- From: John Foliot <john@foliot.ca>
- Date: Sat, 24 May 2014 12:49:14 -0700
- To: "'HTML A11Y TF Public'" <public-html-a11y@w3.org>
- Cc: <skeegan@stanford.edu>, <greg_kraus@ncsu.edu>

Colleagues, Likely overcome by time and events, this is in response to a comment made regarding @longdesc by James Craig, which resulted in an ACTION item on me. (https://www.w3.org/WAI/PF/HTML/track/actions/195) The comment suggested that the specification note (non-normative) that for complex math equations that authors use MathML over an image + @longdesc. Concern was raised that at the current writing (and continued to this date - May 23, 2014) support for MathML in user agents and assistive technology does not meet sufficient enough accessibility support to warrant excluding the use of image + @longdesc as a viable technique for success. I contacted 2 well-respected accessibility experts, Sean Keegan (Stanford University & President of ATHEN - Access Technology Higher Education Network) and Greg Kraus (North Carolina State University) to solicit from them the current state of accessibility support for MathML. Both men are known to be involved specifically with accessibility issues related to MathML. Both respondents noted that overall, support for MathML in browsers, and in combination with Assistive Technology is far from robust. Both were in agreement that moving forward, using (and anticipating continued improvement for MathML support) was the correct response; however both also noted that today using MathML alone does not provide the support to their students that they are legally required to provide, and both institutions are using a combination of MathML and external (non-html) content to provide that support. Both Mr. Keegan and Mr. Krause agreed that using an image with @longdesc *could* be considered a viable technique for success, but both also noted that the effort required, and the final outcome delivered, would likely not constitute a viable solution for them, and both were fairly firm in suggesting that they would not personally adopt such a solution. Conclusion: neither technique has robust support today, and recommending either is problematic. Recommendation is to strike any reference to the suitability or non-suitability of using @longdesc for complex math, but to avoid suggesting that MathML has sufficient support today. Selected comments follow: "The visual rendering issue is a bit muddy right now in that you need to rely on MathJax to get full cross-browser support. Using an image of math would require @alt or @longdesc and *could be* a solution, but practically is not a real solution. There are issues with @alt in terms of the language used, specificity of the description, navigation within the equation, etc. The example you provided of using @longdesc to deliver a secondary HTML document with MathML content would be an improvement over just @alt, but IMO, a less than ideal solution. It might be noted as a success technique, but is not something I would recommend." (S. Keegan - 5/7/14) "For visual rendering, only Safari and Firefox will visually render MathML correctly, so MathJax is really a necessity for presenting Math on the Web. You could present the math as an image and use alt or longdesc to more accessibly describe it, but you run into problems. If you render in alt, you have to vocalize that math, in other words, translate the math equation into how it would be spoken. That is not the easiest thing to do. If you store it at the target of longdesc, the math will still ultimately be HTML, so the same browser/MathJax issues come into play. So if you are going to "display" math in the longdesc, why not just do it in the main page to being with? Unless they want to store the vocalized math in the longdesc, but you run into the same problems as before in trying to produce that text." (G. Kraus 5/7/14) "I think of the @alt/@longdesc solution as a workaround hack in an attempt to deliver MathML, because (I would think) it would be *more work* to implement such a solution than to just deliver MathML. In any case, I agree with Greg this might be a technical solution.although far from what I would want to recommend." (S. Keegan - 5/7/14) "To make HMTL-based math accessible requires the use of MathPlayer from DesignScience. MathPlayer requires IE 9 or less and will not work IE 11. They say it partially works with IE 10. Without MathPlayer the other two options for consuming accessible math are either converting to a DAISY format or MS Word." (G. Kraus 5/7/14) This report closes ACTION 195 JF ------------------------------ John Foliot Web Accessibility Specialist W3C Invited Expert - Accessibility Co-Founder, Open Web Camp

Received on Saturday, 24 May 2014 19:49:43 UTC