Minutes: MathML Full meeting: 19 March, 2026

 Attendees:

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

<https://cryptpad.fr/#cp-md-0-regrets>Regrets

   - Paul Libbrecht

<https://cryptpad.fr/#cp-md-0-action-items>Action Items
<https://cryptpad.fr/#cp-md-0-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports

*ACTION:* NS: Microsoft might be willing to pay Igalia to fix some
Chrome/Edge bugs. Murray has issue #229
<https://github.com/w3c/mathml-core/issues/229> showing errors. Any other
things we can see if MS is willing to pay to fix? If so, add them.

*ACTION:* NS should look through the core issues for MathML problems to
give to Microsoft. He can also look at the Chromium bug list.
<https://cryptpad.fr/#cp-md-0-2-review-a-href-https-w3c-github-io-mathml-docs-charter-2026-html-draft-charter-a->2.
Review draft charter <https://w3c.github.io/mathml-docs/charter-2026.html>.

Issues on it are here <https://github.com/w3c/mathml/issues/568>.

*ACTION:* NS will add pdf association to external organizations.

*ACTION:* NS will put MathML5 back into the charter. We are not going to
write a working draft for MathML5.

*ACTION:* NS: Change the charter to say keep xml entities up to date.
<https://cryptpad.fr/#cp-md-0-3-spec-review-progress->3. Spec Review
progress. <https://cryptpad.fr/#cp-md-0-horizontal-review>Horizontal Review

   -

   Privacy: https://github.com/w3cping/privacy-request/issues/197
   -

   Security: https://github.com/w3c/security-request/issues/121
   -

   Accessibility: https://github.com/w3c/mathml/issues/566
   https://github.com/w3c/a11y-request/issues/151
   -

   Internationalization: https://github.com/w3c/mathml/issues/570

   https://github.com/w3c/i18n-request/issues/299
   -

   TAG: https://github.com/w3ctag/design-reviews/issues/1205

*ACTION:* NS invites the group to look over the above horizontal review
links and see if he left out anything.

*ACTION:* Consider Registries
<https://www.w3.org/policies/process/#registries> next week.
<https://cryptpad.fr/#cp-md-0-our-group-39-s-review-s->Our group's
review(s)?

*ACTION:* DG will run Claude on the core and full specs to look for grammar
errors.

*ACTION:* In the full spec C.4.2.5 Superscripts and Subscripts
<https://w3c.github.io/mathml/#acc_subsup> DG offered to make the text
clearer about where superscripts and subscripts should be placed.
<https://cryptpad.fr/#cp-md-0-agenda>Agenda
<https://cryptpad.fr/#cp-md-1-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports

ACTION: NS: Microsoft might be willing to pay Igalia to fix some
Chrome/Edge bugs. Murray has issue #229
<https://github.com/w3c/mathml-core/issues/229> showing some. Any other
things we can see if MS is willing to pay to fix? If so, add them.

NS: Microsoft has money they must use by the end of June. They might fund
Igalia for accessibility projects. There are bugs Igalia could potentially
fix. They can also work on pdf accessibility. Hopefully MS and Igalia and
work something out to improve math in Edge and Chrome.

NS: Daisy is helping NS with MathCAT.

DG: Accent rendering with <https://github.com/w3c/mathml-core/issues/254>
is still a problem.

NS: We have not decided what the solution is to create proper accents. We
cannot ask people to fix things if we ourselves do not know what to do.

From Deyan Ginev to everyone: accent character issues:
https://github.com/w3c/mathml-core/issues/301
https://github.com/w3c/mathml-core/issues/254

MuS: The tilde is not correct. Firefox is almost always right. Tilda does
not expand correctly.

NS is trying to get a list of bugs, not the ones MuS has listed, that he
can pass to Microsoft for repairs.

DC: Positioning and stretching are wrong. The spec should say what should
happen.

NS: Do we have a proposal we can take to Microsoft and Igalia for the
accent characters?

