Minutes: MathML General/Semantics meeting 23 July, 2020

*Many thanks to Patrick Ion for volunteering to scribe!*





*The meeting was recorded: Attendees:Neil SoifferDavid FarmerDeyan GinevSam
DooleyLouis MaherDavid CarlisleMurray SargentPatrick IonBruce MillerSteve
Noble*[Initial clarifications over the Zoom URL]













*Charter discussionNS: Brian Kardell has done a serious start.  He comes
from the MathML CoreAnd web technology background so that’s reflected in
the first draft he made.  I see the time now as when we try and put down
everything we might want for  a couple of weeks.  Then we need to discuss
what actually can be done, and who we imagine can do it.  We shouldn’t have
anything in there that is a setup for failure.The charter is at:
https://docs.google.com/document/d/1W-oYUbOMueaqb3KFSWkjWVBwR6AzSEBizHwQhvSwfDc/edit#
<https://docs.google.com/document/d/1W-oYUbOMueaqb3KFSWkjWVBwR6AzSEBizHwQhvSwfDc/edit#>[See
the headings some of which are reproduced below]Tests suiteClarifications
on supportInternationalizationIntegration with web technologiesRemoving
speculative elementsOut-of-scope: new elements which would go into Core
Level 2; to get things implemented is a necessity for anything in Core
Level 1, with which is to be begun; E.g., Linking elements; line-breaking
that all want but is difficult;  This can also be done for MathML
Full----constrained by implementationSD: I hate to add that the subject of
Content MathMl is as yet unmentionedNS:: It should be in scope if we want
to tackle that; we might say something about a future in which we might do
that because of the uptake problem, but that just now we’re not taking on
an extensive rewrite.SD: This will depend on the semantics addition that we
are discussing now; NS: My notion is that we can see how the new semantic
markup goes; we have yet to agree on it.SD: The matter needs mentioning at
least.NS: All present have edit access to the charter, or you can make a
comment (for anything that’s controversial).DG: Is the outcome MathML4 or a
3.5?NS: I think it’s MathML4;  it’s going to be  a big rewrite just because
of introducing Core and explaining how it is connected.NS: Deliverables as
in the Charter again--- focussed on a subset of 4 that fits ‘in the
platform’ [of modern web technologies]Core Level 2 explains polyfills,
linking, accessibility issues, ...DC; I don’t think Core Level 2 should be
normative because at present it’s all speculation about what we can’t yet
do.NS: Agreed.NS: Point 4 is revision of MathML3, and mentions the semantic
hint markup on presentation (I added this immediately). Other deliverables
on the list will multiply; in part they are there to shore up the normative
parts; I think a Test Suite is just  a requirement not normative.DC: Yes,
just so.NS: So if we add semantic markup we have to have tests; e.g. in a
new MathPlayer that I’m expecting to write.SD: Presentatipon <-> Content
converters would count there as well.NS: ..Polyfills, DOM APIs, relation
with CSSI’ve been starting on polyfills for Core; tables are a problem for
which the context keeps shifting; Chrome is supposed to be rewriting  their
table code extensively but when it will be finished and what will change is
not clear..I’ve done several things including elementary math and some nice
beginnings of  line-breaking.MS: I don’t see why that last is not in the
basic code.NS: If Microsoft, say, were to ante up enough to Igalia for this
part, it would happen; otherwise it’ll have to happen in level 2.I added a
number of other detailed descriptions of utility code that should be there
for conversions, semantics  to whatever.We can add a lot of possibilities,
and I expect we will.  Then we’ll have to winnow this down to what actually
can be done.MS: I’m very interested in the speech output questions.DG: I
think we should have a section on adding new notations that haven’t been
addressed in the specNS:  I expect there’ll be a list of 300, say, values
all should recognize; then one could explain other cases that are perhaps
known but not required; finally some suggestions for naming
conventions/things to think about for new names. It is good to make the
list non-normative because a non-normative list can be more readily
changed.PI:  …. <mglyph>; what’s need there?NS: I don’t think there’s
anything to do there; it just maps to <img>; similar questions came up in a
PDF call for me recently.DC: yes, <mglyph> is still in there.SN: With my
Pearson hat on;  Are accessibility (A11Y) considerations secondary, and to
