MathML meetings this week: no core meeting, semantics meeting

 Frédéric said he doesn't have any pressing issues this week, so there is
no core meeting this week.

Instead, we will have a *semantics meeting on Tuesday, 28 April, 9am
Pacific, noon Eastern, 6pm Central European Time*. I have let this meeting
slip for too long. I hope this is the first of a few more frequent meetings
where we finalize ideas and begin writing them up in MathML full spec along
with drafting a working note if need be to list out details for subject
area values, mathrole values, and any other large open-ended attribute list
values we feel are needed.

*If you wish to join the meeting, please send me an email at least 10
minutes before the meeting and I will send you the zoom link. *Due to
zoombombing, the link can not be sent to public mailing lists.

As a recap, we have settled on (modulo potential name changes)
<math> tag -- a "subject" attr that changes the default interpretation.
Likely, the values will be based on subject areas taught in high schools
and undergraduate schools, potentially combining subject areas such as
pre-algebra and algebra when it makes sense.

The Chemistry CG has been discussing this and wants a "ChemicalFormula"
value and likely at least one other chemistry subject area value.

"mathrole" attribute -- this gives a general meaning to symbols and
notations, but is not meant to be mathematically precise. For example, if "×"
has mathrole="multiplication", no domain for the operands is assumed nor is
a range implied. However, with that role, it should not be mistaken with
cross product or some other mathematical notation where that symbol might
be used.

An area we did not come to agreement on was how to tag some notations. We
discussed "transpose" at some length as to whether the msup takes on the
mathrole or whether the 'T' does, along with discussing many different
notations for transpose. Other examples include the binomial coef notation
(tag the parens or the mrow), same for matrices, and also some named
functions such as *F**n* (Fibonacci number? -- tag msub or mi). I believe
everyone agreed we want to provide guidance for how to add mathrole so
there is some standardization both for the tools that consume mathrole
(such as speaking MathML or converting it to calculation) and for the tools
that produce it.

I believe that we also agreed that mathrole could be used to subsume
"class=MathML-unit" for dealing with various units such as 's' for seconds.

I hope I haven't left off any important conclusions. If I did, I expect
someone will speak up and remind me ;-)