[MathOnWeb] CSS task force minutes 2018-09-10

Hi everyone,

Here are the minutes from the meeting of the CSS task force two weeks ago.
Sorry for the delay. There was no TF meeting this week.

Best,
Peter.

# MathOnWeb CSS TF, 2018-09-10

* Arno: found an example
https://macrotypography.blogspot.com/2015/11/curly-braces.html
  * Peter: very nice!
  * Peter: I realized that my stretchy+UL example is not actually possible,
even with MathML -- flow content is not allowed inside <math>
* Dani: I added some examples but I'm not quite following the structure
* Arno: let's go through the page?
  * => yes.
* Arno: let's prioritize
  * otherwise we might run out of time
  * Dani: start with simpler ones?
    * Arno: but they might not be the priority?
    * Dani: sounds good
* Arno: what's the difference between centerline and baseline?
  * Dani: it's a different font metric, right?
    * Independent of baseline
    * e.g., align with border
  * Arno: there's the CSS Github Issue
    * they seem amenable
    * do we ask them to recommend this?
    * would that take care of the centerline issue?
      * by aligning box with line
  * Dani: so that baseline is more general than centerlne?
  * Arno: would the proposed issue help?
    * Dani: don't think so
* Peter: is the main problem with centerline that it's about font metrics?
  * Dani: yes.
  * Peter: does it need opentype math table fonts?
    * Dani: we want a CSS way.
* Arno: still don't understand if centerline is different though
  * Peter: if you align baselines of the box for the fraction line, how
does your line align with a minus sign
     * (unless you use a minus to build the line)
* Dani: there's
https://www.w3.org/TR/SVG11/text.html#AlignmentBaselineProperty
  * Arno: right, like TeX's axes
  * Dani: yes.
  [more discussion]
* Arno: putting myself in the CSS TF
  * trying to understand what they think they might help with
* Peter: not sure we'll get them to do something about centerline
  * it's too deep down
* [more discussion about potential push back]
* Arno: could we present it as an extension of the baseline alignment?
  * baseline solves the problem in part
  * then something like the SVG properties
  * Dani: depends on how baseline is done
    * Arno: would Tab's suggestion work?
  * Dani: in SVG you have "baseline" or "mathematical"
    * if there's one word, then we could
* arno: right, Tab doesn't say what the property would be
* Peter: I think it would be good to build on it
  * to ensure that if they're inclined to baseline, do it in a way so that
if they one day become inlcined to centerline, we don't need a completely
different solution
* Dani: baseline problem extends to table
  * you might not have a vertical stack but a 2dim structure (grid)
* Arno: let's move to stretchy constructions
  * Arno: I'd favor proposing a CSS property on how to draw a border shape
    * specify what shape a border should take
    * e.g., left border should be a curley/square/etc bracket
    * then specify top/bottom/left/right border
    * fallback to straight line
  * Dani: I understand the propsal but I feel it might not be the right
solution
    * e.g., for a11y, they don't exist
    * why can't we have it on an element?
  * Arno: so on a span?
  * Dani: so a new kind of border? List of attributes?
    * Arno: yes, the list of unicode characters
    * Dani: fwiw mathml doesn't consider this mandatory; there may be
stretchy as a property but it doesn't matter if you do it
    * Dani: also depends on font
  * Peter: so you're expecting MATH Tables?
    * Arno: yes but implementation detail.
    * Peter: I'd rather not leave that as a detail :-)
  * [more discussion]
  * Peter: I personally don't have a problem CSS-drawing things
    * but whatever they like should be fine
* Arno: so everything can be done, they'll tell us. what do we reply?
  * Peter: imo the focus should be on implementations
    * how to make it easier for devs in their current directions
  * Peter: a good question seems to always be around updating content
    * e.g., when changing the fraction you don't want to re-calculate the
vertical align
    * e.g., if the user zooms, I don't want to have to update my border
values
* Arno: do we want to bring up zoom?
  * Peter: I'd like to but yield to the two of you
  * Arno: agree but it changes so much and worse across browsers
    * seems too huge
* Peter: next steps, focus on wiki page?
  * refine and prioritize?
  * Arno: sounds good.