Minutes: MathML Core meeting, 28 July, 2025

 Attendees

   - Brian Kardell
   - Neil Soiffer
   - Louis Maher
   - Eri Pérez
   - David Carlisle
   - Deyan Ginev
   - Paul Libbrecht
   - Bruce Miller
   - Murray Sargent

<https://cryptpad.fr/#cp-md-0-regrets>Regrets
<https://cryptpad.fr/#cp-md-0-action-items>Action Items
<https://cryptpad.fr/#cp-md-0-high-level-review->High level review:
<https://cryptpad.fr/#cp-md-0--a-href-https-github-com-w3c-mathml-core-issues-q-is-3aissue-20state-3aopen-20label-3aneeds-tests-11-issues-still-labelled-39-needs-tests-39-a->11
issues still labelled 'needs tests'
<https://github.com/w3c/mathml-core/issues?q=is%3Aissue%20state%3Aopen%20label%3Aneeds-tests>

BK: We have eleven labels saying, "needs tests", are they still correct?
Someone needs to look.
<https://cryptpad.fr/#cp-md-0--a-href-https-github-com-w3c-mathml-core-issues-q-is-3aissue-20state-3aopen-20label-3a-22need-20specification-20update-22-13-issues-labelled-as-39-needs-specification-update-39-a->13
issues labelled as 'needs specification update'
<https://github.com/w3c/mathml-core/issues?q=is%3Aissue%20state%3Aopen%20label%3A%22need%20specification%20update%22>

BK: Do these still require specification updates?

*ACTION:* BK: we will have a session, On Monday, August 11, at the usual
time, to go through these issues and see if they still need 'needs tests',
and/or "'needs specification update", or needs a resolution. People are
invited to review these issues and make comments.

*ACTION:* MuS has Microsoft friends who are interested in OpenType issues.
He will invite them to look at the appropriate issues and comment, or work,
on them.
<https://cryptpad.fr/#cp-md-0-new-issue-a-href-https-github-com-w3c-mathml-core-issues-290-add-note-for-authors-about-preventing-whitespace-trimming-and-collapsing-in-elements-290-a->New
Issue: Add note for authors about preventing whitespace trimming and
collapsing in elements #290 <https://github.com/w3c/mathml-core/issues/290>

   - Closeable?

*ACTION:* DC will make a PR for core issue 290: Add note for authors about
preventing whitespace trimming and collapsing in elements
<https://github.com/w3c/mathml-core/issues/290>
<https://cryptpad.fr/#cp-md-0-agenda-items>Agenda Items
<https://cryptpad.fr/#cp-md-0-1-announcements-updates>1.
Announcements/Updates

We will have a special Core meeting on Monday, August 11, 2025, at the
usual time, to review issues marked as 'needs tests', and/or "'needs
specification update".
<https://cryptpad.fr/#cp-md-0-introductions-new-member-from-igalia-eri-pazos-p-rez>Introductions:
New Member from Igalia: Eri Pazos Pérez

The group introduced ourselves.
<https://cryptpad.fr/#cp-md-0--a-href-https-www-igalia-com-2025-07-14-igalia-interop-and-the-sovereign-tech-fund-html-sovereign-tech-fund-a->Sovereign
Tech Fund
<https://www.igalia.com/2025/07/14/Igalia-Interop-and-the-Sovereign-Tech-Fund.html>

What is funded/priorities of work

BK: MathML will be a large part of the funding. There will be funding for
about one year.

EP gave a short presentation on the funding
<https://eri.pages.igalia.com/slides/2025/07_mathml_working_group/>.

EP: We got funding to work on a few things in Web Interoperability, and one
is MathML.

EP: We divided the work into three milestones. One is Regarding new CSS
features that work with MathML. The second one is improving the support for
existing CSS properties across browsers. And finally, general improvements
to MathML layout.

So, for the first milestone, the idea is that we want to replace legacy
features with CSS-specific things that MathML Core added. We already have a
patch ready for Gecko, and a work in progress patch for WebKit.
Additionally, we will be working on things like a math variant and script
level, and other CSS properties.

For the second milestone, the idea is to take existing CSS features that
can be applied to MathML, but the behavior is inconsistent across browsers.
One example of this is RTL, right-to-left direction, and there is shipped
experimental support in Chromium for leaf level and character level
mirroring now, and we are working on a similar patch for Gecko also.

