Minutes: MathML full meeting 3 October, 2024

 Attendees:

   - Neil Soiffer
   - Louis Maher
   - David Carlisle
   - Moritz Schubotz
   - Deyan Ginev
   - Bruce Miller
   - Paul Libbrecht
   - Bert Bos
   - Patrick Ion

<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-regrets>
Regrets

   - Murray Sargent

<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-action-items>Action
Items

*ACTION:* Once we get the current Pull requests merged, DC will ask Bert to
move the draft to TR space. It will still be a normal working draft.

DG: Potential test suites that can use Interop work:

   1.

   presentation-markup test suite:
   https://wpt.fyi/results/mathml/presentation-markup?label=master&label=experimental&aligned&q=math%20%20not%28path%3A%2Fjs%29
   2.

   css-styling test suite:
   https://wpt.fyi/results/mathml/relations/css-styling?label=master&label=experimental&aligned&q=math%20%20not%28path%3A%2Fjs%29

DG: I think we can review something by the group, just not in a meeting. I
was going to suggest making a GitHub issue on one of our repositories and
then anybody else can review it. We can submit the result by Monday evening
October 7.

NS we should put in two separate requests to get the existing tests to
work: one, the presentation case, and two, the css case.

*ACTION:* MoS: Get the interop issue started.
<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-agenda>
Agenda
<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports
<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-content-delivery-network-cdn-problem>Content
Delivery Network (CDN) Problem

DG: Fred Wang helped me diagnose a CDN bug with serving the STIX Two Math
font. He reported it in 2021, but it was left unsolved, while the CDN still
serves the font - so there is no indication it is broken.

The font gets split into subsets, and some of the relevant entries for math
rendering do *not* get transmitted, leading to a variety of
browser-specific degradation: square roots vanishing, operators not
stretching, possibly others.

I have witnessed this behavior also when using the Typekit CDN.

The only reliable solution we know today is self-hosting the single
woff/woff2 file from the official STIX distribution.

   - https://github.com/google/fonts/issues/3773

<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-in-the-9-26-2024-intent-meeting-we-considered->In
the 9/26/2024 intent meeting, we considered:
<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0--a-href-https-github-com-w3c-mathml-issues-72-deprecate-remove-mlabeledtr-72-a->Deprecate/Remove
mlabeledtr #72 <https://github.com/w3c/mathml/issues/72>

DC removed mlabeledtr from full. He changed the spec.

DC made a pull request.

DC: I moved it to The MathML legacy schema, which was pre-existing.

DC: If you use the legacy schema, mlabeledtr will come back.

DC discovered more problems while trying to deprecate mlabeledtr.
<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0--a-href-https-github-com-w3c-mathml-issues-498-xsd-not-allowing-semantics-element-as-child-of-apply-element-498-a->XSD
not allowing -element as child of -element #498
<https://github.com/w3c/mathml/issues/498>

DC said: This is a bug, and I will fix it.

DC Had problems fixing this.
<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-tpac-meeting>TPAC
Meeting

NS sat in on the TPAC management meetings where people said there were too
many hoops that people had to jump through to get things done. But another
person said but that's the value of the W3C Is that we do these other
reviews, and that is important.

Due to these review procedures, documents were not being updated. NS
noticed that we had documents that needed updating.

NS: We have way too many things that sit in GitHub and do not move into the
technical report (TR) status.

NS: Can we take our current draft and move it into TR space?

DC: Bert can do it, but it can be a pain.

DC: Ask Bert to take the editor's draft and send it to TR space.

PL: It is a good thing to do. It is useful to not rely on the current non
released version as an authority.

NS Is it worth pushing things to TR space? Should we spend meeting time on
this?

NS: A majority of our meeting thought it was worth spending time on this.

*ACTION:* Once we get the current Pull requests merged, DC will ask Bert to
move the draft to TR space. It will still be a normal working draft.

BB can show DC the automated method to move a document to TR space.

DC: the MathML core is in the same state.

PL: After we have a new TR, we need to advertise the fact so people can
know we are accomplishing things.

PI: If we do not do marketing, no one will know we exists.

NS: We need to get out a candidate recommendation and advertise that.

PL: Received an answer to Bert's question. we do not have a w3c expert
accessing the ISO documents. We had some criticism that some of our symbols
clashed with the ISO 8002 standard.

PL: Response was that we have trouble relying on documents that are not
public.
<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-interop>
Interop

MathML interop items proposals (closes 9 October) Where to submit:
https://github.com/web-platform-tests/interop/issues/new/choose

Previous Interop issues about MathML:

   - https://github.com/web-platform-tests/interop/issues/556
   - https://github.com/web-platform-tests/interop/issues/197

NS said that BK said that There are no Interop suggestions from MathML.

