name and description calculation - is there a mismatch between the document and what Apple implements?

Joanie,

My apologies for being confused.  I have a few questions -
Are the testable statements in the wiki in sync with what is implemented or
what you believe is implemented? If not, what do I need to fix?

My understanding is sometimes the AXAPI column in the accessible name and
description table doesn't always reflect Apples implementation. Is this a
correct understanding?

If there is sometimes a mismatch between the table and Apple's
implementation can you describe the mismatch?  Again, sorry for being
confused, I know you have described what is going on before - but it would
help me to have it spelled out again.  Thanks



                                                              
     Regards,                                                 
                                                              
    Fred Esch                                                 
 Watson, IBM, W3C                                             
  Accessibility                                               
                                                              
 IBM Watson       Watson Release Management and Quality       
                                                              






From:	Joanmarie Diggs <jdiggs@igalia.com>
To:	Fred Esch/Arlington/IBM@IBMUS
Cc:	public-svg-a11y@w3.org
Date:	04/01/2016 02:02 PM
Subject:	Re: SVG accessibility new issues



Hey Fred, all.

Some clarifications regarding this item:

On 03/30/2016 03:43 PM, Fred Esch wrote:

> *Difference between testable statements in wiki and spec*
> Webkit doesn't follow the spec for name - the spec has requirements for
> exposing labelledby, webkit doesn't bother - neither linux (Joanie) or
> apple see a current need for exposing labelledby

Joanie doesn't see any point to exposing a pointer via AXTitleUIElement
in the case of OS X if Apple doesn't see the need, VoiceOver doesn't do
the wrong thing without this pointer, etc. It's Apple's platform; not
Joanie's. And Joanie respects that. :) And the labelledby value IS
exposed through the computed name (aka AXDescription).

Joanie DOES think the pointer should be exposed in Linux. This exposure
is in addition to exposure through the computed name.

Furthermore, the pointer *is already being exposed* for SVG elements
with aria-labelledby. Joanie fixed that bug (along with some parallel
issues) upstream a couple of weeks ago:
https://trac.webkit.org/changeset/198254.

For those of you who like layout tests (who doesn't love layout tests?):
If you look at
https://trac.webkit.org/browser/trunk/LayoutTests/platform/gtk/accessibility/w3c-svg-name-calculation-expected.txt#L1
,
you'll notice a bunch of "AXTitleUIElement: non-null" lines (In the case
of Linux, "AXTitleUIElement" is layout-test speak for "the object
referenced by ATK_RELATION_LABELLED_BY". Thus that file documents the
fact that WebKit DOES bother for Linux.)

Sorry for any confusion or lack of clarity on this matter in my previous
communications. And please let me know if anything remains unclear.

Thanks!
--joanie

> https://lists.w3.org/Archives/Public/public-svg-a11y/2016Mar/0049.html
> Rich filed a bug on my request and if we agree labelledby doesn't need
> to be expose we need to remove the bug
>
>
>
> Regards,
>
> Fred Esch
> Watson, IBM, W3C Accessibility
> IBM Watson		 Watson Release Management and Quality
>
>
>