meeting minutes, Thursday, August 25 -- next meeting

Dear math-on-webpages CG,

Please find the minutes to our last call below. Thanks go to Jean Kaplansky
for scribing!

The next meeting will be

*September 29, 2016, 6pm UTC*

Best wishes,
Peter.

# Minutes from 8/25/2016 math on the web CG:

IDPF/W3/DAISY call happened on 8/24/2016 to discuss Math on the web. There
were a lot of responses from many people who were not all concerned with
math.

The struggle is that the EPUB 3.1 is about to be finalized. 1 topic is
recommendations, and a11y-related recommendations. So they had a discussion
about how to embed MathML alongside fallback information and how to make
this work without the Switch - which is deprecated.

Jean clarified the statement about switch support - it’s out there with
some RS’s and not with others.

Now they’re talking about using data slice MathML with images.

Now the discussion is about using CSS to hide and not hide the MathML. It’s
not a very neat solution. Jean thinks that this could cause some a11y
problems down the line. Then there’s the whole aria-label and
aria-lableledby issues - they don’t necessarily act the way people expect
them to.

Philosophically, Peter hates to admit that he doesn’t see that this
conversation is going anywhere. The assumption is that MathML is good for
a11y, but this is not a good assumption.

Charles joined the call: I don’t think any one of us can fix the problem of
Math a11y all at once.

People started discussing things like MathML + ARIA label, even if it’s
duplicate content in the a11y rendering, it wouldn’t be so bad if it was
the same. The problem is if someone creates a label you could get something
different in AT if it actually understands the MathML (which isn’t likely
to happen that often, but when it does…)

For there EPUB content - the base page is SVG so the math expression
becomes clickable - and then you get the MathML, where you can go see the
MathML - SAS uses SVG approach to, and they aren’t using MathJax - they
embed the images with the raw MathML that let’s people get to the MathML.

Conclusion - they thought that they had a solution yesterday after the tele
conference, but further thought after the call re-thought whether or not it
would work. JavaScript is still not the answer because schools may block
it. The reading systems were not on board with the idea. Rick Wright - “If
others bought into the approach, he would also buy into the approach.”
Charles doesn’t know how difficult a battle it’s going to be.

data attributes- a lot of people are already doing this, but it’s not
necessarily HTML5 best practice.

Peter has trouble recommending inclusion of MathML in EPUB. Charles
mentioned that you could also put data-LaTeX, then you could include the
LaTeX.

The people on the list don’t like using data for MathML because they’re
using “weird” solutions. They should go back and reconsider the data
solution again-  a very good solution for including the MathML in
association with the RS’s.

What about deeply nested labels and what happens when there are a lot of
them? They work off the MathML, and can work with the tree structure, and
then do an abstraction to do an analysis to create alt text out of it. It’s
not such a leap to say that we could just label the SVG or HTML tree that
we generate that would provide an a11y and nav rendering. An approach to
consider.

What is the disadvantage of using SVG instead of MathML in the content
instead? You could wind up with missing information unless the MathML is
also included. But wouldn’t this be an exception? Reuse is a good use case
- safety net, but it still seems to be an extreme. Exposing it while not
requiring it to be shown or rendered in any way.

Dani - would like the recommendation to store the MathML with the PNG or
the SVG in more or less a standard way, then a11y technologies could use
the MathML or not - but you’re not precluded from using the MathML later.

Steve - Some elements of the general discussion - I don’t really understand
the import of - I’ve been advocating for MathML as a11y for the last 10
years. There are lots and lots of cases for a11y - speech rules required
for both blind students and learning disabled - need the ability for
localization in the speech role process, the ability to output braille, so
there are still a lot of possibilities for places where MathML can be used.

Peter - disagrees with including MathML in the EPUB.

Bottom line - Math in EPUB is an EPUB problem, and we should let the
IDPF/W3 figure it out. We’re all EPUB’d out in this forum. ;)

TPAC: Math on the web has an hour reserved at TPAC,
https://www.w3.org/2016/09/TPAC/schedule.html#cgs-tue, - this is a very
late breaking happening. Ivan Herman set up the slot. But now we need
people to actually do a presentation. If people are going to TPAC then they
should participate in the program - Charles, Peter, and Dani are planning
to attend. Do a joint Digital Publishing Interest Group session with the
Getting Math Onto Web Pages. So… If you want to go to Lisbon in September.

2 or 3 potential work areas for the Math on the Web group- what we might be
doing as a group to contributing to the standards process.

Peter had the following on his list:
* layout (Dani had an idea to collect shortcoming, tricks, hacks, and stuff
that could use help in this area. Peter has some ideas about inline
mathematics);
* All things computational and procedural - screenshare from Desmos - not
based on MathML - more on TeX-like things - drawing graphs and calculating
things - common problems and challenges and exchange problems that might
need solving?

Jos has done a lot on computation (mathJS). He doesn’t have concrete ideas
but AsciiMath/Ascii is easier to parse, also by various graphic libraries
for graphing. MathML is too complicated to be used by libraries. Something
JSON-based has possibilities. Also thinking about how editors handle
formulas - OpenOffice has an equation editor that turns out ascii that is
very usable by libraries. More simple than MathML. Note - these Ascii
languages are meant for display and not for computation. The big thing is
what’s the standard below? OpenOffice does some Ascii thing that is like
AsciiMath, and like LaTeX without the backslashes. Not everything can be
salvaged, and not everything should be salvaged.

Peter - don’t get bogged down by the past too much. Think forward.

Charles - DIAGRAM developers group - a goal is to come up with something
that can show a11y math on the web. Might be worth exploring what they can
do on that.

Peter - people are welcome to comment on the EPUB thread, but do we want to
say anything about this as a group to the IDPF before the W3C merger? It
doesn’t make sense for the group to make a comment. But people should chime
into the conversation if they are able and have something to say.


ACTION Peter to reach out to DPUB IG
ACTION Collecting resources for computational Ascii-related math - Peter
and Jos to pursue.
ACTION add minimal layout examples