Minutes: MathML Full meeting, 18 Sept, 2025

 Attendees:

   - Neil Soiffer
   - Louis Maher
   - David Carlisle
   - Murray Sargent
   - Paul Libbrecht
   - Bert Bos
   - Bruce Miller
   - Patrick Ion
   - Murray Sargent
   - Moritz Schubotz

<https://cryptpad.fr/#cp-md-0-regrets>Regrets

   - Deyan Ginev

<https://cryptpad.fr/#cp-md-0-action-items>Action Items
<https://cryptpad.fr/#cp-md-0-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports

*Action Item for the next agenda:* what is the role of "
https://w3c.github.io/mathml-docs/unicode-speech/".
<https://cryptpad.fr/#cp-md-0-2-progress-on-spec-writing->2. Progress on
spec-writing? <https://cryptpad.fr/#cp-md-0-a-some-cleanups>a. Some spec
cleanups

NS says that we do not have a consensus on using images versus polyfills.
<https://cryptpad.fr/#cp-md-0-agenda>Agenda
<https://cryptpad.fr/#cp-md-1-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports

NS: I will be traveling on Thursday, Oct. 2 and Thursday, Oct. 9 and will
not be available for our intent meetings.

*Action Item for the next agenda:* what is the role of "
https://w3c.github.io/mathml-docs/unicode-speech/".

BB believes that we will be granted a six-month charter extension. He is
not sure if we can get a one-year charter extension. He will continue to
try for a one-year charter extension.

On 9/17/2025, DG wrote: Hi everyone,

As of this morning, arXiv's experimental HTML pages are now using an intent
attribute for the first time. Namely, each math element is annotated with
the ":literal" property.

":literal" values are the only reliable starting point for graduate level
texts in STEM, where simple heuristics fail as often as they succeed (if
that). They fit Louis' pragmatic stance of "just show me what the author
has written", while also allowing more informed intent attributes to be
deposited on the subtrees, when known.

Additional intent upgrades are interesting future work for arXiv - and
LaTeXML. We would like to gradually enrich subtrees, whenever we have
sufficient confidence we can infer the mathematical concepts.

These new attributes for arXiv are only starting to trickle in with the
article batches announced daily. It will take a full regeneration of the
collection to have them everywhere. That could happen near the end of 2025,
maybe just in time for the MathML 4 CR phase. We'll see.

Here is one random recently announced article, to illustrate today's
changes: https://arxiv.org/html/2509.12354v1

arXiv updates and MathML Intent from Deyan Ginev on 2025-09-17
(www-math@w3.org from September 2025)
<https://lists.w3.org/Archives/Public/www-math/2025Sep/0005.html>

We had a discussion on how certain characters were pronounced such as an
up-side-down A which means for all, and the integral sign.

From Patrick D F Ion to everyone: Names like LongDoubleLeftRightArrow date
back to trying to find neutral names in the early TeX era and ISO 12083
copied them.
<https://cryptpad.fr/#cp-md-1-2-progress-on-spec-writing->2. Progress on
spec-writing? <https://cryptpad.fr/#cp-md-1-a-some-cleanups>a. Some cleanups

DC worked on some of the cleanup issues. He closed some and pinged other
people about other issues.

NS: Are we still aiming to use polyfills instead of images, especially for
non-core things?

NS: Sometimes MathML will not work because of browser problems.

NS: Firefox, at the moment, doesn't have CSS support, so it means core
doesn't work properly there either.

DC: If we put in an image and not a polyfill, it is like saying that MathML
is not usable.

DC: We should aim to have polyfills. If that is not possible, use images.

PL would do an image because it is the safe way. He says to have both.

MoS says the browsers may work properly in five years. Until then, he is ok
with having images.

DC: We should ask the browsers to get it right.

DC: Chrome works well. The Firefox people merely need to match Chrome.

MoS: We should not mix the tests with the spec. So if somebody wants to
test if it works in his browsers, he can go to the browser test.

BM: It would be good to have images to show what you are trying to get.

From Moritz Schubotz to everyone:
https://www.w3.org/Math/testsuite/mml2-testsuite/index.html

DC: When we started MathML 4, we decided to drop all the images.

DC: We should decide exactly what we want to do with polyfills including
where to house them.

DC: Once we had: if it is green, it is rendered, if it is gray, It is an
image.

DC: Currently we have images for anything not in core.

NS likes to have both.

NS did not find many examples in the SVG specification. Mostly he found
images.

NS says that we do not have a consensus on images versus polyfills.

On another issue: MuS: Is the mfence more general than the mrow? It is,
because you can make the outer parentheses or brackets larger than the
inner ones, and it's easier to see the meaning when you have a difference
in size.

