Hi, Pete. I think the idea behind explicitName is to allow AT to announce real name instead the name generated from subtree. So that AT can use IAText to traverse through subtree and expose child markup to user. So let's we have accessible name and IAText interface then the question is how could AT know if it's real name or name generated from subtree. For that they should copy the logic from server side, i.e. traverse to subtree and calculate the name by IAText, IAccessible::name and then compare strings. I think this is main concern.
There are several roles that allow name calculation from subtree like row, tooltip, pushbutton. In theory they can have complicated hierarchy and this could perf issue, I don't have numbers, but name calculation is noticable part in Firefox profiling. Example, <div role="row"><div role="gridcell">cell</div><div role="gridcell">quite complicated stuffs like <a aria-label="link">click me</a></div> So AT can't just compare name and IAText::text because IAText::text returns two embedded objects in Gecko, while name is generated from role="row" subtree and equals to "cell quite complicated stuffs like link". Hope that helps. Thank you. Alex. On Sat, Jul 23, 2011 at 12:12 AM, Pete Brunet <[email protected]> wrote: > Hi Alex, It would help if you can an example that exposes the problem. > > Also, I am assuming that accName is always the content that the AT should > use in normal playback. IAText::text is useful if a user needs to examine > the actual contents, e.g. when closely reviewing or editing content, but > that would be done in a different reading mode than normal reading and then > the user would be fully aware of the context with respect to titles, labels, > etc. > > Are you saying that it's a performance issue to generate the accName due to > the hierarchical nature of the a11y tree of Firefox? How much of an issue > is it? Is it perceptible? Your example will help us understand this > aspect. The title example Jamie gave in the bug seems simple and > performant. The label example would require a search for the label, but I > assume the use of labels is not prevalent and there is already logic and > possible caching already in place for this used by the implementation of > relations. > > Pete > > On 7/22/2011 9:48 AM, Alexander Surkov wrote: > > Hi, Pete. It may be a tricky way to get name from IAText::text because > of hierarchical text representation (traverse through whole tree and > build name). Also this is not quite the same, subtree that name is > calculated from can contain accessible having name different from > their IAText::text what affects on the result name. Another point of > view this approach is not too perf way to check whether it's real name > or computed. > > Thank you. > Alex. > > > On Fri, Jul 22, 2011 at 11:35 PM, Pete Brunet <[email protected]> wrote: > > Regarding the explicit name issue > https://wiki.mozilla.org/Accessibility/IA2_1.3#Explicit_name > which originated at > https://bugzilla.mozilla.org/show_bug.cgi?id=637578 > and was discussed on the IA2 list starting at > > https://lists.linux-foundation.org/pipermail/accessibility-ia2/2011-June/thread.html#1338 > > The traditional approach to this problem is that the explicit name is > provided by IAText::text and when needed the server provides an override via > accName. The two values match when there is no override. Jamie mentioned > two cases where accName would differ from IAText::text, i.e. when a title > attribute or a label would be more meaningful than the actual content. > > Does this approach resolve the issue? > > Pete > -- > Pete Brunet > > a11ysoft - Accessibility Architecture and Development > (512) 467-4706 (work), (512) 689-4155 (cell) > Skype: pete.brunet > IM: ptbrunet (AOL, Google), [email protected] (MSN) > http://www.a11ysoft.com/about/ > Ionosphere: WS4G > > _______________________________________________ > Accessibility-ia2 mailing list > [email protected] > https://lists.linux-foundation.org/mailman/listinfo/accessibility-ia2 > > > > -- > Pete Brunet > > a11ysoft - Accessibility Architecture and Development > (512) 467-4706 (work), (512) 689-4155 (cell) > Skype: pete.brunet > IM: ptbrunet (AOL, Google), [email protected] (MSN) > http://www.a11ysoft.com/about/ > Ionosphere: WS4G > > _______________________________________________ > Accessibility-ia2 mailing list > [email protected] > https://lists.linux-foundation.org/mailman/listinfo/accessibility-ia2 > > _______________________________________________ Accessibility-ia2 mailing list [email protected] https://lists.linux-foundation.org/mailman/listinfo/accessibility-ia2
