W3C

– DRAFT –
Making math a first class citizen on the Web

29 October 2020

Attendees

Present
Bert, CharlesL, fantasai, George, ivan, jeff, jensimmons, myles, plh, rego, wendyreid
Regrets
-
Chair
-
Scribe
DavidC

Meeting minutes

<bkardell_> https://docs.google.com/presentation/d/18oam2cageqx_8LZ5J9S6sUihygwaCWUgi1EOjuBUaCM/edit?usp=sharing

<plh> Slides

<plh> Slides

NS: Davidc Scribe in irc

NS: links to presentations in chat and zoom

<rego> present

NS: Neil and Brian to do initial presentations

PLH: starting recording

Neil: Brian will talk about status of mathml, them I will talk about mathml 4 full core and charter

BK: math and graphics mathml and svg two earliest w3c standards, integrated into html5 parser

BK: some issues and when blink split from webkit, dropped mathml, left questions about status of mathml

BK: mathml has strong infrostructure outside of browsers wikipedia has thousands of pages and lack of mml one of most starred bugs on chrome

BK: math is text and ability to render math accepted

BK: eg home education and needing to do math

BK MathML refresh community group: split old mathml spec into full and core. Core is focussed on things supported by browswes and have existing content

BK@ core draft aims to specify many things not previously specified eg integration with html/css/dom

BK apporiach in style of modern standards with interated tests

BK: Igalia working on imlementation and tests

BK: existing browser implementtaions don't pass all tests as previously not interoperable

BK: high level of interop even in current implemetation and improvined test coverage

BK: Please jon community group and/or file bugs on implementations.

NS: what is in Core..? scripts limits fractions stretch characters, matices and math typosigraphy (low level math layout details)

NS: Not in core (core level 1): linebreaking, indentation, bevelled fractions, elementary math (long division etc) meclose (strikeout etc) labelled table rows, links, som eother little used attributes and some obscure alignment features.

<plh> Intro

<plh> charter draft

NS: CharteNS: scope develop recs and test suites, clarify integration with javascrip/css, accessibility requirements (already quite good but want to clarify) Removing some unused elements.

NS: out of scope adding new elements, nor eemove elements that have significant usage. Goal to maintain most existing cmathml documents.

<plh> Getting Math Onto Web Pages

<MikeSmith> I have seen discussion about dropping particular features, and response from Apple that they don’t support dropping those features

NS: not removing content mathml, this has not had wide adoption and we are looking at intent annotations in presentation mathml

NS: deliverables: Core level 1 already have reasonable start of a spec. Also Core level 2, this is pitems planned for future core releas. and MathML full, other parts of MathML3, and use outside of teh web.

<fantasai> MikeSmith, I think it's mostly about deferring not dropping features? Similar to CSS2.1 effort.

NS: specify accessibility techniques. Hope to have at least one tex to mathml convertor presrving accessibility annotations., and one a convertor from content mathml to presentation with accessibility and search annotations.

NS: sample conversions to speaech.

<MikeSmith> fantasai, what I saw was about dropping some features actually. I am pretty certain Fred Wang has proposed doing so

NS: will open up to discussion, would like feedback on polyfilling, and shadow dom issues.

NS: There is a start of accessibility mapping for mathml mad searching not really covered.

NS: who will do the work? Igalia has done most of the work recently, not much support from browser vendors.

NS: recording stopped, open discussions

Judy Brewer: thanks for presentation I could not agree more how important it is to get accessibility on the web.

<MikeSmith> fantasai, see https://lists.webkit.org/pipermail/webkit-dev/2019-March/030537.html and that whole thread

JB: Particular;y looking for conformation that content mathml will continue. Was going to ask how could W3C help with guidence, resources. You started to cover this in final part of your presentation.

NS: I think a lot of this is bottom up so we mainly need to get vendor involvement, eg apple had done a nice job of mathml in safari and also voiceopver on safari but would b eimportant to get updates for new intent markup

Florian_Rivoal: some time there was pushback agaisnt mathml as content mathml not widely implemented and presentation mathml not rich enough, it appears that you are agreeing with that but rather than drop eveything you are looking to extend presentation mathml

<MikeSmith> along with the other problems that florian_irc mention is the problem that the underlying MathML layout model does not match well with the CSS layout model

NS: there is a technique for posting linked trees of content and prersentation mathml, so current plan is to address presentationsmathml's lack of semantics/intent eg knowing if a superscript is a power or Transpose, by supplying a simople way for authors to specify a general subject area (eg calculus or chemistry and to provide a way more similar to contnet mathmk where the author intent is specified by attributes on individual terms)

BK: HTML itself only has weak text semantics, eg is this a poem etc, so a lot of things fall into hints and heuristics so that given 2 decades of experience so right answer seems to be that right answer is to build on mathml not start over.

Bert Bos: You mentioned some things not going into core 1.

BB: why not links?

BK: we asked about this in tag review and two browsers support this already (and that won't be removed) but will be deferred to level 2 for specification

BK: many things not specified in mml, eg tab index, security principles (has to match html links)

BK: SVG backed out its original anything can be an (x) link

<Zakim> MikeSmith, you wanted to comment

BK current plan to allow links on token elements and to add a ne "a" element more like html a

<MikeSmith> https://lists.webkit.org/pipermail/webkit-dev/2019-March/030537.html

Mike5: It was asserted that there was no plan to remove features, but in the past theer was discussion on webkit about removing features and apple said they would not do that as they have mathml enabled in webview which is also used out of the browser and it is less clear there which features are being used.

