Re: Requesting Clarification re CR Exit Stacks Criteria

This would ultimately be a Working Group decision for close calls, but here is my personal interpretation.

On Sep 20, 2012, at 12:28 PM, Janina Sajka <janina@rednote.net> (janina@rednote.net) <janina@rednote.net> wrote:

> I'm looking at the independent stacks criterion for CR exit and want to
> ask about it in a11y related real world terms ...
> 
> 1.)	We have cross platform user agents, e.g. Firefox available for
> Windows, Android, Linux, etc.
> 
> Are these separate stacks? Or one stack? How do we know for sure, i.e.
> the actual UI code is pretty different on each OS, is it not? Certainly,
> the assistive technology involved is different.
> 
> On the other hand, the point of a cross-platform app like Firefox is to
> reuse code as much as possible.

Firefox for different platforms would not count as independent, since the majority of code is shared, and even what is not shared is developed by the same party. Here is the definition of "independent" from the draft CR exit criteria:

"Each implementation must be developed by a different party and cannot share, reuse, or derive from code used by another qualifying implementation. Sections of code that have no bearing on the implementation of this specification are exempt from this requirement."

Clearly, Firefox ports for different operating systems would not count.

> 2.)	Are AT-SPI/ATK and IAccessible2 one stack or two? The latter is
> a derivative of the former implemented in a separate OS. The AAPIs look
> very similar, however.

I would say two different browsers using the same accessibility API counts as independent, just as two browsers on the same OS would be independent (assuming they otherwise meet the criteria). But the same browser using two different APIs would not be independent.

> 
> 3.)	What about KDE and GNOME on Linux/Unix? Two stacks or one? For
> accessibility both use AT-SPI, though KDE does not use ATK in order to
> talk to AT-SPI. The browser will clearly be different, yet the AT may be
> the same for KDE and GNOME browsers, because of the common AT-SPI layer
> (which is designed to support AT interoperability among toolkits).

If the assistive technology has a role in implementation of the specification (other than just processing generic commands from the browser), then I would not consider two different browsers talking to the same AT to be independent. In my opinion, two different, independent browsers talking to two different, independent assistive technologies would count as independent even if they used the same accessibility API as a connecting layer, so long as that API is just a general-purpose interface and does not itself materially relate to implementing the feature.

Regards,
Maciej

Received on Thursday, 20 September 2012 20:00:15 UTC