From Paul Libbrecht to everyone: Replying to "Does the mfenced pol..."

Yes! https://hoplahup.net/tmp/mfenced-grows.png

MuS: The Microsoft software makes the outer ones bigger than the inner ones.

DC: Mfences have been defined to be equivalent to mrow for all the time
MathML was released.

MuS: That was an oversight of the original people.
<https://cryptpad.fr/#cp-md-0-b-a-href-https-github-com-w3c-mathml-issues-524-524-naming-the-arguments-to-intent-new-property-needed-a->b.
#524: Naming the arguments to intent: new property needed
<https://github.com/w3c/mathml/issues/524>

NS: DC has a PR for this.

NS: The idea here is what happens when you navigate into something with
intent. Do you just say, in Part 1, in Part 2, or do you say in numerator,
in denominator, because someone used a name, or in this example, in premise
or in conclusion.

You should name what an argument is called.

NS said that this is not mentioned in core spec so AT cannot know about it,
and it would be difficult to translate into another language.

DC: You only know the ones you know.

NS: We could give names for things in the core concept list.

PL: We should spend time on making it possible for authors to include new
roles, new navigation.

BM: Two mechanisms: You can say that for random stuff with random concepts,
you may have names for them using arg name which may not be translatable. A
separate mechanism is to say that an implementer is free to give names.

From Patrick D F Ion to everyone: A good example of semantic nuances with
no definite convention defined to disambiguate!

NS listed the names of the slots numerator and denominator upper and lower
limit.

Ns: We need to go through all the core concepts and see which ones can get
names to describe their use.

NS congratulated DC for working on these open issues.
<https://cryptpad.fr/#cp-md-0-3-a-href-https-github-com-w3c-mathml-issues-449-intent-properties-ordering-amp-references-issue-449-a->3.
Intent Properties: ordering & references Issue #449
<https://github.com/w3c/mathml/issues/449>

Postponed to next week when DG can attend.
<https://cryptpad.fr/#cp-md-0-4-a-href-https-github-com-w3c-mathml-issues-523-523-missing-property-to-indicate-needed-grouping-a->4.
#523 Missing property to indicate needed grouping?
<https://github.com/w3c/mathml/issues/523>

NS: This is not about naming the arguments, but knowing when the
two-dimensional structure of it begins and ends. This is easy to do with
linear equations, but difficult to do with a 2D structure. If it is 2d like
the binomial coefficient, you need to bracket it in some way. Intent does
not have a way to show the beginning and ending of structures. A 2d
expression needs bracketing words.

MuS has a function that says check parentheses to add or remove parentheses
if necessary.

MuS: If you want to see what I've been doing, you can run my little app,
and you can see what I do for all of these cases, and I do make it clear
with a binomial coefficient.

DC: It is difficult to find the beginning and ending of structures with
unknown functions.

From Patrick D F Ion to everyone: Can we envisage the 3-dimensional case
for notations, as in commutative diagrams (or more dimensions)?

MuS: This kind of logic is sort of part of Unicode math, and so, I have a
function that says, check parentheses. And it gets rid of them, or it adds
them as need be. And so, it was pretty easy for me to make a distinction
with a binomial coefficient, whether, you know, if it's n choose k, or a
binomial coefficient n plus 1 choose k plus 2, and binomial coefficient.

DC: This would be difficult to do for unknown open content.

NS: I do not think we can decide on this yet.

DC: If you make a PR, then you are forced to write it up.
<https://cryptpad.fr/#cp-md-0-zoom-meeting-summary-9-18-2025>Zoom Meeting
Summary 9/18/2025 <https://cryptpad.fr/#cp-md-0-quick-recap>Quick recap

The team discussed LaTeX's math command implementation and MathML
attributes, including potential updates to AMS packages and the handling of
literal intents for mathematical expressions. They explored challenges with
Unicode character naming and rendering examples in the MathML
specification, ultimately deciding to include both polyfill renderings and
images for clarity. The group addressed various technical issues related to
mathematical specifications, including argument navigation properties,
translation considerations for mathematical concepts, and challenges with
bracketing and two-dimensional notation, while confirming a 6-month charter
extension.
<https://cryptpad.fr/#cp-md-0-next-steps>Next steps

   - Bert: Follow up with Philippe regarding the charter extension duration.
   - Deyan: Review MathCAT's character-speaking approach and provide
   suggestions for literal intent naming.
   - David/LaTeX team: Consider how to implement intent attributes in LaTeX
   packages.
   - Neil/team: Develop a "literal" column for the Unicode speech page to
   indicate differences between literal and common naming conventions.
   - Neil: Ask Deyan to review MathCAT's character-speaking features.
   - Paul: Add the link to Deyan's message about Archive using literal
   intents in the meeting minutes.

