Minutes_ MathML Core Meeting_ 30 March 2026

03/30/2026 MathML Core Meeting-core#319<https://github.com/w3c/mathml-core/issues/319>
Attendees

  *   Brian Kardell
  *   Neil Soiffer
  *   Louis Maher
  *   Eri Pazos Pérez
  *   Bert Bos
  *   David Carlisle
  *   Deyan Ginev
  *   Paul Libbrecht
  *   Bruce Miller
Regrets
Action Items
2. charter<https://w3c.github.io/mathml-docs/charter-2026.html> discussion
ACTION: BK would look into the charter to see if he wanted to make evergreen recommendations.
CONSENSUS: NS: In the charter: Delete linking and change that to Line breaking, and delete the line of c's. Then send a note to Bert asking Bert to submit the charter.
3. Some visiting guests with questions...
ACTION: BK: It would be good if you could convince someone at Microsoft to join the group. PW: We will discuss that internally.
4. Review new issues
User Agent style for href hyperlinks and focusable elements #128<https://github.com/w3c/mathml-core/issues/128>
ACTION: Update the spec with actual user agent styles for links now that we have links.
Any feedback from browser developers for: combining chars<https://github.com/w3c/mathml-core/issues/311>
ACTION: EP will investigate this.
Add a template for reporting issue #318<https://github.com/w3c/mathml-core/issues/318>
ACTION: BK: People should put comments on this issue to say what we will need, then BK will make a template for this.
Document why mtable sets displaystyle to false #314<https://github.com/w3c/mathml-core/issues/314>
ACTION: DC will put a line in the pr why displaystyle is defaulted to false, then he can close this issue.
Agenda Items
1. Announcements/Updates
2. charter<https://w3c.github.io/mathml-docs/charter-2026.html> discussion
NS: Our charter runs out in the middle of April. In the MathML full meeting we agreed to renew the charter.
NS reviewed the changes that were discussed in the March 26, 2026, intent (full) meeting.
At PL's request, NS added: Continued development of the polyfills to bridge the current gap.
BK asked NS to fix some linking issues.
NS to BK: Did you want to discuss some evergreen recommendations? Do we need to change the text in the charter to echo this?
ACTION: BK would investigate the charter to see if he wanted to make evergreen recommendations.
BK found a line of c's in the charter.
PL: The charter says that the group is expected to be guided by the following documents, ethical web principles, privacy principles, and web platform design principles, and those are technical architecture group designs documents, and it appears they do not have one about accessibility.
CONSENSUS: NS: In the charter: Delete linking and change that to Line breaking and delete the line of c's. Then send a note to Bert asking Bert to submit the charter.
3. Some visiting guests with questions...
PW: I am Peter Wu from Microsoft, and Brian invited me to this meeting. I have been working on improving the math support in PowerPoint and Word and other Microsoft 365 apps in the past year, and we made a lot of improvements to MathML import and export and integrated it with our HTML import.
And now, like, in the range of 100 million math equations are copy-pasted into Word and PowerPoint each month.
PW: I am trying to determine how far along in the adoption curve are various websites in adopting MathML, and moving away from other ways of rendering math that were around before the browsers supported it natively such as rendering HTML with lots of CSS and SVG. Are there other blockers?
PW: Is there anything we can do to speed up the adoption process?
PW: So, for example, I think one of the discussion topics we had was that Wikipedia uses MathML, but they use SVGs for their display representation.
PW: And there was a list of blockers for them to switch to native MathML for rendering.
DC: You can turn that on if you have an account there for there is a preference to see MathML, but it is off by default.
EP: But they have been working for a few years on a new MathML-only renderer, but it is not finished.
NS: Certainly there's things that need to be improved in Chrome's rendering of MathML that are blockers.
PL: The biggest enemy is aesthetics.
NS: Murray has several bugs that have been reported, and so they just need to get fixed.
PW: I have done a little comparison of MathJax versus native Math in Chromium and things are quite a bit more spaced out with the native rendering.
PW: In general, you get larger glyphs that are easier to read with MathJax, because they are a little tighter-spaced.
DC: There's alignment in tables, which Core does not really do, and CSS kind of does.
