Gitter Channel Archive 2nd of March 2016

I think MathML is a failed web standard. It never had support from browser vendors (all rendering code is volunteer contributed, almost entirely unpaid). No browser ever sat on the MathWG (ok, Murray Sargent from MS Office but that doesn't help). MML is also not being developed, just minor bug fixes at this point. Content MathML is a complete failure for providing semantics for the web (the only known use cases are procedural / computation / CAS). PresentationMathML is supposed to only do layout but since it's the only thing that even barely works, it's brutally squeezed into providing semantics -- something that people think is feasible only because Neil Soiffer was able to write heuristics to make sense of it. so it's a bit like saying creating TeX was easy because Knuth could do it or that images are fine for a11y because Abby can do OCR on them. Add to that that few languages have the history of English with Nemeth/MathSpeak being much older than any of this and providing a basis (whereas I hear here in Germany blind students learn LaTeX in middle school). So PresentationMathML is not supposed to provide semantics, it also doesn't help with layout much on the web -- and imho it really shouldn't because CSS should and MathML has some minor and some major incompatibilities with CSS while all the while CSS is gettting better and better allowing (with ugly work) perfect math rendering (no JS involved). Besides, when you look at HTML5 you can throw out roughly 90% of MathML anyway; it's just duplicated. And to top things off the MathWG charter is running out this month with no continuation in sight and MathML is still stuck in print-content for math, not even providing anything even slightly graphical (e.g. commutative diagrams). Finally, its community is badly broken, with the "document language people" (aka XML folks) being adamant that it's brilliant (and yes, it's ok there) whereas those people that have enabled math rendering for years on the web are not being helped one bit by the perennial promise "once MathML is implemented, all will be fine"; it's not because even the best implementation in Gecko is terrible for development as it is shoddily integrated, with missing or broken API integration for style, dataset, event handlers. IMHO, what we need is a bottom up effort, helping those doing rendering with improvements in CSS, help those doing accessibility by providing a proper ARIA module and an extensible way of embedding a11y relevant information, and help those wanting to take the document language further to do that without messing up the progress of math and science on the web.