Minutes: MathML Full Meeting, 21 May, 2026

 Attendees:

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

<https://cryptpad.fr/#cp-md-0-action-items>Action Items
<https://cryptpad.fr/#cp-md-0-3-a-href-https-github-com-w3c-mathml-issues-576-horizontal-privacy-review-feedback-discussion-a->3.
Horizontal Privacy review feedback discussion
<https://github.com/w3c/mathml/issues/576>

*ACTION:* In response to the privacy group's comments: NS will thank them
for their thorough review, and he will say that we agree with their
comments, and we will include all of their issues in the appropriate
sections. We will also add some clarifying language elsewhere.

*ACTION:* BB recommended to NS that NS contact the chairs of groups who
should have sent us comments, but have not sent them.
<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-0-2-charter-status-update>2. Charter status
update

BB: I can send out the message to the AC that the official review has
started. That will probably be in the next few days, and then the AC review
will last for four weeks. And unless there are formal objections, which I
don't expect, we can restart the working group.

PL told BB that Brian Kardell is in the technical architecture group. BB
could contact Brian if BB needs to talk to that group.
<https://cryptpad.fr/#cp-md-1-3-a-href-https-github-com-w3c-mathml-issues-576-horizontal-privacy-review-feedback-discussion-a->3.
Horizontal Privacy review feedback discussion
<https://github.com/w3c/mathml/issues/576>

Privacy review Finding 1 said that we could not rely on the core
specification because the full specification added features which are not
present in core such as intent. They were referring to sections D.4 and D.5
in the full specification.

Finding 2 — href on all MathML elements reintroduces link-model risks
outside Core

Finding 3 — AT-use detection via intent divergent content (primary new
privacy concern) The W3C Security and Privacy Questionnaire explicitly
flags features that allow authors to serve different content to AT users as
a privacy concern, because sites can infer AT use from subsequent user
behavior.

We can use the language concerning alt text for this issue.

Finding 4 — intent requires explicit non-observability guidance MathML Core
reserves intent and arg as valid attributes but does not define their
processing behavior. As a result, MathML Core’s privacy review does not
cover their privacy implications. MathML 4 should therefore add explicit
privacy guidance for intent.

DG says we could put some of the necessary text into the main part of the
spec and not the intent. We should consider putting the corrected text in
two places, one such place would be the appendix.

PL: We can put it in two places so that the critics can know their question
was resolved.

Finding 5 — intent literals should be safely handled in speech and braille
pipelines

We may just use their language.

Finding 6 — intent processing should not expose user locale or AT
preferences MathML 4 introduces author-provided intent values that may be
interpreted differently depending on language, locale, speech rules,
braille rules, or assistive-technology preferences.

NS: Okay, they have a suggested addition to the text.

Finding 7 — Clarify fetch behavior for external annotation references
MathML 4 allows annotation and annotation-xml elements to reference
external annotation content using src. The specification appears to discuss
this mainly for processors that expand, export, or transform annotations,
rather than for ordinary visual rendering. However, because src is a
URL-bearing attribute, MathML 4 should explicitly define when, if ever,
these external references may be dereferenced in web contexts.

Finding 8 — mglyph adds external image resource loading outside Core mglyph
is not in MathML Core. It includes a src attribute for external glyph
images, and the spec notes a JavaScript polyfill implements it using img.
This creates image-like network requests not present in Core's baseline.

Finding 9 — Content MathML semantic identifiers should not be resolved
automatically Content MathML is outside MathML Core and introduces semantic
identifiers such as definition URL, cd, and csymbol. These identifiers can
refer to external or application-defined semantic definitions. While such
references may be useful for specialized tools, MathML 4 should clarify
that web user agents must not automatically resolve or deference them
during ordinary parsing, rendering, or accessibility processing.

Requested addition to §D.4: Content MathML semantic identifiers such as
definitionURL, cd, and csymbol should be treated as opaque identifiers in
web contexts. User agents must not automatically fetch, resolve, or
dereference them during parsing, rendering, or accessibility processing
unless an application explicitly requests such resolution subject to the
host environment’s normal fetch and privacy controls.

*ACTION:* In response to the privacy group's comments: NS will thank them
for their thorough review, and he will say that we agree with their
comments, and we will include all of their issues in the appropriate
sections. We will also add some clarifying language elsewhere.

