues have
elicited so much response, Ron suggested that it might be useful to
reverse the order of topics.

In response to Bob's posting on the email list, Ron acknowledged that
macros are indeed part of the WP.  Given that this is perhaps not
all that well-known to newer ERB members and that Raman posted a note
asking for further discussion of extensibility, Ron asked Bruce to
comment on the macro facility of the WP.  Bruce said that he was
in basic agreement with Sutor's suggestion of allowing semantical
annotations, and that this might be done either via macros or by
embedding such annotations within the ambient mathematical notation
(using perhaps attributes).  Ron suggested that a small message
to the mailing list describing some more details of all this would
be helpful (and Bruce agreed).

Raman argued that macros allow new objects to have "first-class"
status.  Stephen Watt mentioned that macros aid the ability to give
alternative renderings.  Ron mentioned that, despite all the wonderful
things one imagines authors accomplishing with macros, they do present
management difficulties (they introduce `new notation' which is
generally difficult for people other than an author to handle).  There
followed a discussion on ways to avoid problems with macros, and there
was some feeling expressed that macros are potentially so helpful in
any case that we don't want to eschew them.  Bruce will further the
discussion with his posting of some details about macros and
template-matching.  [[It's also difficult for me to do justice to all
the remarks made about macros.  I encourage others to further the
discussion in email. -RW]]

Raman spoke so vigorously for macro use that Ron asked about his views
on MINSE.  Raman said he couldn't characterize his reaction at the
moment, but would look in further detail at MINSE.

Ron moved the conversation on to discussion of potential agenda items
for the Sep30/Oct1 meeting.  Stephen Watt inquired about clarification
of milestones that the committee has set for itself.  Ron spoke about
the proposal we have aimed at delivering to the W3C by the end of the
year.  In addition, we aim to be providing a "reference
implementation" along with the proposal.  Milestones will be clarified
and adjusted at the Sep30/Oct1 meeting.  [[I had proposed putting this
discussion toward the end of the meeting, but we can move it forward
as we review the proposed agenda to start the meeting. -RW]] Ron also
said he would discuss with Dave the detail of official milestones we
must meet for the W3C (which are presumably loosened when we change to
Working Group status).

It seems clear that the issues of extensibility and the role of the
"display list" need clarification.  There was a discussion of what
"display list" now means, but I must admit I'm confused on where we
stand.  Ralph Youngen and I had different takes on what was said in
the discussion.  Bruce had discussed last week that "display list" in
his original specification meant a data structure he intended to be
suitable for both visual and audio rendering.  Robert indicated today
that he had been taking the display list spec to be one suited to
describing characteristics for visual display.  Others seemed to
agree that such a structure was appropriate.  It would help me to have
a written update on how the WP now considers transformation from
expression list to visual display will be achieved.  Some of the
process will involve template-matching, but I'm unclear as to whether
we are targeting a lower-level entry point for visual-layout data
(such as TeX or ISO 12083).

We ended the conference in the middle of a discussion about the need
for operator-precedence parsing.  Stephen Watt suggested that the need
for this depends upon how we see data-entry being handled (the
suggestion being that "smart" editors could be used to generate
encodings which parse more easily than the WP's operator-precedence
scheme).  Bruce said that it has been felt that human-generated and
human-readable syntax is important and that the operator-precedence
parsing is a key to this.  Stephen and Bruce may continue the
conversation in email.