- From: Peter Krautzberger <peter.krautzberger@mathjax.org>
- Date: Tue, 4 Oct 2016 09:47:30 +0200
- To: www-style list <www-style@w3.org>, W3C Digital Publishing IG <public-digipub-ig@w3.org>
- Message-ID: <CABqxo82T6AeUJRNYkq7smeye9tZD0NJuL+_=L6AZ32Vt=jikAg@mail.gmail.com>

Hi, For context, I wrote that blog post. Some inline comments. Re Avneesh, > - Publishers are already using MathML in their production workflow. But they have to comment it due to lack of support on consumption side. so, MathML is already available. I do not believe that's a sufficient argument. Publishers use lots of formats (XML or otherwise) that have no place on the web. Ignoring for a minute that some publishers do not use MathML, it might be worth noting that authors mostly do not use MathML. Instead, they use MS Office XML, math editing software (with proprietary internal formats) or text-based formats such as TeX/LaTeX or from computational software. It's almost always the publisher who converts material to MathML. I don't think it's a burden on publishers to convert MathML to HTML with CSS or SVG. They do this with all their non-web formats anyway (e.g., JATS, OASIS etc). There are several open-source tools that can convert MathML on both server and client (e.g., mathjax-node / MathJax). Once you look beyond MathML, there are even more tools out there that prove that mathematical layout is not difficult to realize in CSS or SVG. Similarly, it is not a burden on publishers to make such output accessible. Again, there are open-source tools for this. > - JAWS and NVDA, the two most popular Windows screen readers support MathML. It is also supported by Voiceover, but I remember the statement from Peter that support in Voiceover is not up to the mark. So, assistive technology support MathML. It would need some more refinements, but the support is there. >From my experience (of testing content in the wild), the reality appears rather different. First off, NVDA does not support MathML. It can leverage MathPlayer if installed -- but MathPlayer is dead in the water since its developer retired. Granted, NVDA received some funding recently to integrate Volker Sorge's speechruleengine (which we at MathJax contribute to as well and which originated in Volker's work in ChromeVox) so likely this might come true at some point in the future. >From testing, JAWS and VoiceOver support is barely usable beyond the simplest expressions. From an outside perspective, I can only conclude that they are unmaintained (e.g., look at bugzilla items for VoiceOver). In the end, none of this changes the fact that MathML is not accessible in itself. To make a comparison, math accessibility tools are like OCR or automated image captions. Just because they can generate something vaguely sensible from a given source material does not mean that the source material is accessible. Re Paul. > MathML's domain is a couple of orders of magnitude simpler but its attempt to capture math notation and be computable, accessible, look good, while supporting a range of math typesetting styles, multiple languages, etc. Presentation MathML (which is what this discussion is about) is for visual layout. These other ideas are nice of course (albeit a bit unrealistic). > While MathML is flawed, it is the best we have right now. We can improve MathML and its implementations, or develop a replacement, but let's not dump MathML until its replacement exists and can be seen to be an improvement. CSS and SVG are already superior when it comes to realizing math layout on the web and offers ARIA to make such rendering accessible. There are also tools that do not use MathML and yet provide universal rendering (e.g., Desmos's editing interface). Re Ivan > Ie, if the browser does include a MathML renderer, this is what one would want to use, otherwise (as a fallback) an image. It would help if people stopped suggesting that the only possible fallback for mathematics is an image (which mostly triggers "binary" alongside it). Unless we're specifically talking about IE8 or older Kindle devices, SVG and HTML+CSS is the obvious "fallback". Both of these formats can provide universal rendering and thus are really not "fallbacks". Best regards, Peter. On Tue, Oct 4, 2016 at 8:36 AM, Paul Topping <pault@dessci.com> wrote: > > > I do not disagree with what you write, but I am not sure this is > relevant for > > the original problem. The whole issue is to *use* MathML whenever > > possible, not the contrary... > > Glad to hear it. There are those who feel MathML has not been totally > successful and would like to see it replaced. I'm just trying to defend it > until such a replacement becomes a reality. Sorry if I overreacted. > > Paul Topping > > Design Science, Inc. > "How Science Communicates" > Makers of MathType, MathFlow, MathPlayer, Equation Editor > http://www.dessci.com > > > >

Received on Tuesday, 4 October 2016 07:48:03 UTC