NS: Part of this issue is there's a whole bunch of characters, and we do
not know which one's the one that should work?

DC: The positioning, especially for the combining characters, is not great.

NS: We should ask that, at least, get the combining characters to be
centered correctly over the character, even if it doesn't get to stretching.

DG: Everything reported for archive has an associated MathML Issue with it.

ACTION: NS should look through the core issues for MathML problems to give
to Microsoft. He can also look at the Chromium bug list.
<https://cryptpad.fr/#cp-md-1-2-review-a-href-https-w3c-github-io-mathml-docs-charter-2026-html-draft-charter-a->2.
Review draft charter <https://w3c.github.io/mathml-docs/charter-2026.html>.

Issues on it are here <https://github.com/w3c/mathml/issues/568>.

NS: We need start and end dates.

NS has sent out all the horizontal requests. Reviewers have two months to
make comments.

NS is pushing for May 2026 as a candidate recommendation for MathML4.

NS wants to show a first full implementation of intent, core concepts, and
properties in a screen reader.

DG and MoS are developing the Community Repository for Intent Values and
Properties. NS is saying that it will be open for contributors by August
2026. DG and MoS say that this date is possible.

NS: Core is out there as a CR, but we are still making some changes.

NS: The full spec may be a recommendation by January 2027.

NS: We will need the implementation reports and test suites.

NS: Our last thing to get ready would be the polyfill implementations, so
that would give us, say, 2 months of working on only the polyfills.

NS: They had a question of who implements MathML 4. I listed MathCAT as
implementing it and being used in all these, screen readers and learning
disability tools. I listed the LaTeX project generating the intent, and I
listed Wikimedia.

NS: We need to document who is producing MathML4 and who is consuming it.
We must have two implementations of all of the features.

From Moritz Schubotz to everyone: Wikimedia has experimental authoring and
rendering support.

From Murray to everyone: My intent-oriented implementation is in
Unicode-MathML GitHub - MurrayIII/UnicodeMathML: JavaScript-based
translation of UnicodeMath to MathML 4.0. The facility supports dictation,
speech, Nemeth braille, and LaTeX. The interactive playground lets you
experiment with UnicodeMath, LaTeX, speech, braille, and Markdown.

MoS: We have an implementation for authoring and rendering in Wikimedia,
but it's not enabled anywhere. This might count towards the MathML4
implementation requirements.

NS: I said that one of the authoring tools in Wikimedia is taking advantage
of intent. I did not say that it had not been released. I can fix that if
that is needed.

MoS: What about MathJax. MathJax currently implements MathML3.

NS: They will not implement MathML4 until browsers get better.

MoS: You can convert mathml4 to mathml3 and can use MathJax.

NS: That could be a way for polyfills to work.

DC: So in external organizations, you could mention either the PDF or LaTeX.

*ACTION:* NS will add pdf association to external organizations.

MoS: If we start to implement intent and the open lists say how we want to
distribute it.

From Moritz Schubotz to everyone:
https://www.wikidata.org/wiki/Wikidata:Data_access

MoS Where do we want to go with content MathML considering using intent?

MoS: Do we start with MathML5 in this time frame?

DC: Perhaps discuss MathML 5. I mean, work on MathML5 requirements, I don't
think we should commit to making a working draft.

*ACTION:* NS will put MathML5 back into the charter. We are not going to
write a working draft for MathML5.

*ACTION:* NS: Change charter to say keep xml entities up to date.

New process document
https://www.w3.org/policies/process/#transition-rec-requirements
<https://cryptpad.fr/#cp-md-0-3-spec-review-progress>3. Spec Review progress
<https://cryptpad.fr/#cp-md-1-horizontal-review>Horizontal Review

   -

   Privacy: https://github.com/w3cping/privacy-request/issues/197
   -

   Security: https://github.com/w3c/security-request/issues/121
   -

   Accessibility: https://github.com/w3c/mathml/issues/566

   https://github.com/w3c/a11y-request/issues/151
   -

   Internationalization: https://github.com/w3c/mathml/issues/570

   https://github.com/w3c/i18n-request/issues/299
   -

   TAG: https://github.com/w3ctag/design-reviews/issues/1205

*ACTION:* NS invites the group to look over the above horizontal review
links and see if he left out anything.

