Minutes: MathML Full meeting 12 May, 2022

 Attendees:

   - David Carlisle
   - Sam Dooley
   - Bert Bos
   - Steve Noble
   - Louis Maher
   - Bruce Miller
   - Deyan Ginev
   - Cary Supalo
   - Murray Sargent
   - Paul Libbrecht
   - David Farmer
   - Patrick Ion

<https://sandbox.cryptpad.info/code/inner.html?ver=4.14.1-00#cp-md-0-regrets->
Regrets:

   - Patrick Ion
   - Moritz Schubotz

<https://sandbox.cryptpad.info/code/inner.html?ver=4.14.1-00#cp-md-0-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports

Neil's issue on actions for FPWG https://github.com/w3c/mathml/issues/342
PL:

DC: Lots of work on Chapter 4. Making Pull Requests when words change.
Markup changes, such as folding, are done directly in the source. Required
some rearranging to move multiple examples into a single collapsing
section. No need for examples that are basically the same as others, but
with different operators. Folding the examples does not make sense
everywhere. Still going through them.

SD: In F2, the value of "tendsto" is not defined anywhere anymore, not even
in the schema.

DC: "int" and "diff" may always have been wrong.

SD: Somebody (me?) needs to go through the table in F and see if everything
makes sense.

SN: I started collecting a11y issues in a Word document. Plan to do more
next week and then put it on GitHub.

PL: Draft Media types https://w3c.github.io/mathml-docs/mathml-media-types/
<https://sandbox.cryptpad.info/code/inner.html?ver=4.14.1-00#cp-md-0-2-tpac-meeting-what-do-we-want-to-do->2.
TPAC meeting (what do we want to do)

https://www.w3.org/2002/09/wbs/1/LogisticsTPAC2022/

NS: do we want a meeting to try and nail down some issue where the group's
attention is focused for half a day or a full day?

NS: Do we need to liaise with ARIA or CSS?

https://www.w3.org/2002/09/wbs/83726/ARIATPAC2022attendance/

NS: If you plan to attend in person or virtually, please fill out the above
survey.
<https://sandbox.cryptpad.info/code/inner.html?ver=4.14.1-00#cp-md-0-3-continue-reviewing-the-mathml-4-doc-and-issues>3.
Continue reviewing the MathML 4 doc and issues

a) Allow intent attribute on content mathml Deyan has a 15 minute
presentation on why this isn't a good idea. If anyone wants to present a
counterpoint, please let me know and I'll add you to the agenda.

DG: I think this is complicated and experimentation needs to be done before
we know whether it is a useful addition.

DG: On the topic of experimentation: There is an example where someone
(randomservices org) extended CMML to include <binomial/>, adding custom
schema changes and custom upgrades to the content-to-presentation XSLT.
data-intent is valid HTML5 that can be experimented with. GPT-3 can do
experimentation without even using a schema.

DG: If we add intent, on the surface it looks like a redundant tree to the
Content tree. We started with a premise of "baseline narration" being
derived from walking the Presentation MathML tree. Hence the missing
information added through intent largely mirrors the Content tree.

DG: If we start designing an AT capability for Content MathML, the baseline
readout will be the Content operator tree. Then we are missing the visual
"notation" information and maybe should add a "notation" attribute to
convey that.

DG: Small set of examples on binomial content that illustrates these
points: https://gist.github.com/dginev/8db966d043992883dfb5ebeb0fa308c3

PL: I agree we should allow experimentation, but adding it allows
experimentation. It is simple.

DG: It is not simple. And introducing AT to Content MathML is a significant
change. It will distract from the work we did with presentation and intent
and confuse messaging to adopters.

SW: I think adding intent would simplify things because content and
presentation can be mixed. We need to figure out how the two interact.

DC: I sympathize with Deyan, but agree with Paul. It would help with
conversion. I don't think we need to specify much because this is different
from core. We just need to say it is allowed.

PL: Not allowing it hinders research into using it. Notation is a good
idea, but that is much harder.

DG: To me, it makes a big difference to have a useful description of how
the new attribute is leveraged. It should be fully specified, or I am
confused myself.

DC: It won't be more undefined than CMML is currently.

BM: If you don't like it, don't use it. There seems to be some implicit
conflict -- using intent conflicts with content. But if it's just for a11y,
that's not a conflict.

BM: Seems premature to say there is a conflict/problem when we haven't
nailed down intent

DG: It seems wishful thinking to add it when we haven't ever thought
through a bunch of examples to make sure it is useful in practice. I
believe that burden falls on the WG, and not on spec adopters.