Mike5: I'm Mike Smith (in Tokyo) been followeing MathML for a long time, at one point google removed mathml support from Chrome, and what has prevented this from getting higher prioritization amongst browser engines, part of the plan has to be someoen taking responsibility for future maintenance

BK: I did address some of these, but I do think that things have changed

<bkardell_> thank you iank_

Ian Kilpatrick: From our perspective (google) we have a high bar for adding features, well tested but work in last couple of years has been fantastic in addressing these issues and we feel a lot more comfortable about maintenance

<iank_> florian_irc: That's part of it - the code now from our perspective is very easy to maintain.

<iank_> bkardell_: No problem :)

Charles_laPierre: I'm recalling a demo of mathml in chrome at a previous tpac. The author intent has a strong correlation between the purpose annotation in personalisation WG, my question: where does this leave mathjax?

<bkardell_> MikeSmith: I wish I had some more time to address your stuff about removal and webkit comments

<MikeSmith> bkardell_, we can take it ot eh issue tracker

NS: won't replace need for mathjax, MJ ytakes several inputs tex, mathml, asciimath, and output jas being html, svg and (previously) mathml. I suspect the html output or a new output jax will incorporate mathml core. I have spoken with DC (main implementor) and he's in favour of this when time ready.

<Zakim> tzviya, you wanted to ask what you need from publishers

Tzviya_Siegman: as a publisher what do you need from us? I can not promise money but I have loads of content.

NS: content is important, I've been surprised that publishers haven't pushed harder for elementary math support.

NS: Pearson has been involved in cmathml refresh and been looking at what is important in educational contexts 9eg bevelled fractions didn't seem so important andc crerent;y pushed to level 2)

BK: It would be good to support the charter and join the group if chartered, providing content helps.

<CharlesL> s/personalization WG/personalization TF within the APA

BK: NISO and Pearson have given support

<Zakim> MikeSmith, you wanted to ask about prioritization of features and developer pain points

TS: At Wiley we are more in academic publishing and linebreaking likely to be an issue

Mike5: one thing I have not seen in previous years is documenting what are the pain points due to lack of support eg Neils example with bevelled fractions, but it would be good to document where the descisions were made

BK: mathml refresh has a huge (github) issue tracker

Mike5: People can't go through the whole tracker but need a simpler documented list of main decisions

BK: Until we gt past current pain point that not implemented in all the main engines difficult to get into specific issues

George_Kerscher: Chemistry WG committed to working with ypu on pronunciations for Chemisty

Neil: yes we really appreciate input from Chemisty WG

<bkardell_> MikeSmith: would you be willing to talk with me after about how we can answer your point better? We have a lot of data, an explainer, a spec, etc -- how can I help?

<MikeSmith> bkardell_, yeah, definitely. And I can raise an issue in the tracker too

<bkardell_> MikeSmith: perfect!

Florian_Rivoal: unlike most texts math is multilingual same math may apply to multiple languages. Is authoring intent sufficiently abstract or does in insert English descriptions.

NS: unlike core which is already specified this is not yet final propsal, but current ideal is that levels 1,2,3 would be common language independent terms eg "binomial coefficient"

NS: translators would need to translate these level 1 terms, but for wider defined terms (level 3) they would be or may be in English, a translator may try to translate as text on the fly or the author may supply more than one language.

FR: would it be possible to make the dictionary multilingual to save each translator doing the work

NS: yes we could have multilingual fields and ask people to contribute

<bkardell_> plh I have a question in closing

NS: but in advanced specialised fields, perhaps showing English bias, English is very predominant

<plh> Proposal: are people supportive of the new charter?

<fantasai> +1 to chartering this group

BK: Could we tale a sense of teh room of whether peopel are supportive of a new charter for a math WG

<florian_irc> +1

<CharlesL> +1

<Bert> +1

<tzviya> +1

<Judy> +1

<plh> +1

<ivan> +1

<maxf> +1

+1

<atsushi> +1

<George> 1+

<Deyan> +1

NS: we have had a huge number of participitants here and I really appreciate this

<ivan> Lots of things in Hungarian

<NeilS> +1

<bkardell_> thank you everyone

Minutes manually created (not a transcript), formatted by scribe.perl version 124 (Wed Oct 28 18:08:33 2020 UTC).

Diagnostics

Succeeded: s/lonks/links

Succeeded: s/linke/like

Succeeded: s/ niot / not /

Succeeded: s/securrity/security/

Succeeded: s/priotization/prioritization

Succeeded: s/maintainence/maintenance

Succeeded: s/testsed/tested

Failed: s/personalization WG/personalization TF within the APA

Succeeded: s/Charles laPierre:/Charles_laPierre:

Succeeded: s/Tzviya Siegman:/Tzviya_Siegman:

Succeeded: s/NS Pearson/NS: Pearson

Succeeded: s/BK mathml refresh/BK: mathml refresh

Succeeded: s/George Kerscher:/George_Kerscher:

Succeeded 2 times: s/Florian Rivoal:/Florian_Rivoal:/g

Succeeded: s/alrready/already

Succeeded: s/sufficient;y/sufficiently

Succeeded: s/peopel/people

Succeeded: s/predominent/predominant

No scribenick or scribe found. Guessed: DavidC

Maybe present: BB, BK, Charles_laPierre, Florian_Rivoal, FR, George_Kerscher, JB, Mike5, Neil, NS, TS, Tzviya_Siegman