Additionally, we will be working on other CSS properties and then handling
special position elements in MathML layout.

Finally, the idea with this last milestone is to improve the general
testing scores related to Mathematics.

We already merge support for unknown elements being rendered as an mrow.

EP is doing the work.

NS: How can we help you move forward quickly? Should we be writing more
tests?

EP: I think test cases are useful here for the things that we already did.
They were essential to know if we were getting things right.

EP: You could review my work to see if we are paying attention to the
important issues.

MuS: You mentioned MathVariants, and I have a polyfill that I have not
checked it in. It is written in JavaScript, and it translates the
MathVariants.

EP Is not currently working on MathVariants but is interested in doing so.

NS: Polyfills bridge the differences between core and full. The polyfills
may not work on all browsers.

EP could get test from the polyfill failures.

From Deyan Ginev to everyone: side question: Do we know if the MathJax team
is interested in updating for MathML 4 Full input?

BK: You have different browsers with different tests to work on each year
and try to see their interop scores improve each year. This is how interop
works. EP would like to create a dashboard to show the tests as a function
of browsers. You want tests to pass all three browsers.

BK: The more tests we have the better, and the sooner we get the tests the
better.

DC: Do you have access to Safari along with chrome and Firefox?

EP: Yes. I am working on WebKit.

PL: We work with Arabic notation and RTL issues.

DC: If you get something into WebKit, does it get to Safari?

BK: It will make it into Safari.

DG: Is the scope of the work cover MathML core level 1? Yes says BK.

NS: We must start writing our charter in August. Perhaps level 2 items
might get to the charter?
<https://cryptpad.fr/#cp-md-1-high-level-review->High level review:
<https://cryptpad.fr/#cp-md-1--a-href-https-github-com-w3c-mathml-core-issues-q-is-3aissue-20state-3aopen-20label-3aneeds-tests-11-issues-still-labelled-39-needs-tests-39-a->11
issues still labelled 'needs tests'
<https://github.com/w3c/mathml-core/issues?q=is%3Aissue%20state%3Aopen%20label%3Aneeds-tests>

BK: We have eleven labels saying, "needs tests", are they still correct?
Someone needs to look.
<https://cryptpad.fr/#cp-md-1--a-href-https-github-com-w3c-mathml-core-issues-q-is-3aissue-20state-3aopen-20label-3a-22need-20specification-20update-22-13-issues-labelled-as-39-needs-specification-update-39-a->13
issues labelled as 'needs specification update'
<https://github.com/w3c/mathml-core/issues?q=is%3Aissue%20state%3Aopen%20label%3A%22need%20specification%20update%22>

BK: Do these still require specification updates?

NS: Perhaps some of these issues have not been fully resolved?

BK: Perhaps we should go through them and see if they still need work?

MuS: Integral and parentheses sizes were really giving trouble.

BK: Some of these issues are level two.

*ACTION:* BK: we will have a session, On Monday, August 11, at the usual
time, to go through these issues and see if they still need 'needs tests',
and/or "'needs specification update", or needs a resolution. People are
invited to review these issues and make comments.

*ACTION:* MuS has Microsoft friends who are interested in OpenType issues.
He will invite them to look at the appropriate issues and comment or work
on them.

From Deyan Ginev to everyone: +1 to getting the Microsoft folks to comment
on GitHub. That way they will stay notified of updates.
<https://cryptpad.fr/#cp-md-0-issues->Issues:
<https://cryptpad.fr/#cp-md-1-new-issue-a-href-https-github-com-w3c-mathml-core-issues-290-add-note-for-authors-about-preventing-whitespace-trimming-and-collapsing-in-elements-290-a->New
Issue: Add note for authors about preventing whitespace trimming and
collapsing in elements #290 <https://github.com/w3c/mathml-core/issues/290>

   - Closeable?

From David Carlisle to everyone:
https://github.com/w3c/mathml-core/issues?q=isAissuestateAopenlabelAopentypeFtex

DC: We resolved last time that, that is the way it had to be, so we should
just document it.

DC added the labels: "needs specification update", and "needs-tests".

