Minutes_ MathML Core Meeting_ 27 July 2026

07/27/2026 MathML Core Meeting-core#336<https://github.com/w3c/mathml-core/issues/336>
Attendees

  *   Brian Kardell
  *   Louis Maher
  *   David Carlisle
  *   Eri Pazos Pérez
  *   Paul Libbrecht
  *   Deyan Ginev
Regrets
Action Items
1. Announcements/Updates
ACTION: NS: We will discuss the date of the August core meeting by email.
ACTION: Can we delete overview.html in the repo? DC: Yes. He will delete overview after this meeting.
3. Ignore variation selectors in operator category determination<https://github.com/w3c/mathml-core/pull/328>
ACTION: DC will respond to Ignore variation selectors in operator category determination<https://github.com/w3c/mathml-core/pull/328> DC wrote: Thanks for this, The PR has merge conflicts but the intention is clear (it might be better to first have an issue so we could discuss the exact version of the enhancement and then make a new PR rebased against the current head) That said, we discussed this at today's MathML-core meeting (7/27/2026), there was general support for the idea, but we would need to check implementation details to decide if this can be fitted into this cycle or would need to be mathml-core level 2.
4. Browser Status<https://github.com/w3c/mathml-core/issues/320>
Resolution: We should move Browser Status<https://github.com/w3c/mathml-core/issues/320> to a discussion. We should thank the submitters for their work and to say that it's not actionable and that we would like our issues to be actionable. We need to convince people to identify which spec bugs they point to, or which implementation bugs they point to, and write additional web platform tests.
ACTION: In Browser Status<https://github.com/w3c/mathml-core/issues/320>, DC wrote: We discussed this issue again at today's meeting.
In its current form it's not easy to have any action that would enable us to close it.
I am going to convert it to a discussion (which is a general GitHub feature, although not previously enabled for mathml-core). This allows the thread to remain open for comments without having it showing up in the open issues.
It would be good if many of the specific problems shown could be made into individual web platform tests and then issues raised at the appropriate place (either the browser bug reporting sites, or as issues against the mathml-core spec if the spec needs clarifying)
DC converted this issue into discussion #338<https://github.com/w3c/mathml-core/discussions/338>
8. Document why mtable sets displaystely to false<https://github.com/w3c/mathml-core/issues/314>
DC merged add text re displaystyle for #314 <https://github.com/w3c/mathml-core/pull/337> which closed Document why mtable sets displaystely to false<https://github.com/w3c/mathml-core/issues/314>
Agenda Items
1. Announcements/Updates
ACTION: NS: We will discuss the date of the August core meeting by email.
BK Said that Agalia had a program featuring the work done by Microsoft to make office products accessible (https://www.igalia.com/chats/mathml-midnight).
Get an update on the progress Igalia has made for browser compatibility?
________________________________
EP referred to the Slides from Hackfest<https://webengineshackfest-2026-mathml-core-065775.pages.igalia.com/>
and Android Math Fonts<https://issuetracker.google.com/issues/498266576?pli=1>
It covers most of the things. Basically, we are passing many more tests, and we have more tests, so it's a bit of both.
We have been removing some legacy MathML features, especially in Safari. In Safari, most of them are still enabled by default, because we are discussing with Apple the best way of removing them without breaking support for things they have, but it's mostly everything under one feature flag that you can turn for MathML Core. It's enabled by default in the Linux ports, but not in the Apple ones yet.
We made available across all three engines, things like from Family Math, Math Shift, Math Dev. The script level attribute, Math Auto with the text transform. We still haven't fully deprecated the math variant things, because we are waiting on Wikipedia to change their MathML renderer to the new one, hopefully.
You can do it with Math Auto now, and it's under a feature flag that can be disabled in all of the browsers, except in Chrome, which never had it in the first place.
We improved some of the existing things, like some CSS features, like margins, some padding, some borders. We added a right-to-left mirroring to all of the browsers.
There's one thing missing here, which is handling floats and positioned elements. Positioned elements are handled in Safari, but not in Firefox and floats are not handled in any of them according to the spec because it either does a very bad hack that doesn't really work on all cases, and is a lot of code or properly implementing display math.
Implementing Display Math is quite a big task, so it will have to be done in a follow-up, somehow, either by the browsers or something. Once Display Math is implemented, doing both of these things is very easy.
Finally, we did a few different fixes, like rendering unknown elements as MRO, fixing the Cambria Math thing both using the Half Boss fix in all of the browsers instead of each browser doing a different thing.
Right now, we've paused the major work on MathML, because we completed the grant that we had for this year, but we had a coding experience student that is working on the A element support across all of the different browsers. We have made some nice progress in that regard. There are some patches that I can link if you want to see them. But yeah, I think that's most of it.
NS: In Firefox and Safari, did you get the CSS support in there? Because I have some polyfills that depend upon that.
EP: Most of that is there. We worked a bit on improving the layout of all of that. There may still be some browser differences. It would be great if you find a browser difference or something like that, and you open an issue so we can keep track of it.
NS: Do I open it on each browser, like Firefox and WebKit?
EP: Yes if it's one of the browsers clearly not following the spec.
BK: How much of that work for supporting general CSS staff in MathML is done for macOS and iOS releases today?
EP: I think most of these changes landed in WebKit a few months ago.
BK: If you identify something, you should see if there's a web platform test for it. And, if not, if you can create one.
DG: Do you have a grant pipeline for MathML level two work?
EP: Not at the moment.
BK: If anyone is interested in writing grants or has leads for grants, please let him know.
ACTION: Can we delete overview.html in the repo? DC: Yes. DC will delete overview after this meeting.
BK: It would be a pretty huge thing to get Microsoft to join the working group. I would like to get implementations from them.