<https://cryptpad.fr/#cp-md-0-summary>Summary
<https://cryptpad.fr/#cp-md-0-latex-math-commands-and-extensions>LaTeX Math
Commands and Extensions

The group discussed LaTeX's implementation of math commands and packages,
with David explaining that while Mathcat doesn't guess interpretations,
LaTeX will include a math command with MathML attributes in the next
release. David noted that while packages could redefine these commands, the
kernel would only define the basic commands, and mentioned that while AMS
packages could potentially be updated, it would require a political
decision rather than a technical one. Bert reported confusion about his
group's charter extension, as Philippe had offered only a 6-month extension
despite earlier indicating they could request more time. The conversation
ended with a brief discussion about progress reports and Deyan's use of
literals in LaTeX, which Bruce and Deyan handle through LaTeXML.
<https://cryptpad.fr/#cp-md-0-debating-mathml-intent-approaches>Debating
MathML Intent Approaches

The team discussed Archive's implementation of literal intents for MathML,
with Neil explaining that this approach generates code directly on the math
element rather than using semantic interpretations. The group debated the
challenges of using literal interpretations, particularly for Unicode
characters where the default names like "Unicode multiplication sign" or
"Unicode for all" would be unwieldy for users. They agreed that while some
symbols like "times" or "for all" have clear alternative names, many
Unicode characters don't have better alternatives, and the team needs to
determine where to draw the line on literal versus semantic
interpretations. Paul suggested this change might be a provocation to
encourage discussion about improving the intent system overall.
<https://cryptpad.fr/#cp-md-0-mathml-rendering-strategy-discussion>MathML
Rendering Strategy Discussion

The team discussed how to handle rendering examples in the MathML
specification, particularly regarding the use of polyfills versus images.
David and Neil debated whether to include both polyfill renderings and
images, with David arguing for including polyfills to encourage browser
implementation, while Paul and Moritz suggested using images as a safer
alternative. The group ultimately agreed to include both polyfill
renderings and images, with the understanding that if polyfills don't work
properly in certain browsers, images would serve as a fallback to ensure
clear display of the intended rendering.
<https://cryptpad.fr/#cp-md-0-math-specifications-and-unicode-discussion>Math
Specifications and Unicode Discussion

The team discussed several technical issues related to math specifications,
including the proper use of minus signs and character encoding. Neil
mentioned fixing a small minus sign issue and noted that AI in Visual
Studio Code automatically suggested using the proper Unicode minus sign.
The group debated the relationship between MFence and MROW, with Murray
arguing that MFence is more general than MROW, though David and Neil
maintained they are equivalent based on existing definitions. The
discussion concluded with concerns about the format of the issues being
reviewed, with David suggesting they might be better separated into
individual issues.
<https://cryptpad.fr/#cp-md-0-argument-navigation-property-implementation>Argument
Navigation Property Implementation

The team discussed implementing a property for argument navigation,
focusing on how to handle argument names and translation considerations.
David proposed using a list of known argument names that could be
translated, similar to core concept lists, while Neil expressed concerns
about translation challenges for custom argument names. Paul clarified that
the primary goal was to support navigation functionality, suggesting that
the property should start with a navigation indicator like "nav" and
include arg names as a special case. The group agreed that the spec should
allow authors to specify custom argument names and that these should
override any automatic translations, though Bruce noted the importance of
avoiding problematic markup that could break rendering.
<https://cryptpad.fr/#cp-md-0-mathematical-term-translation-standards>Mathematical
Term Translation Standards

The team discussed translation issues for mathematical concepts,
particularly focusing on French translations of terms like "lower limit"
and "upper limit." They agreed to implement a static list of named slots
(including numerator, denominator, upper limit, and lower limit) in YAML,
with the option to add more names as needed in the future. Neil committed
to reviewing all core concepts to identify additional names for argument
slots and will proceed with approval after the meeting. The discussion
about Unicode was deferred to a future meeting due to time constraints.
<https://cryptpad.fr/#cp-md-0-mathematical-notation-bracketing-challenges>Mathematical
Notation Bracketing Challenges

The team discussed challenges with bracketing and announcing
two-dimensional mathematical notation, particularly for expressions like
binomial coefficients. Neil explained the difficulty of automatically
identifying when bracketing is needed for unknown mathematical concepts,
while Murray shared his approach of using Unicode math and parentheses
handling functions. The group agreed that while implementing explicit
begin/end announcements (like "begin numerator, end numerator") would work,
it would be cumbersome and could be improved with better syntax. The
conversation ended with a brief discussion about the group's charter
extension, which was confirmed for 6 months with potential for additional
extension.