NS has not yet received any horizontal review comments.

NS: in filling out the TAG horizontal review, I discovered the W3C has
something called

Registries <https://www.w3.org/policies/process/#registries> that seem like
they might be appropriate for the intent concepts. There is even a registry
track (parallel to a rec track?). Should we be considering this?

From Deyan Ginev to everyone: operator dictionary and the accent characters
are also interesting candidates for registries

*ACTION:* Consider Registries
<https://www.w3.org/policies/process/#registries> next week.
<https://cryptpad.fr/#cp-md-1-our-group-39-s-review-s->Our group's
review(s)?

DG used Claude AI to look at the intent section to look for grammar errors.

NS used Gemini to look over the spec for grammar errors.

DG says Claude is better than Gemini.

DG said that Claude caught Unicode typos.

*ACTION:* DG will run Claude on the core and full specs to look for grammar
errors.

LM: I was asked to proof-read appendix C on accessibility in the MathML 4
full spec. Everything looks good except for: C.4.2.5 Superscripts and
Subscripts <https://w3c.github.io/mathml/#acc_subsup>

It says: "It is important to apply superscripts and subscripts to the
appropriate element or sub-expression. It is not correct to apply a
superscript or subscript to a closing parenthesis or any other grouping
symbol. Important for navigation"

I disagree, how about: (a+b)^2

I would reduce this section to: "It is important to apply superscripts and
subscripts to the appropriate element or sub-expression. Important for
navigation."

DC: If you brace it and put the superscript on the whole expression, it
floats away from the closing bracket, then it will snap.

DG: I had a second suggestion to clarify that we're talking about a
standalone parenthesis in the sense of, even, not a parenthesis, a
standalone MO element, because it's really about, don't put superscripts on
the standalone MO elements, put them on the sub-expression.

DG: I can modify it to say: so don't put superscripts on a standalone MO
element, except in the cases where you're compensating for missing Unicode
characters, because I think that's the exception.

*ACTION:* In the full spec C.4.2.5 Superscripts and Subscripts
<https://w3c.github.io/mathml/#acc_subsup> DG offered to make the text
clearer about where superscripts and subscripts should be placed.
<https://cryptpad.fr/#cp-md-0-zoom-summary-for-the-intent-meeting-of-3-19-2026>Zoom
Summary For The Intent Meeting Of 3/19/2026
<https://cryptpad.fr/#cp-md-0-summary>Summary

The meeting focused on discussing potential accessibility improvements for
MathML, with Neil announcing an opportunity to secure funding from
Microsoft for bug fixes and accessibility enhancements by June 30th. The
group reviewed specific rendering issues including problems with fractions,
parentheses, and accent positioning in various browsers, with Murray and
David demonstrating current display problems. The team also reviewed and
updated their charter for 2026, discussing timelines for MathML Core
recommendations and implementation reports, while considering the addition
of MathML5 requirements and registry tracking for intent concepts.
Additionally, the group discussed using AI tools like Claude for spec
review and editing, with Deyan sharing his positive experience using Claude
for finding and correcting grammar issues in the specification.
<https://cryptpad.fr/#cp-md-0-mathematical-font-character-positioning>Mathematical
Font Character Positioning

The team discussed issues with character positioning and stretching in
mathematical fonts, particularly focusing on accents, and combining
characters. They observed that Firefox generally handles these characters
correctly, while other browsers like Edge and Chrome have inconsistencies.
Murray shared that Cambria Math has supported stretchy characters since
2007, but other fonts do not always implement this feature consistently.
The team agreed to propose a specification for correct character
positioning and stretching to Microsoft and Galia, with the goal of
improving support across browsers. Neil suggested focusing on getting
combining characters to center properly over the base character as a first
step.
<https://cryptpad.fr/#cp-md-0-mathml-core-issues-discussion>MathML Core
Issues Discussion

