Minutes: MathML Full meeting 2 Dec, 2022

 Attendees:

   - Neil Soiffer
   - Louis Maher
   - David Farmer
   - Bert Bos
   - Deyan Ginev
   - Bruce Miller
   - Steve Noble
   - Dennis Müller

<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.0-rc2#cp-md-0-regrets>
Regrets

   - David Carlisle

<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.0-rc2#cp-md-0-agenda>
Agenda
<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.0-rc2#cp-md-0-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports

NS: The beta of chrome, containing MathML, has been released. MathML works
in Edge beta also. The production version of Chrome with MathML should be
released around January 10, 2023.

NS: Things are not perfect. Firefox has not implemented all the MathML
features. Neither has Safari.

DG: Says this is fantastic news. He has already submitted Chrome MathML
bugs. So has NS.

SD: I can save a document, containing math, from the Chrome print option,
into a pdf document. The math is readable in the pdf document.

PL: Other browsers, which are based on Chromium, are picking up these
changes.
<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.0-rc2#cp-md-0-2-deyan-will-present-what-he-calls-his-a-href-https-dginev-github-io-mathml-docs-intent-to-content-simple-algorithm-for-producing-content-near-expressions-from-intent-a->2.
Deyan will present what he calls his simple algorithm for producing
content-near expressions from Intent
<https://dginev.github.io/mathml-docs/intent-to-content>

DG: This note suggests a two-step algorithm for transforming intent
expressions into Content MathML.

DG said that this was a preliminary draft, and he thanked DC, NS, and BM
for their feedback.

DG: walked us through the document.

DG: would rather not use hints. He would like intent to give the full
meaning of the expression.

LM: What do underscores do.

DG: If you have an underscore in front of something, that thing is part of
the language, and you just speak it. You do not have to understand it.

DG: Start with intent that is close to speech, which is the problem we are
trying to solve first. Then form more complex expressions.

NS: What do you do with expressions that do not have the standard meeting
such as sin to the minus one power? The program might interpret the
superscript minus one as a power instead of an inverse.

DG: It is up to the author to give clear guidance as to the meaning of the
superscript. If the author wants it treated as an inverse, then he can
annotate the expression to say that.

NS: There is a back-end processing step which cleans things up. The back-in
process can be used to correct errors in the content.

SD: The need for that back-end step is not really such a big obstacle,
simply because anytime a user is interacting with content MathML, they're
probably not going to be very accurate in the way in which they enter the
content MathML; therefore, a clean-up process will be needed.

SD: So that their meaning will be clear, imaginary numbers may need extra
annotations just like the exponentials do.

PL: It is nice to have a content generator accept that we have not yet
stabilized our intent grammar. We do not want DG's algorithm to affect our
it intent grammar. We must keep in mind that the primary use of intent is
to make things accessible.

DG: means for his algorithm to help with content and internationalization.
He does not mean for his algorithm to give normative guidance to intent.

PL: What if an author uses intent to help speech, but the output is not
content compatible? Should we have two intents—one to generate speech, and
another to generate content?

SD: This is the situation we have now. the author must know the consumer's
needs. If the author wants to generate speech and content, then he must use
the proper intent language that will produce both.

DG: It might be difficult to have validators. There is a worry that every
consumer of content MathML would decide on what intent he needs to support
his project.

SD: The author has one target in mind either speech or some math engine.

BM: In actual practice, there will be people who barely get MathML. Their
first intention will be to do as little accessibility tweaks as possible.

BM: Extracting content is a bonus you get for free.

DF: People will not hand write MathML. They will use a system to generate
MathML and change the input to produce the output correctly.

BM: People will be concerned with accessibility and not content.

NS: DG should write a note that will be publicly released when we release
MathML 4.

DG: Are people ok with this algorithm not being normative?

PL: The version number of DG's algorithm should not match the version
number of MathML. We want to update the algorithm without causing a number
change in MathML.

We discussed units and chemical elements.

NS asked the group if we should be working on the "isa" grammar along with
our normal intent descriptions.

DG: briefly covered his ideas about incorporating "isa" or "type" into
"intent" via adding a ":" syntax as is used in some computer languages. For
example: intent="m:unit" on an mi element.

NS: Next week, we will talk about deprecating or keeping mstyle. After
that, we will discuss "isa".

SD: Cannot mstyle be replaced with mrows using the same attributes?

MUS was assigned a bug saying that Voiceover did not tell users about
fonts. He wondered if Voiceover should be able to indicate fonts and text
sizes.

CS and LM said no. They said that each letter would have to be scanned
individually to get text sizes and fonts. While this might be done
infrequently to make proper headings, that in general it was not done.

There was a discussion about using color to emphasize certain areas of a
document. The blind are almost totally incapable of finding color changes
in a document without scanning and interrogating each word, or character,
individually.

MUS pointed out that a bold "B" can be different from just a "B". The Bold
"B" might be a vector. Changing the font of a character might change the
character. Using font styles to indicate the nature of a character is very
unfortunate to the totally blind.

DG: said that some math uses color.

NS: suggests that we should investigate the use of CSS generated color.

NS will prepare sample text for LM to see if CSS generated color and font
sizes can be detected.

LM: It can be detected if the blind user is willing to scan, and
interrogate, each letter and word individually.

Next week: columnspan->colspan
<https://github.com/w3c/mathml-core/issues/180>, mstyle, and isa