NS There are superscript positioning issues in the native rendering such as that for accent marks. MathJax does it better.
DG: In Archive We have 1 billion math formulas on our site in MathML. We are rendering them in MathML Core without MathJax. So the good news that I can give you is that everything is very well documented at this point, so pretty much anytime something has annoyed somebody, a report has been filed. There are tens of issues that need to be resolved.
DG: The native MathML is almost as good as MathJax or anything else.
PW: We unfortunately do not have equation numbering in Office.
DC: It is easier to put equation numbers where mathematicians expect the numbers to be in html than in the native rendering.
NS: He wants to get the baselines aligned on the math label and that is not easy because math is not text.
BK: We should capture that in an issue if we have not already done that.
NS: That was a MathML issue and not a core issue.
DC: Between DG and me, we have a lot of examples of the baseline not lining.
ACTION: BK: It would be good if you could convince someone at Microsoft to join the group. PW: We will discuss that internally.
4. Review of new issues
User Agent style for href hyperlinks and focusable elements #128<https://github.com/w3c/mathml-core/issues/128>
BK: This is an old issue about styling and hyperlinks. It is not level two anymore; it is getting done.
BK: So, I believe that the spec just has a default UA style sheet, and it is probably using Focus right now instead of Focus-Visible.
BK: I worked on this with Alice Boxhall. It is an accessibility feature. So, when browsers first did focus, The rules when you draw the focus indicator were not trivial.
BK: This has never felt right to users which is why browsers did not do that natively. And there never was a way to plug in to the correct heuristic., So Focus-Visible is that. So, if we are using Focus in the default UA style sheet, it's almost certainly it should be Focus-Visible instead.
NS: So, the user-agent style sheet needs to be updated.
BK: It should be updated with links now that we have links.
BK: We must leave the issue open.
ACTION: Update the spec with actual user agent styles for links now that we have links.
BK: Remove level two from the issue labels.
BK: update the default style sheet to include user agent styles for hyperlinks.
Any feedback from browser developers for: combining chars<https://github.com/w3c/mathml-core/issues/311>
ACTION: EP will investigate this.
Add a template for reporting issue #318<https://github.com/w3c/mathml-core/issues/318>
BK: So, Fred is just suggesting that it would make sense to have a template for opening issues on this task to make sure that we are capturing the things that we need.
ACTION: BK: People should put comments on this issue to say what we will need, then BK will make a template for this.
NS: We should have people say what browser and operating system they are using. People should add which font is being used, and if a font is forced or default.
NS put these comments on the issue.
Problems with "Root with index" #317<https://github.com/w3c/mathml-core/issues/317>
Unclear how to use Operators_fence special table #315<https://github.com/w3c/mathml-core/issues/315>
Document why mtable sets displaystyle to false #314<https://github.com/w3c/mathml-core/issues/314>
DC: You always enforce compact rendering unless explicitly told otherwise.
NS: Nothing in core says why it is defaulted to false.
ACTION: DC will put a line in the pr why display style is defaulted to false, then he can close this issue.
DC You must set it on for displayed equations.
Clarify when token elements should fall back to non-MathML layout #313<https://github.com/w3c/mathml-core/issues/313>
5. Status of HTML Or SVG Or MathML Element
Consider renaming HTML Or SVG Element to HTML Or Foreign Element what/html#4702<https://github.com/whatwg/html/issues/4702>
Zoom Summary for The Core Meeting Of 3/30/2026
Summary
The meeting focused on discussing MathML implementation and adoption, with Peter Wu from Microsoft sharing insights about their integration of MathML into Office apps, which now handles hundreds of millions of equations monthly. The group discussed various technical challenges preventing wider adoption of native MathML rendering, including aesthetic differences compared to MathJax, alignment issues in tables, and accent positioning problems. The team reviewed their charter, which needs to be reauthorized by mid-April, and agreed to modify it by removing "linking" and adding "line breaking" to Level 2 features, while also addressing a typo. Several technical issues were discussed, including combining characters, styling for hyperlinks, and the display style behavior of mTable elements, with the group deciding to update the default user agent stylesheet and create a better issue reporting template.
