March 3, 2019 Core meeting notes

The recording can be found at:
https://benetech.zoom.us/recording/share/ZwmQw7bHajvQS4qPJDkvk-g3aHL1hAnKPxrUOc1a7NWwIumekTziMw


Here's the summary:
*Discussion began with handling of invalid elements (wrong number of
children):*
 https://github.com/mathml-refresh/mathml/issues/15

General proposal is to convert any error to an mrow and issue a message on
the console. Google likes this as it shows all content and is easy to
implement. The disadvantage is that (e.g., mfrac) would no longer look like
a fraction and the error message is hidden to most users (who never look at
the console).

Neil: what does SVG do since that has parallel handling to MathML.
Fred: SVG says rendering stops, message to console.
Fred: -- math is closer to text, so should behave more like HTML and just
use mrow. Eg., invalid mfrac would be an mrow.
Neil -- could we use merror instead?
Fred -- merror is treated as mrow at layout level, so no merror layout code.
David -- mtable is an HTML example with structure
Fred -- mtable adds anonymous children
Fred & Rob: experiment more to see if CSS can deal with error handling
Bruce -- maybe errors should happen at HTML5 parser level
David -- wouldn't work since DOM directly created by MathJax and others
Rob -- group should say what we want to achieve and then see how hard it is
to implement
*Action Item*: Fred discuss with Google with feedback: it would be really
helpful to indicate error to the user (console is for developers), but
simplicity counts

*Related item is empty elements that have children*
Proposal was to not render children, but that contradicts google's rec to
"render everything"
Neil: we should be consistent (except where I don't like it ;-)
*Action Item*: Fred to discuss with google


*New CSS properties*
https://github.com/mathml-refresh/mathml/issues/31
displaystyle -- not much said, moved quickly to
scriptlevel --
   affects mulitplier and minsize
   Neil: getting rid of minsize has accessibility considerations for low
vision
   Someone: TeX approach is allowing only two levels of decrease
   Fred: Mozilla doesn't like minsize because it is technical dept
   David: could use CSS to limit sizes
   *Action Item*: Fred investigate CSS; Neil investigate what accessibility
groups low vision

mathvariant --
   Fred: some software generates mathvariant in mstyle:
        <mstyle mathvariant="..."> <mi>F</mi> </mstyle>
   Fred: also TeX allows the font change to be around more than a char
   David: modern Unicode-based TeXs don't allow that
   Murray: it doesn't really make sense as there is (for example) no
Fraktur chars for digits.
   Scribe: Not sure if we resolved removing it from mstyle
   Fred, David: let's push for it being part of CSS's text-transform. The
CSS group seems to like that and it makes the implementation easy
   *Action Item*: Fred to add that to CSS-drafts issue.

*Remove from core*
General agreement that all named lengths and unitless precents (#24, #4,
#7), should be removed from the core. Probably stay in full, but to be
discussed in general group.

*Weekly meetings*
Neil: There is a lot of activity around core, which is great. But that
means we need to meet more often to resolve issue. We'll move to having *weekly
meetings* for the next few months until we work through the backlog.