- From: Dr. Olaf Hoffmann <Dr.O.Hoffmann@gmx.de>
- Date: Wed, 2 Jul 2008 17:56:58 +0200
- To: "Jeff Schiller" <codedread@gmail.com>, j.chetwynd@btinternet.com, www-svg@w3.org
Jeff Schiller: > Jonathan, > > On 7/2/08, Jonathan Chetwynd <j.chetwynd@btinternet.com> wrote: > > my accessibility concerns are far removed from this issue, though > > naturally I welcome your suggestions. > > Your email suggested that user agents did not display a tooltip (though > Opera, in fact, does). Thus, I assumed you were interested in getting this > evaluated and fixed across browsers. One step in this direction is to > generate a test suite that identifies what browsers should be doing, > opening and tracking bugs, etc. I was hoping you could help in this > regard. Can you clarify why your concerns are "far removed from this > issue" ? > Squiggle displays both title and desc of single elements, this is already even better than only displaying the title ;o) Amaya provides an alternative text view - currently without much structure but the best what I have currently seen, unfortunately without a direct connection to the related graphical elements. And because it is currently not specified, how to structure these presentations of text alternatives, it could be maybe tested, whether all title and desc in a document are accessible at all or no or even some of them - why to restrict, if one can get it all? It would be pretty surprising for example to get such an 'hover-tooltip' from Squiggle or Opera type for a SVG document printed on a sheet of paper - but it would be less surprising to get a complete and structured text alternative output from the printing program on a sheet of paper. And for an optional hover-tooltip-technique about the suggested tests in http://www.w3.org/Graphics/SVG/IG/wiki/Accessibility_Activity for example this one: <svg> <g> <title>FAIL</title> <circle> <title>PASS</title> </circle> </g> </svg> why should it be wrong to display both titles in a structured way? An advanced user agent could for example transform this to something like this (using XHTML2/HTML5 as pseudo-code): <html ...> <body> <section role="g"> <h1>FAIL</h1> <section role="circle" class"current"> <h1>PASS</h1> </section> </section> </body> </html> or old HTML4: <html> <body> <div class="g"> <h1>FAIL</h1> <div class="circle current"> <h2>PASS</h2> </div> </div> </body> </html> This is even more useful as to show only 'PASS' as tooltip-output and could be an approach for a non-visual presentation at the same time. Or what about such documents: <svg ...> <title>Basic Shapes</title> <desc>Some samples for basic shapes in SVG.</desc> <circle cx="0" cy="0" r="10"> <title>Circle</title> <desc>The circle element has the attributes cx and cy related to the center of the circle and r for the radius of the circle.</desc> </circle> <ellipse cx="0" cy="20" rx="10" ry="5"> <title>Ellipse</title> <desc>The ellipse element has the attributes cx and cy related to the center of the circle and rx and ry for the radii of the ellipse in x respectively y direction.</desc> </ellipse> ... </svg> An advanced user agent could provide the complete structure as output on demand. If for example the user wants to see a tooltip for the ellipse, the circle text alternative is grayed out, document title and description and those of ellipse are in black to indicate, what is currently relevant - or an advanced user agent may minimize information in other desc elements to an icon like '>' to indicate, that there is more information available on demand. I think, it would be not very clever for tests to exclude those advanced and much more helpful possibilities, to scarifice all this, just to ensure that a minimal tooltip is popped up at all ... Another approach could be to write a simple parser script (for example in PHP) and to provide it to interested authors to generate alternative text views automatically until viewers can do it on their own in a useful and usable way? Together with some suggestions for authors how to write and to structure documents to get some useful output with even a simple parser this might show, what viewers/browsers should implement ...
Received on Wednesday, 2 July 2008 16:00:04 UTC