*ACTION:* DC will make a PR for core issue 290: Add note for authors about
preventing whitespace trimming and collapsing in elements
<https://github.com/w3c/mathml-core/issues/290>
<https://cryptpad.fr/#cp-md-0--a-href-https-github-com-w3c-mathml-issues-120-top-remove-deprecate-simplify-the-ms-element-120-a->Remove/deprecate/simplify
the ms element #120 <https://github.com/w3c/mathml/issues/120#top>

NS summarized: This was discussed today at the MathML Core meeting. The
main points that were brought up were:

   - Full and Core are in conflict: full defaults to assuming open/close
   quotes and core defaults to not assuming quotes.
   - ms is in the parser and it is a leaf, and therefore an integration
   point with HTML. Therefore, it must stay in core (it cannot be an unknown
   core element treated as mrow)
   - Changing full breaks legacy MathML. While ms might not be that common
   in general math, in specific contexts, it is quite common. @davidcarlisle
   <https://github.com/davidcarlisle> mentioned in another call that his
   company's documentation has tens of thousands of instances of it because
   they are documenting code that does mathematical computations.
   - Using :before and :after can be problematic for copy/paste, but not
   for accessibility. However, copying MathML as plain text is going to fail
   in many circumstances (e.g., for fractions and scripts), and even when
   copying linear text, italics and spacing will be lost. For ms, if it is
   copied in a MathML context, then the MathML should be copied and the quotes
   will be preserved.
   - We did not discuss the problems around RTL.

We did not resolve what should be done.
<https://cryptpad.fr/#cp-md-0-related-a-href-https-github-com-w3c-mathml-core-issues-289-clarify-modify-usage-of-quotes-in-elements-for-compatibility-with-full-mathml-289-a->Related:
Clarify/modify usage of quotes in elements for compatibility with full
MathML #289 <https://github.com/w3c/mathml-core/issues/289>

NS: We talked about this in full. Core and full disagree. Core says ignore
them, and full assumes they are double quotes.
<https://cryptpad.fr/#cp-md-0-zoom-meeting-summary-for-math-wg-core-2025-07-28->Zoom
Meeting Summary for Math WG Core (2025-07-28)

The meeting began with technical discussions about Unicode Technical Report
25 and MathML conversion processes, including challenges with encoding and
transformer tools. The group welcomed new members and discussed the history
and development of MathML, including its accessibility features and ongoing
work on interoperability funded by the Sovereign Tech fund. The team
explored various technical issues related to MathML specifications and
polyfills, discussed testing approaches and visual presentation methods,
and planned a working session for mid-August to address ongoing projects
and specification updates.
<https://cryptpad.fr/#cp-md-0-unicode-technical-report-25-discussion>Unicode
Technical Report 25 Discussion

The meeting began with technical discussions about Unicode Technical Report
25, where Murray explained he was using XML for editing and HTML
conversion, which David had previously worked on. Murray encountered issues
with DTD errors when using math alphanumeric, which David suggested might
be due to encoding problems. The group discussed Saxon as the transformer
tool for the conversion process. Neil then attempted to start the meeting
properly by introducing a new member, Eri Pérez , and suggested doing an
"around the house" introduction, but the transcript ends before this could
be completed.
<https://cryptpad.fr/#cp-md-0-mathml-working-group-introduction>MathML
Working Group Introduction

The meeting introduced members of the MathML working group, including Neil
Soiffer and Brian Kardell, who co-chair the group. Participants shared
their backgrounds and experiences with MathML, with some members having
been involved since its inception in 1999. The group discussed the history
and development of MathML, including accessibility features and the
conversion of mathematical content to speech and braille. New member Eri,
from Igalia, was welcomed to the group, bringing his experience to the team.
<https://cryptpad.fr/#cp-md-0-mathml-interoperability-funding-plan>MathML
Interoperability Funding Plan

The team discussed funding received from the Sovereign Tech fund for work
on MathML interoperability, with Eric presenting a three-milestone plan
focusing on CSS features, existing CSS property improvements, and general
MathML layout enhancements. Eri explained that while he is currently
working alone, Fred is providing initial support, and the team is working
on patches for both Gecko and WebKit browsers, with tasks tagged for all
major browsers. Neil inquired about ways to help the project progress, and
Eric suggested that reviewing test cases and highlighting important issues
could be beneficial.
<https://cryptpad.fr/#cp-md-0-javascript-math-polyfill-discussion>JavaScript
Math Polyfill Discussion

