- From: Matt May <mattmay@adobe.com>
- Date: Thu, 21 Jan 2010 23:23:52 -0800
- To: David Singer <singer@apple.com>
- CC: John Foliot <jfoliot@stanford.edu>, Maciej Stachowiak <mjs@apple.com>, HTML WG <public-html@w3.org>, "public-html-a11y@w3.org" <public-html-a11y@w3.org>
On Jan 21, 2010, at 11:56 PM, David Singer wrote: > John asked to turn things on their head: do I take it that people think the spec. should say that the AT MUST NOT try to find EXIF tags, do OCR or scene analysis, inspect the file name for possible useful information, etc.? I'm thinking you mean UA when you said AT. Either way, the answer is no. I would encourage the development of all kinds of client-side repair. I just want the guidance in the document itself to be limited to what can be reproduced consistently across browser/AT toolchains, and I don't want authors to get any ideas anywhere in the document that they're getting a free pass. Anything called "repair" implicitly means signal has been lost. How that signal is patched up is going to vary, in some cases wildly, from implementation to implementation. That makes it a good subject for research, and a really bad subject for standardization. The worst possible scenario I can think of is where the spec hints at image analysis as a replacement for @alt, no browser actually implements such support, but authors take that nonexistent support for granted, because hey, it's in the spec, and stop worrying about @alt. - m
Received on Friday, 22 January 2010 07:24:34 UTC