Minutes: MathML Full WG 23 Feb, 2023

 Attendees:

   - Neil Soiffer
   - Louis Maher
   - David Carlisle
   - Bruce Miller
   - Dennis Müller
   - Bert Bos
   - Sam Dooley
   - David Farmer
   - Cary Supalo
   - Deyan Ginev

<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.2-0#cp-md-0-regrets>
Regrets

   - Steve Noble
   - Steven Watt

<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.2-0#cp-md-0-agenda>
Agenda
<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.2-0#cp-md-0-1-announcements-updates-progress-reports>1.
Announcements/Updates/Progress reports

LM: Due to AT issues with the Zoom transcript system, I dropped out of the
meeting from 12:35 PM until 12:53PM. There was no transcription during this
outage. Because I could not pick up the thread of the discussion when I
reentered the meeting, I pasted an edited version of the restarted
transcript into these minutes from 12:53 until the end of the meeting.

LM: I am going to investigate a third-party transcript provider called
(Tactiq)
"https://tactiq.io/learn/save-a-transcript-of-zoom-local-recording".

*ACTION:* NS will invite other AT developers to our meetings.
<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.2-0#cp-md-0-2-continue-intent-discussions-lot-of-great-discussion-on-github-that-is-getting-to-the-core-of-intent-pun-intended->2.
Continue intent discussions. Lot of great discussion on github that is
getting to the core of intent (pun intended):
<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.2-0#cp-md-0-a-david-farmer-39-s-intent-list-ideas-action-item-from-last-week->a)
David Farmer's intent list/ideas (action item from last week)

DF: is developing a system called "Space Math".

DF: He has an easier way to write math than LaTeX or ASCII math, but it is
backward compatible with both. He Showed some examples.

DF: showed a display where you enter the math, you see what the AT would
speak, you see the MathML grammar, and you see how the MathML grammar is
rendered in the Firefox browser. There are no intents on this.

DF: showed an example with a plus sign. He said it did not need intents.

In his system, the space character means something. He showed (a b) which
he said was "a" times "b".

He chose to use HTML entities and not Unicode characters. If people think
that is not a good idea, please let him know.

His next example involved "5 by n". He meant "by" to be interpreted as
times. He did not put an intent on it.

DF is trying to stick with core.

His next example involved absolute values and determinants.

We discussed putting the intent on the "mrow" as opposed to putting the
intent on the one character down at the bottom.

From Deyan Ginev to Everyone: <mrow intent="_($a, _by, $b)"> ...

DF: wondered how those two positions would be different.

NS: Around two years ago, we decided to push intent as far down the tree as
is possible. But you cannot always do it.

DC: said that in the "by" example, you could put it in both positions.
However, in the case of the absolute value, you had to put the intent early
on the mrow.

DC: To be consistent, you might always put the intent on the mrow.

DF: is trying to stick to core where things are better defined.

DF: His next examples involved superscripts. He considered the case of "RR".

From Deyan Ginev to Everyone: <msup intent="_($base, $script)"> ... </msup>

DF: put RR and asked how it would be pronounced.

DF: also considered the case of H sub 25 where he was referring to the
matrix "H" row two column five and not H sub twenty-five.

From David Carlisle to Everyone: <mn>25</mn> <mn>2</mn><mo>&invisiblecomma;
</mo><mi>5</mi>
<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.2-0#cp-md-0-b-referring-to-an-element-39-s-content->b)
Referring to an element's content (

#426 <https://github.com/w3c/mathml/issues/426#issuecomment-1437308818>,
#436 <https://github.com/w3c/mathml/issues/436>

NS: Well, we've had a lot of great conversation on the GitHub among at
least DC, DG, BM, and me.

LM: If the AT sees a word starting with an underscore, should the AT read
it literally, or should the AT ignore it?

NS: The AT should read it literally, ignoring the underscores, unless the
AT has a reason to over ride it.

LM: How do we reference the current element.

DC: You can use "this", "self", ":", "*", or nothing. This is not in the
spec. It is part of our current discussions.

DC: had assembled many parallel intent grammars in one document. He wanted
to discuss them and show his own grammar. This document was not in his main
branch.

DC: put name (like absolute value) then put its properties. Put a colon
between the name and the properties. Then you could attach a hint after the
properties.

DC: discussed using "*" to indicate something like "self".

*NOTE* LM: At this time (12:35 PM) I had technical difficulties and was not
able to follow the meeting. The transcript for this time was also lost. I
rejoined the meeting at 12:53 PM, but I lost the thread (intent) of the
discussion. For this reason, I am going to paste an edited version of the
transcript. from around 12:53 PM until the end of the discussion.
<https://sandbox.cryptpad.info/code/inner.html?ver=5.2.2-0#cp-md-0-begin-transcript>Begin
Transcript

SD: But we do have a model to draw from in terms of what we, how we can
decide, such as well.