MathML Support and Adoption Discussion
Peter Wu from Microsoft presented on their recent improvements to MathML support in Office 365 apps, reporting hundreds of millions of equations processed monthly through their new HTML import feature. The discussion focused on understanding barriers to wider MathML adoption on websites, particularly regarding Wikipedia's transition from their current SVG and legacy MathML renderers to a new MathML-only renderer following MathML Core standards. The group noted that while Wikipedia has been developing this new renderer for several years, there is no clear timeline for implementation, and some technical challenges remain around MathVariant deprecation.
MathML Support Improvement Discussion
Peter Wu from Microsoft discussed improvements to MathML support in PowerPoint and Word, noting that approximately 100 million math equations are copied into these applications monthly. The discussion focused on challenges with native MathML rendering in browsers, particularly aesthetics and spacing issues compared to MathJax. Neil and David highlighted specific technical problems including table alignment, accent positioning, and superscript issues that need addressing. Deyan from Archive reported on their experience rendering 1 billion math formulas in MathML Core without MathJax, indicating that while there are high tens of issues to resolve, most problems are well-documented and the rendering is nearly usable.
Math Equation Display Challenges
The team discussed challenges with displaying math equations in browsers, particularly regarding HTML tables used for equation numbering and their limitations in Word and PowerPoint. David and Bruce explained that while HTML tables work for equation positioning, they create formatting issues when pasted into these applications. The group agreed that moving toward CSS grid could improve the display, though current CSS capabilities have limitations with math baseline alignment. Brian suggested creating a new issue to better capture these display challenges and explore potential solutions, including making CSS more powerful for math formatting needs.
Work Group Charter Update Discussion
The group discussed rechartering their work, with Neil presenting an updated charter that expires in mid-April. The proposed deliverables include continued development of polyfills, MathML Core Level 2 features, test suites, and implementation reports, with a timeline suggesting completion of test suite and implementation reports by November and recommendation status by December of the following year. Brian noted that linking, previously mentioned as a Level 2 feature, is in Level 1, and Neil agreed to fix this in the charter. Peter Wu agreed to discuss internally about having someone from Microsoft join the group.
Charter Review Discussion Meeting
The group discussed moving forward with a charter review, with Neils seeking approval to proceed. Brian identified a typographical error in the text and suggested checking about "evergreen" language in the charter. Paul raised concern about the lack of an accessibility guiding document in the charter, though it was confirmed this was boilerplate text and no changes were needed. The conversation ended with Brian asking about wrapping functionality for Level 2, though this point was not fully addressed in the transcript.
Text Formatting and Link Issues
The team discussed several issues including line wrapping functionality and combining characters. They agreed to modify the document by replacing linking with line breaking and fixing a typo involving multiple Cs. Eri reported that while they investigated combining character issues, they need to reach out to browser vendors for more information and will do so next month. The discussion then shifted to issue 128 regarding styling hyperlinks, with Brian noting that WebKit has recently changed to use Focus-Visible instead of Focus, which may require updates to the default UA style sheet.
User Agent Stylesheet Updates
The team discussed updating the user agent stylesheet to include styles for hyperlinks, with Brian explaining that Focus-Visible is an accessibility feature that should be used instead of the default focus pseudoclass. They agreed to leave the issue open and update the spec with actual user agent styles for links. The group also briefly touched on extending the HTML or SVG element to include MathML, with Eri planning to follow up with Luke about the status of this work. Finally, they noted that issue 314 appeared to be fully resolved, though they needed to decide whether to include anything in the spec.
Mtable Display Style Setting Discussion
The team discussed the display style setting in mtable, specifically why it is set to false by default. David agreed to add a line in the PR explaining the rationale, either by referencing standard tech layout or noting that matrices and arrays are typically presented more compactly. The group decided this addition would help clarify the specification and allow them to close the related issue.
Issue Template Creation Planning
The team agreed to create a template for opening issues to ensure relevant information is captured. David volunteered to create the template, with input from Neil about including browser type, operating system, and math font information. Neil was tasked with adding comments about these missing details to the template. The conversation ended with plans to review and comment on the template, and Bert was informed he would receive an update about pushing out relevant information later.