be got around to some day?NS: Charter drafts go to groups who might care
about math; it’s certain that several groups have a11y concerns; we have
pretty clear continuing commitment on that front; it may end up with
non-normative description though.SN: Are there normative considerations
that should be fixed for a11y purposes? After all the principle is that you
should consider a11y at the outset and then it’s much easier to provide.NS:
Big additions for MathML4 are the semantics additions that certainly affect
a11y; ‘in scope’ should add a little bit about a11y; perhaps, Steve, you
could add a bullet point appropriately?BK did a yeoman job of producing a
basic document but his PoV isn’t a11y but other concernsSN: I’ll do some
addition.SD: It’s right we see to it that the basic elements are there to
support a11y.DC: The fact is that you can only add whatever the browser
makers will implementNS  There’s an APA (A11y Platform Architecture) WG of
the W3C.NS: We need to add APA as an interested working group; but also the
Publishing WG; The chem WG;  ARAI WG (who have an Accessible Architecture
Mapping); and others to need to review the final MathML charter. SVG, Web
Incubator and so on are already listed; we promised to work with other
WGs.NS: If we had Shadow DOM for elementary mathematics elements it would
make polyfills for such things much easier.NS: I encourage everyone to
examine and change the draft, say over the next couple of weeks; then we’ll
need to refine it (cut it down).Continue discussion on semantic
proposalsNS: I found very interesting the commentary showing many notations
for inner products.DG: Terry Tao had a great discussion on 15+ inner
product notations on MathOverflow, which trended to the top of HackerNews,
giving us an extra feeling of how timely our semantics a11y work is:
https://mathoverflow.net/questions/366070/what-are-the-benefits-of-writing-vector-inner-products-as-langle-u-v-rangle/366118#366118
<https://mathoverflow.net/questions/366070/what-are-the-benefits-of-writing-vector-inner-products-as-langle-u-v-rangle/366118#366118>SD:I’ve
worked my way through (⅓ so far) the comparison document, then I made a
smaller document to work on with; this is still a work in progress;
generalizations  seem to be quite possible when I actually analyze the
complicated examples; these are examples where good naming and clarity over
argument positions work out well.  I’ll share a case on-screen with $x_i’$
and it’s possible interpretations. \dots   It’s the nested subscript cases:
...BM: Can you point out what is different from what DG and I propose?SD:
It will be great if we really converged; I am just working through it
still.BM: I added msub and mover cases this morning in the comparison
document.SD: The cases were where the presentation had alternative
semantics:$\sin^2$ [sin squared] is another problem case; we had ‘sin
inverse’ I want to be able to represent sin applied to x and that squared,
and to point to the right places in the tree.NS: I think a good example of
this is got by replacing sin with capital D (a pretty common notation); if
D is derivative is it any different from just a function called D?DC: It’s
only the trig functions that have the wacky power notation; for any other
function powers, written as a superscript, are repeated application.PI cf
$\sin^n$ vs.  $\log^n$<much discussion of cases, including mention of DC’s
thesis where the superscripts are mostly indices>SD: Speech engines and
brailing engines are going to have to deal with all these things and they
will be different.<much more discussion, in some cases going back to school
days>MS: nit: : sin shouldn't be in math italicNS: There’s a non-functional
aspect in Sam’s draft as well as functional notation like BM/DG’s.MS: I’d
like to see a column 4 with the desired speech form(s)in EnglishPI: How
about a Speech Button for a shorter column.BM: Regrets maybe next week as
I’m co-chair at CICMPI: DG will be there too; me too.NS: Continue next
week; with this as well as the charter; From the Chat:DG: btw, we did not
invent "functional power", it's used in practice:
https://en.wikipedia.org/wiki/Function_composition#Functional_powers
<https://en.wikipedia.org/wiki/Function_composition#Functional_powers>*

<http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
Virus-free.
www.avg.com
<http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail>
<#m_565389086321861302_DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>