NS: If we want interop to do things for Math, we must put out suggestions.
Last year, DG did that issue 556
<https://github.com/web-platform-tests/interop/issues/556> And BK
submitted issue
197 <https://github.com/web-platform-tests/interop/issues/197> in 2022.
These are potential starting points.

NS: Instead of asking for small things, we should go bigger.

NS: Let us ask for full CSS compatibility among the browsers.

NS: We need tests. There are interop tests for CSS.

DG: You must have tests. You can start with existing tests.

DG: There is a presentation test suite on Interop.

DG: Also, there is a Css styling test suite on interop.

DG: Both of these are large. And I would love to have both of these
completely green, because they are core functionality in a way, especially
the presentation one.

DG: Interop must have tests, but you can just ask that the existing test
work.

NS: Anyone can submit a pull request saying this is a new test.

NS: We could do this in a meeting to show how it is done.

NS: We need to get proposals in.

DG: It is not possible to get new tests before the October 9 deadline. We
should ask that existing test pass.

NS: Who wants to do a new interop issue?

MoS will look into it.

PL offered to review it.

DG: I think we can review something by the group, just not in a meeting. I
was going to suggest making a GitHub issue on one of our repositories and
then anybody else can review it. We can submit the result by Monday evening
October 7.

NS we should put in two separate requests to get the existing tests to
work: one, the presentation case, and two, the css case.

DG: Potential test suites that can use Interop work:

   1.

   presentation-markup test suite:
   https://wpt.fyi/results/mathml/presentation-markup?label=master&label=experimental&aligned&q=math%20%20not%28path%3A%2Fjs%29
   2.

   css-styling test suite:
   https://wpt.fyi/results/mathml/relations/css-styling?label=master&label=experimental&aligned&q=math%20%20not%28path%3A%2Fjs%29

*ACTION:* MoS: Get the interop issue started.
<https://sandbox.cryptpad.info/code/inner.html?ver=2024.9.0-3#cp-md-0-3-issue-a-href-https-github-com-w3c-mathml-issues-473-core-concepts-how-should-the-multiple-forms-of-differentiation-be-handled-473-a->3.
Issue: core concepts: how should the multiple forms of differentiation be
handled? #473 <https://github.com/w3c/mathml/issues/473>

DG said use properties to define it. You would have the property, then the
derivative name.

NS: There are the terse and verbose modes.

NS: Do people feel that there's a useful verbose translation or verbose
pronunciation, or that it really should just be pronouncing the notation as
it is?

BM: We should not over complicate things. We have two ways of reading
expressions, the semantic (derivative of F. With respect to X,), versus
purely a structural where you just walk through the tree and end up with X
dot or X over or etc., etc. with minimal, derivative specific, aspects to
it.

BM: Following DG's ideas, the system might shift between a more or less
semantic reading mode as you traverse the expression.

BM: Short form properties are helpful, but we could end up with a huge
number of complicated properties if we're not careful.

LM prefers terse readings as long as he has a Braille display. If he does
not have such a display, LM prefers a structural reading so that he can
copy and study the expression later.

DG: Archive hosted accessibility forums using complex mathematical material
whose equations were paragraphs. They interviewed several blind individuals
reading these complex papers.

DG: One blind individual told DG that he needs to navigate archive formulas
because archive formulas are basically paragraphs of information on 2 lines
or one line. They're very dense and very custom. And the symbols and the
way they're used tend to be above high school level. So, you really need to
be able to stop, pause and go into the fragments and look at individual
pieces, so he convinced me navigation is an important feature. And so yes,
navigation is plainly important for higher level materials.

NS: Yes. And so, the issue is, does he just want to navigate and know what
the notation is? Or does he want to navigate semantically?

DG: He needs to understand what's written. That's what semantics is for.
It's the understanding of the formula. So, knowing the argument holes the
actual arguments to the mathematics, to the operation you're performing
helps. You understand it. Going through the slashes and the fraction lines
it helps. You understand the two-dimensional layout of where things are.
But it doesn't tell you anything about what it means.

DG: So, I think the answer is both. You need both.

DG: The problem is that if an expression is long and complicated, you
cannot hold it in your mind.

DG: You don't stop when a large argument is traversed. So, I think it's the
boundaries to larger components that are the crucial things in navigation.

NS: Several papers have said, 7 to 9 syllables is what your working memory
is. And once you pass that, you can't keep it in your working memory.