2. Exhaustively list the attributes accepted by the <a> element.<https://github.com/w3c/mathml-core/pull/335>
BK: Is this list ready?
EP: Yes.
DC merged PR 335.
3. Ignore variation selectors in operator category determination<https://github.com/w3c/mathml-core/pull/328>
ACTION: DC will respond to Ignore variation selectors in operator category determination<https://github.com/w3c/mathml-core/pull/328> DC wrote: Thanks for this, The PR has merge conflicts but the intention is clear (it might be better to first have an issue so we could discuss the exact version of the enhancement and then make a new PR rebased against the current head) That said, we discussed this at today's MathML-core meeting (7/27/2026), there was general support for the idea, but we would need to check implementation details to decide if this can be fitted into this cycle or would need to be MathML-core level 2.
4. Browser Status<https://github.com/w3c/mathml-core/pull/320>
BK: This is a status report. What is the purpose of keeping this issue open.
BK: To be acted upon, it must be split into individual requests.
NS: Can we point AI at this and have individual tests created?
DG: If I can supervise it, it would.
EP: Some of these issues may be already fixed.
NS: All of these issues would make good tests.
NS: Can we device a test that checks to see if the result of an individual action is the same across all browsers? We would not have to supply a reference picture that defined correctness. The test fails if you get different results from different browsers.
PL: Writing tests is a competency that we should all have. Perhaps we should have a meeting on how to write tests.
BK: We had such a meeting a couple of years ago.
BK: Perhaps we should make such a class and have it recorded so that individuals can review it when convenient.
EP: We do not have funding for MathML right now. We do have funding for web platform tests, especially if they fail in Chromium.
EP: If you have Chromium-only failing tests related to MathML, you can send it our way.
BK: What should we do with this.
NS: We need to make web platform tests out of it.
NS: Perhaps the spec is under specified.
EP: Close this issue but keep the discussion open on other issues.
Resolution We should move Browser Status<https://github.com/w3c/mathml-core/issues/320> to a discussion. We should thank the submitters for their work and to say that it's not actionable and that we would like our issues to be actionable. We need to convince people to identify which spec bugs they point to, or which implementation bugs they point to, and write additional web platform tests.
ACTION: In Browser Status](https://github.com/w3c/mathml-core/issues/320), DC wrote: We discussed this issue again at today's meeting.
In its current form it's not easy to have any action that would enable us to close it.
I am going to convert it to a discussion (which is a general GitHub feature, although not previously enabled for mathml-core). This allows the thread to remain open for comments without having it showing up in the open issues.
It would be good if many of the specific problems shown could be made into individual web platform tests and then issues raised at the appropriate place (either the browser bug reporting sites, or as issues against the mathml-core spec if the spec needs clarifying)
DC converted this issue into discussion #338<https://github.com/w3c/mathml-core/discussions/338>
8. Document why mtable sets displaystely to false<https://github.com/w3c/mathml-core/issues/314>
DC merged add text re displaystyle for #314 <https://github.com/w3c/mathml-core/pull/337> which closed Document why mtable sets displaystely to false<https://github.com/w3c/mathml-core/issues/314>
Zoom Core Meeting Summary 7/27/2026
Summary
This was a MathML working group meeting where participants discussed browser compatibility updates, specification changes, and ongoing implementation work. Eri provided a comprehensive update on browser status, reporting significant progress including improved test coverage, removal of legacy MathML features, and implementation of CSS support across browsers. The team discussed several pull requests including one to exhaustively list attributes accepted by the A element (which David merged during the meeting), and another regarding variant selectors in operator category determination. The group also addressed an open issue (320) that contained multiple browser compatibility concerns, deciding to move it to a discussion category rather than leaving it as an actionable issue. David committed to merging a pull request about mTableDefaults display style after receiving approval, and the team noted that current MathML funding had ended in June, though some ongoing work continues with a coding experience student.
MathML Implementation Progress Update
The team discussed removing an outdated overview file from their system, which David agreed to delete after checking it doesn't break the build. Eri provided an update on browser compatibility, reporting progress on MathML implementation across browsers including improved test coverage and feature support, though some legacy features remain enabled in Safari under feature flags. The team paused major MathML work due to completing their current grant, but continued progress is being made on A element support across browsers.
MathML Implementation Progress Discussion
The team discussed progress on MathML implementation, with Eri reporting that while many CSS features are supported in WebKit, some items like accent transforms and minus sign changes still need specification work. Neil inquired about CSS support in Firefox and Safari for his polyfills, and Brian suggested creating web platform tests to identify and address any browser differences. The discussion concluded with Deyan asking about future funding opportunities, to which Brian responded that while no immediate grants are planned, they welcome help writing grant proposals and are exploring implementation commitments from potential new working group members.
MathML Funding and Development Updates
The team discussed funding for MathML work, which ended in June and will be suspended after that time. They reviewed a pull request from Fred that exhaustively listed attributes accepted by the A element, which David approved and merged during the meeting. The discussion also touched on related work being done by Luke regarding interfaces with HTML and SVG, though specific details about this work were not fully explored in the meeting.
MathML and Browser Compatibility Updates
The team discussed three main topics: ignoring variation selectors in operator category determination, browser compatibility issues, and MathML testing infrastructure. For the first topic, they decided to respond to a pull request by acknowledging it's reasonable but needing to evaluate implementation costs. Regarding browser compatibility issues (pull request 320), they agreed to move it to a discussion category since it contains multiple unactionable items and lacks specific testable issues. For MathML testing, Erie confirmed they have funding for general WPT tests, particularly for Chromium-only failing tests related to MathML. The team noted that several members will be on holiday in August, so they decided to keep the next meeting slot open and discuss scheduling via email.