NS: Yeah, so you know, we can come up with an ordering.

SD: This is good and is clever enough to get around whatever syntax we come
up with. And, you know, seems like we're going to have to deal with that,
anyway.

BM: I do fear that if we restricted it to one property, then we're going to
gradually feel pushed to provide a rather long list of compound names.

NS: Yeah, Cartesian product type, I think.

BM: Yeah, and you know things that a preferred core list of properties.
Otherwise, people the multiple gives people the option of saying things
like well, the unit length as I unit, or English unit, or whatever it is,
and kind of cover themselves with plenty of options.

BM: But if they don't, if they only have one option, then they better
choose wisely.

NS: The issue is, what if someone says that I unit and English unit?

BM: Yeah. Well, garbage?

NS: We must resolve this.

BM: Do we?

NS: Yeah, because I must implement something.

NS: Do you want me to implement something like the spaces thing?

NS: Do you want me to implement it one way and someone implements it
another, and it works in my system and not their system?

NS: I don't think we want that.

BM: Well, I've thought that in this fine environment, reproducibility makes
a lot of sense in mathematical core where we are. You know, we have an
experience with this stuff to finally PIN down, this is exactly what is
right.

BM: So, a little bit of sloppiness seems allowable from my perspective.

DC: Someone is going to ask what you do for units in chemistry.

NS: So again, we're proposing something that imposes implementation costs
and spec costs. I hope we can get this into a Github issue.

DG: So, yeah, one comment is that I have tried quite meticulously to leave
examples and very complete technical counterproposals to a lot of the
things that were said today in the Github shows.

DG: So, if anybody is interested in counterproposals, please read the
Github issues.

DG: There's a lot of comments on the naming of Isa values or property
values. That is why the original is a suggestion of had basically a
guideline of what the name represents. So, the way I was thinking of open
values as encyclopedic concepts. And yes, those get infinitely granular. If
you really press your luck and want to say too much in one type. So, I gave
in the chat, an example of the an odd, prime natural number is technically
a very valid property of natural numbers that you can say they have to be
able to have to be natural, and that could be an isa value, a single value
the problem is that that satisfies one of the 2 nice things you can do with
these are so. One was an accessible description, where an unsided user can
press a shortcut and hear that back out. So that is always valuable, even
if it's that weird. So, you will have to start doing ontology work on top
of these values, which I think nobody wants to do, and that is in favor of
Bruce's suggestion to have multiple simple ones.

DG: This is a matrix which happens to be diagonal, so call them Colon, and
you just have many values. And then you can do simple checks for each of
the values to say, okay, is it a matrix? And then the diagonal matrix, the
identity matrix and all the other upper left triangular matrix would still
have a single column matrix so that they're all easy to spot rather than
you must have this list of possible matrices that you need to then downcast
to a matrix type so it's going to be messy either way, and I think this is
a suggestion from last week. It requires some thought. So, I think that
meets the most thinking, since it's the newest suggestion. A lot of the
other discussions are older. And yeah, that's it.

NS: You know it's a good set of examples. So, can I ask you in terms of
merging hint?

NS: And is it although we really let's, I think property is maybe a better
name than is, although it's longer so.

NS: The examples you have there. Odd, prime, whatever. And they're
describing.

NS: They're describing mathematical concepts as opposed to the, you know,
Eridy, the infix whatever.

NS: And as David pointed out, is, maybe there's other things that affect
the way something's read but aren't really about mathematical properties.
So, they, you know this. I want you to read this as if it were a matrix and
not just a table. Or I want to read this as maybe lines or case numbers, or
whatever, and things hints towards the reading. So maybe the hints are
expanded, and that we have 2 separate ideas. One is geared toward the
mathematics, and the others geared toward the reading.

NS: is that a good distinction?

DG: I disclosed, In Github, that I personally do not intend to use the
hints. So, to me, merging the 2 syntaxes has the value of I don't have to
make that clear to anybody, because it's the same space. So, I'll be using
lots of visa. It makes my stomach queasy that they are artifacts that are a
custom dialect that we define.

DG: It's this list will not be complete and to me it's more natural to use
the underscore overrides.

NS: But I'm still kind of curious, cause you you've proposed several times
using an underscore. And yet that's absolutely against the notion of you
know I want to explain the concept. So that's still an issue.
------------------------------

Transcript Ends.

*ACTION* NS: will arrange a two-hour meeting for next week, March 2, 2023.