The team discussed MathML core issues and Chromium bugs, with Deyan noting
that most reported issues have at least one MathML core issue associated
with them. Neil agreed to review core issues for rendering problems and
mentioned that Brian is in contact with someone to explore potential
solutions. The group also reviewed and updated the charter, with Neil and
Brian remaining as chairs and Bert as team contact. They decided to hold
weekly to monthly meetings based on issue needs, with no face-to-face
meetings planned. Neil expressed concern about explaining the value of
these bugs to non-technical executives and asked for help in framing the
importance of these issues.
<https://cryptpad.fr/#cp-md-0-mathml-specification-review-meeting>MathML
Specification Review Meeting

Neil and David reviewed changes to a specification document, focusing on
deliverables and timelines. They discussed the status of MathML Core and
MathML4, noting that MathML5 might not be feasible given the current pace
of work. Neil outlined several deliverables, including documentation of
core concepts, test suites, and community repository work. They also
touched on the development of a MathML converter, which David had already
made progress on.
<https://cryptpad.fr/#cp-md-0-mathml4-timeline-planning-discussion>MathML4
Timeline Planning Discussion

The team discussed the timeline for MathML4, with a target recommendation
date of May 2026. They agreed on an August 2026 deadline for a community
repository for intent values and properties, which Moritz and Deyan
confirmed was realistic. Neil mentioned the need to update the timeline
regarding test suites and implementation reports for MathML Core, which
Bert clarified no longer requires a proposed recommendation phase. The team
also discussed the potential for Microsoft to fund a first full
implementation of intent core concepts and properties in a screen reader.
<https://cryptpad.fr/#cp-md-0-recommendation-transition-process-updates>Recommendation
Transition Process Updates

The team discussed the process for transitioning a recommendation, which no
longer requires a proposed recommendation but instead needs implementation
reports, test suites, and evidence of addressing previous issues. They
agreed to prepare these reports by January and focus on polyfill
implementation afterward. Neil mentioned the need to document authoring
tools and consuming tools that use intent, and the team discussed MathML4
implementation challenges, with Moritz suggesting extending MathJax to
render MathML4 input by converting it to MathML3 internally.
<https://cryptpad.fr/#cp-md-0-charter-review-and-updates>Charter Review and
Updates

David and Neil reviewed changes to a charter, noting that most sections
remained unchanged except for updates to the external organizations
section, where they agreed to add the PDF Association. Moritz raised a
question about starting with MathML5 and content management, suggesting it
would be beneficial to discuss and potentially make adjustments, even if
major changes to content management are not implemented immediately.
<https://cryptpad.fr/#cp-md-0-mathml5-requirements-discussion>MathML5
Requirements Discussion

The team discussed MathML5 requirements and decided not to commit to
creating a working draft, as MathML4 is nearly complete. David noted that
the entity specification would not change significantly due to Unicode
updates. Neil was tasked with updating the charter to reflect keeping XML
entities up to date. Deyan inquired about the bold text in the scope
section, which Neil explained was due to a class styling issue. The
conversation ended with a reminder for participants to review horizontal
review links for any issues.
<https://cryptpad.fr/#cp-md-0-registry-formalization-and-specification-review>Registry
Formalization and Specification Review

The team discussed several topics, including the use of registries for
intent concepts and properties, which was introduced as a potential
formalization method. Bert explained that registries are
recommendation-level documents suitable for non-normative tables of
keywords or items that may change over time. The group agreed to explore
this idea further in the next meeting. Additionally, they addressed the use
of AI tools like Claude and Gemini to identify and correct grammatical and
spelling errors in the specification, with Deyan emphasizing Claude's
capabilities. The team also debated the placement of subscripts and
superscripts on closing parentheses, with David and others providing
insights from typesetting practices. Finally, they noted the need for
feedback on horizontal reviews and confirmed the next meeting's schedule.