Murray shared his JavaScript polyfill for math variance, which translates
mathematical symbols into Unicode code points, and inquired about its
potential use. Eri mentioned that he is currently focusing on font family
math but will consider Murray's work in the future. Neil and Brian
discussed the differences between MathML full and MathML core, as well as
the challenges polyfills face due to inconsistencies between browsers. Paul
suggested that knowing the version number might be sufficient to disable
polyfills, but the group acknowledged the need for more explicit methods,
especially for deprecated features.
<https://cryptpad.fr/#cp-md-0-css-polyfills-and-browser-testing>CSS
Polyfills and Browser Testing

The team discussed the development of CSS polyfills and their inconsistent
behavior across different browsers. Brian explained the interop project's
goals, emphasizing the importance of increasing interop scores through
comprehensive testing. Neil agreed to compile a list of test cases, while
Eri mentioned ongoing work on WebKit patches. Paul suggested a visual
approach to presenting test results and emphasized the importance of
visually appealing math notation. Deyan clarified that the scope of the
work was limited to level one mathematical core, and Brian noted some
flexibility in prioritizing tasks within the sovereign tech fund.
<https://cryptpad.fr/#cp-md-0-issue-label-review-and-prioritization>Issue
Label Review and Prioritization

The team discussed the need to review and update labels for various issues,
particularly focusing on those marked as needing tests or specification
updates. Brian raised concerns about the large number of issues requiring
specification updates, many of which were from 2019, and suggested adding
an "open questions" label to help prioritize work. David acknowledged that
he had been busy with other tasks but agreed to review the core issues,
while Murray mentioned ongoing work on OpenType table features. The group
considered organizing a small working session to address these issues more
effectively.
<https://cryptpad.fr/#cp-md-0-opentype-working-session-planning>OpenType
Working Session Planning

The team discussed scheduling a working session for mid-August, with David
confirming his availability on August 11th. Murray suggested focusing on
OpenType issues and proposed involving Microsoft experts Sergey Malkin and
Peter Constable, with Brian suggesting a split-meeting format to
accommodate their schedules. It was decided to have the Microsoft experts
review the issues and make their comments on GitHub. The team agreed to
review the 'needs tests' and/or "'needs specification update", issues list.
<https://cryptpad.fr/#cp-md-0-spec-updates-and-css-alignment>Spec Updates
and CSS Alignment

The team discussed two key issues: core spec updates and CSS alignment
properties. David agreed to create a PR for issue 290 to add a note to the
spec. Neil raised concerns about a contradiction between core and full
specifications regarding attribute handling, but David clarified this was
not the issue he was addressing. The team also discussed issue 120 and its
relationship to core issue 289, with Neil suggesting that removing ms from
core might be a solution, though Deyan noted this had not been officially
decided.
<https://cryptpad.fr/#cp-md-0-mathml-integration-and-compatibility-discussion>MathML
Integration and Compatibility Discussion

The team discussed the handling of MathML in core and its integration with
HTML parsing. They explored the implications of removing MathML support,
including potential compatibility issues and the impact on copy-paste
functionality. The group agreed to consider simplifying the description of
differences between core and full specifications, though they acknowledged
that any solution would likely have some drawbacks. They decided to
continue the discussion in the next meeting on the 11th, with Neil
summarizing the key points in the issue tracker.
<https://cryptpad.fr/#cp-md-0-next-steps>Next Steps

   1. Eri to continue working on CSS features for MathML, including font
   family math and CSS properties for MathML layout.
   2. Eri to coordinate with Fred on WebKit patches for MathML.
   3. Murray to check in his MathML variance polyfill for review.
   4. David to review MathML core issues and create PR for issue 290 to add
   author note about white space trimming/collapsing.
   5. Murray to contact Sergey Malkin and Peter Constable regarding
   OpenType math table issues.
   6. Brian to post meeting notes to the issue tracker.
   7. Neil to summarize discussion about MS element into the issue tracker.
   8. David to review OpenType issues before August 11th working session.
   9. Brian to create calendar invite for August 11th working session.
   10. Team to review and comment on issues labeled "needs tests" and "need
   specification update" before next meeting on August 11.
   11. Team to consider removing MS element from MathML core specification.