Minutes_ MathML Full meeting_ 25 Sept_ 2025

> 20250925
Attendees:

  *   Neil Soiffer
  *   Louis Maher
  *   David Carlisle
  *   Bruce Miller
  *   Paul Libbrecht
  *   Bert Bos
  *   Murray Sargent
Regrets

  *   Moritz Schubotz
  *   Deyan Ginev
Action Items
1. Announcements/Updates/Progress reports
MoS: Interop 26<https://github.com/web-platform-tests/interop/issues/1140> ACTION: People are invited to give this a thumbs up.
Core meeting on Monday September 29.
The next Intent meeting will be on Thursday, October 16.
3. #100 Simplification of the <semantics> element<https://github.com/w3c/mathml/issues/100>
CONSENSUS: DC: Use polyfills versus trying to get some CSS into a style sheet somewhere.

  1.  Review of Unicode character names<https://w3c.github.io/mathml-docs/unicode-speech/> with emphasis on literal vs semantic names
DC: We will add a literal: text where needed using the existing choose: markup.
Agenda
1. Announcements/Updates/Progress reports
Microsoft added lots more support for MathML in Word and PowerPoint<https://techcommunity.microsoft.com/blog/microsoft365insiderblog/writing-and-presenting-about-math-is-easy-and-inclusive-with-microsoft-365/4439216>
MoS: Interop 26<https://github.com/web-platform-tests/interop/issues/1140> ACTION: People are invited to give this a thumbs up.
\MathMLintent in standard latex-dev release (today)
Core meeting on Monday September 29.
NS: We may have some working group people there to talk about the link issues.
The next Intent meeting will be on Thursday, October 16.
NS: Paul has added German to MathCAT.
2. Progress on spec-writing?
NS corrected a meclose polyfill where he used a hyphen instead of a minus sign.
3. #100 Simplification of the <semantics> element<https://github.com/w3c/mathml/issues/100>
DC: Core just uses the first element.
DC It can be done with CSS.
PL: I would just minimize CSS, because CSS conflicts are super hard to handle.
DC: A js version will help integrate with other polyfills.
CONSENSUS: DC: Use polyfills versus trying to get some CSS into a style sheet somewhere.
4. Intent Properties: ordering & references Issue #449<https://github.com/w3c/mathml/issues/449>
(progress during the week?)
NS: I think that issue 449 is the last significant issue With the spec before we can go to CR. We do have a lot of writing to do.
5. Review of Unicode character names<https://w3c.github.io/mathml-docs/unicode-speech/>
with emphasis on literal vs semantic names
We have a table whose column heads are:
Unicode, Character, Speech Template
DC obtained these values from MathCAT.
NS: MathCAT got the speech template values from their common or legacy versions.
We discussed adding a column entitled "_literal" for the cases when the literal value disagreed with the MathCAT value.
NS thought that this would happen only around ten times. This table has several thousand entries.
NS: We could just add a comment when the MathCAT entry differs from the literal entry.
NS: There is no GitHub issue on this topic.
DC: We will add a literal: text where needed using the existing choose: markup.
Zoom Meeting Summary
Quick recap
The team reviewed several technical submissions and discussed ongoing improvements to MathML support in Microsoft Word, including accessibility features and Unicode support. They explored implementation approaches for polyfills and discussed terminology around mathematical symbols, particularly focusing on the correct representation of multiplication and division symbols. The conversation ended with discussions about documentation processes and concerns about speech generation capabilities in real-time systems, along with some personal travel plans.
Next steps

  *   David to update the issue regarding using JavaScript instead of CSS for polyfills.
  *   David to push the working roll-up for packaging polyfills.
  *   Neil to release Mathcat version 0.7.x before his trip.
  *   Paul to submit a pull request for Unicode Full support in Mathcat.
  *   David to update the YAML table to include literal pronunciations for special characters.
  *   Neil to fix the missing vertical bar entry in the character pronunciation table.
Summary
Design Feature Implementation Challenges
David and Neil discussed the challenges and potential solutions for implementing a specific design feature, expressing concerns about its impact on PDFs and the need to appeal to design sensibilities. Moritz's submission to Interop 26 was reviewed, with Brian's approval to resubmit previous work, aiming to encourage browser collaboration. Neil encouraged others to support the submission by giving it a thumbs up on GitHub.
MathML Improvements and Accessibility Updates
The group discussed recent improvements to MathML support in Microsoft Word, including work by Peter Wu who fixed several issues related to XML namespace tags and mFenced. Murray reported on progress with accessibility features, noting that Mathcat will be used for larger expressions while arrow-key navigation will use his speech engine, with plans to extend announcements for blind users. David announced that MathML and Math from L intent commands have been added to the standard distribution, though they currently do nothing without automatic conversion tagging enabled. Paul mentioned completing Unicode support and preparing a pull request, while Neil indicated he would release a new version (likely 0.7.x) after addressing some navigation issues.
Semantic HTML Polyfills Implementation
The team discussed implementing polyfills for semantic HTML features, deciding to use JavaScript rather than CSS to avoid conflicts and simplify the approach. David mentioned having a working Rollup setup that could serve as a single JavaScript entry point for all polyfills, which he agreed to push to the repository. The group also discussed the possibility of setting up GitHub Pages to provide a CDN for the polyfills, with Paul offering to help with this process.
Character List and Quote Mark Bias
The team discussed two main topics: intent ordering and references, which they decided to postpone until Dan returns, and a list of characters created from MathCAT that shows English language bias, particularly regarding quote marks. They noted that certain quote marks, like the left pointing double angle quote, are rarely used in mathematics and may need to be simplified or removed from the specification.
Mathematical Symbol Terminology Discussion
The team discussed the correct terminology for mathematical symbols, particularly focusing on the "times" symbol (�) and its alternatives like "cross" and "times sign." They debated whether "times" was the most appropriate literal representation, with Murray and David arguing that "times" is the correct term when not used in vector contexts. The discussion also touched on the Unicode representation of division symbols and the presence of a broken bar symbol (A6) in their reference materials.
Handling Pronunciation in Documentation
The team discussed how to handle literal versus non-literal pronunciations in their documentation, focusing on a table of Unicode characters. David suggested adding a flag or comment marker in the YAML file to indicate when the literal pronunciation differs from the common pronunciation, rather than creating a separate column for literals. Neil confirmed that most characters have semantic meanings rather than literal descriptions, with only about 10 special cases where the literal pronunciation differs. The team agreed this approach would be the easiest solution, with David planning to implement it in the YAML file.
Generative AI Limitations and Applications
The group discussed a table-generation process, with David confirming he likely created it by manipulating YAML data. Paul raised a question about speech generation in courses, expressing concern about the lack of navigation capabilities with generative AI. Neil clarified that real-time systems require responses within 100 milliseconds, making generative AI unsuitable for such applications. The conversation concluded with Neil sharing details about an upcoming trip to Japan, including dietary restrictions and concerns about monsoon season.