- From: Joanmarie Diggs <jdiggs@igalia.com>
- Date: Wed, 19 Aug 2015 13:54:04 -0400
- To: Joseph Scheuhammer <clown@alum.mit.edu>
- Cc: W3C WAI Protocols & Formats <public-pfwg@w3.org>
On 08/19/2015 01:16 PM, Joseph Scheuhammer wrote: > Hi Joanie, (et al) > > On 2015-08-19 11:57 AM, Joanmarie Diggs wrote: >> An example of accessible names going (imho) horribly awry with list >> items can be seen in doxygen-generated pages, such as this one: >> http://freedesktop.org/software/pulseaudio/doxygen/ >> >> If I view that page in Firefox and use Accerciser to get the name of the >> first list item of the accessible list associated with the left-hand >> pane, I get: >> >> 'PulseAudio\xa0Introduction\xa0Simple API\xa0Asynchronous >> API\xa0Threads\xa0Error Handling\xa0pkg-configSimple APIAsynchronous >> APIChannel MapsSample Format SpecificationsVolume Control\xa0Deprecated >> ListData StructuresFilesExamples' > > That looks like a name-from-contents computation, which follows from the > spec: > - Listitems and treeitems require an accessible name. > - both roles allow name-from-contents. > > In the absence of an author supplied name (e.g., @aria-label), browsers > are thereby required to create the name from the contents of the item. > > Your results are an argument for the author to supply a name to avoid > these long names-from-contents. What are the odds that we can get most authors to supply the name? Could we not instead modify the name-from-contents computation of list items so that it is limited to the displayed text for that list item; not all of that item's visible descendants? --joanie
Received on Wednesday, 19 August 2015 17:54:42 UTC