BB says the internationalization group is sending us comments.

BB: The internationalization should be the last group to send us comments.

*ACTION:* BB recommended to NS that NS contact the chairs of groups who
should have sent us comments, but have not sent them.

DC sent a link that was producing an odd message on his browser.

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

DC said that this was a Firefox issue.
<https://cryptpad.fr/#cp-md-0-zoom-intent-meeting-summary-for-5-21-2026>Zoom
Intent Meeting Summary For 5/21/2026 <https://cryptpad.fr/#cp-md-0-summary>
Summary

The meeting focused on reviewing the status of the MathML 4 charter process
and addressing privacy considerations raised by the privacy working group.
Bert reported that the charter extension announcement was sufficient and
the official review process would begin in the next few days, with a 4-week
review period expected to conclude by the end of June. The team discussed
privacy and security concerns raised in the review, including issues with
href attributes, intent attribute handling, and Content MathML semantic
identifiers. They agreed to accept all privacy recommendations and update
sections D4 and D5 of the specification accordingly. Bert also mentioned
that the internationalization working group would provide three additional
issues regarding language handling in the intent attribute.
<https://cryptpad.fr/#cp-md-0-charter-review-and-development-updates>Charter
Review and Development Updates

The team discussed the status of charter review, with Bert confirming that
the announcement to membership was sufficient and the 4-week AC review
process would begin in the next few days. David reported merging pull
requests for round-hand transferring script variants and updating the
MathML full spec on GitHub. Paul raised a question about the polyfill
roll-ups, noting that the generated file doesn't fire immediately as
intended in the demo version.
<https://cryptpad.fr/#cp-md-0-mathml-polyfills-implementation-discussion>MathML
Polyfills Implementation Discussion

The team discussed implementing polyfills for MathML, with Paul explaining
that different versions would be needed and questioning whether users
expect immediate application or packaging through platforms like NPM or
CDNs. David suggested using GitHub as a hosting solution, while Paul
advocated for local packaging over CDNs due to privacy concerns. The group
agreed that Paul would package the four proposed versions with a README
providing a simple URL for implementation.
<https://cryptpad.fr/#cp-md-0-mathml-privacy-and-security-clarifications>MathML
Privacy and Security Clarifications

The team discussed feedback on sections D4 and D5 of their specification
regarding privacy and security considerations. The reviewers were concerned
that they didn't provide enough explicit information about features,
particularly regarding href on MathML elements and link navigation. The
team identified that they need to clarify how href in MathML relates to the
core link model and ensure it doesn't create navigation schemes beyond what
the host environment allows. They agreed to add more explicit information
about how href works, potentially by explaining it as equivalent to adding
an M-row around the element and an A tag in core, ensuring only
core-allowed behaviors are permitted.
<https://cryptpad.fr/#cp-md-0-mathml-security-concerns-discussion>MathML
Security Concerns Discussion

The team discussed security concerns related to MathML, focusing on
removing href attributes and adding guidance to prevent injection attacks
and behavioral profiling. They agreed to update the intent section and
privacy section (D4) with guidance similar to what exists for image alt
tags in HTML, preventing assistive technologies from exposing intent
consumption signals to page scripts. The team also addressed concerns about
safely handling intent literals in speech and braille pipelines to prevent
potential injection risks, with Microsoft expressing particular security
concerns about these issues.
<https://cryptpad.fr/#cp-md-0-content-mathml-annotation-references>Content
MathML Annotation References

The team discussed handling external annotation references in Content
MathML, with Paul suggesting a first pass to ensure all relevant elements
are sanitized. They agreed that mglyph elements should be converted to
image tags using the same approach as for links. The group also confirmed
that Content MathML semantic identifiers should not be resolved
automatically, aligning with current practices where only interested
parties should handle these references.
<https://cryptpad.fr/#cp-md-0-privacy-and-internationalization-recommendations>Privacy
and Internationalization Recommendations

The team discussed accepting privacy group recommendations and agreed to
include them in the sections with clarifying language. Bert reported that
the internationalization group will send three issues regarding language
determination and character replacement, which the team will address. The
group noted that security and other groups that didn't submit reviews will
be contacted to check if they have any additional feedback.