Minutes: MathML Full Meeting, 13 Aug, 2026

 Attendees:

   - Neil Soiffer
   - Louis Maher
   - David Carlisle
   - Deyan Ginev
   - Bruce Miller
   - Paul Libbrecht

<https://cryptpad.fr/#cp-md-0-regrets>Regrets
<https://cryptpad.fr/#cp-md-0-agenda>Agenda
<https://cryptpad.fr/#cp-md-0-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports
<https://cryptpad.fr/#cp-md-0-1-1-arxiv-html-intent-update>1.1 arXiv HTML +
Intent Update

Deyan reported that all arXiv HTML articles have now been fully generated
and posted online.
This includes approximately 1 billion math formulas, each with a literal
intent attribute.

“The 1 billion math formulas I keep mentioning are now properly on the live
site, and they have 1 billion intent attributes.”

Literal intent is inserted during LaTeX → HTML conversion. The group
discussed how literal readings behave for symbols such as vertical bars or
perpendicular signs.

Deyan shared updated conversion statistics:

   - 86% of documents convert without errors
   - 98.8% produce some HTML output

He also noted 3,400 open user issues providing valuable feedback.
<https://cryptpad.fr/#cp-md-0-1-2-accessibility-review-of-intent-editor-site>1.2
Accessibility Review of Intent Editor Site

Neil said that Deyan’s intent editor site scored 10/10 on an automated
accessibility checker, with only minor issues (missing labels, missing main
landmark).
<https://cryptpad.fr/#cp-md-0-2-a-href-https-github-com-w3c-mathml-issues-584-horizontal-review-for-security-intent-a->2.
Horizontal Review for Security (Intent)
<https://github.com/w3c/mathml/issues/584>

The group discussed the W3C horizontal security review concerning potential
denial of service risks from deeply nested or extremely large intent
expressions.
<https://cryptpad.fr/#cp-md-0-2-1-directed-acyclic-graph-structure>2.1
Directed Acyclic Graph Structure

David reiterated that intent references form a directed acyclic graph,
preventing recursive expansion loops. Hence the worry that was expressed
isn't an issue.
<https://cryptpad.fr/#cp-md-0-2-2-resource-limits>2.2 Resource Limits

Neil described existing MathCAT safeguards:

   - Reject intent expressions over 1 MB
   - Reject expressions nested deeper than 512 levels

<https://cryptpad.fr/#cp-md-0-2-3-empty-math-element-abuse-scenario>2.3
Empty Math Element Abuse Scenario

Deyan raised a hypothetical scenario: placing a huge intent expression
inside an empty <math> element to hide large content from sighted users.

The group concluded this is mitigated because:

   - Intent cannot contain links
   - Systems may reject oversized intent
   - The spec already prohibits interpreting URLs/scripts inside intent

<https://cryptpad.fr/#cp-md-0-2-4-spec-changes>2.4 Spec Changes

David will draft a PR adding language to Appendix D.5.2 and Section 5.6
clarifying that:

   - Systems may reject intent due to resource limits
   - Failure to expand intent should result in ignoring intent

He will tag the Horizontal Review for Security (Intent)
<https://github.com/w3c/mathml/issues/584> issue when submitting the PR.
------------------------------
<https://cryptpad.fr/#cp-md-0-3-continuation-of-a-href-https-dginev-github-io-mathml-intent-open-editor-open-intent-editor-a-discussion-appropriate-keys-in-yaml->3.
Continuation of open intent editor
<https://dginev.github.io/mathml-intent-open-editor/> discussion
(appropriate keys in YAML)

Deyan reported only incremental progress since last week, mainly
incorporating prior feedback.
------------------------------
<https://cryptpad.fr/#cp-md-0-4-ms-office-no-longer-generates-malignmark-maligngroup-deprecate-drop-from-spec>4.
MS Office no longer generates malignmark/maligngroup — deprecate/drop from
spec

The group revisited whether to remove malignmark and maligngroup from the
MathML spec, since Microsoft Office no longer generates them.

Bruce, Paul, and Deyan supported deprecation.
David noted historical attachment but agreed deprecation is reasonable if
polyfills continue supporting legacy content.

It appears that there are no polyfills supporting legacy content.

Neil will contact Murray Sargent once more before finalizing the decision.
------------------------------
<https://cryptpad.fr/#cp-md-0-5-other-notes>5. Other Notes

   - Paul contributed via chat due to audio issues.
   - David mentioned upcoming deadlines for ISO TR25 and Unicode 18.
   - Charter approval expected August 14.

------------------------------
<https://cryptpad.fr/#cp-md-0-6-action-items>6. Action Items

   -

   Add resource limit language to Appendix D.5.2 and Section 5.6
   *David Carlisle* — Will draft PR and tag the issue.
   -

   Follow up with Murray Sargent regarding malignmark/maligngroup
   *Neil Soiffer*