Bug 61013 – Keep linked images in the accessibility tree.

• Component: Accessibility
• Status: RESOLVED INVALID

[Chris Guillory]

It's useful for the user to know that the selected link contains a graphic. If we keep graphics that are children of links in the accessibility tree then windows screen readers like JAWS and NVDA will read out the more descriptive "link graphic" instead of simply "link" when the link is focused.

[Chris Guillory]

Created attachment 94559 [details]
Patch

[Chris Guillory]

Created attachment 94712 [details]
LayoutTest now passes on Mac

Chromium issue is http://crbug.com/36581. This issue is also present on Safari for windows.

[chris fleizach]

Comment on attachment 94712 [details]
LayoutTest now passes on Mac

i assume this is meant to be a patch since its review?

[chris fleizach]

Comment on attachment 94712 [details]
LayoutTest now passes on Mac

View in context: https://bugs.webkit.org/attachment.cgi?id=94712&action=review

> Source/WebCore/ChangeLog:10
> +

I may be dense, but this comment doesn't make sense to me. 
Are you saying... "don't expose the src of the link.... and to do that, you override the image src?"
a) I don't understand why you don't want to expose the source of the link.
b) I don't understand why you want to change what the URL is for the image. Right now it should be the the location of the image. That seems correct in my mind, because the src of the link is returned in the link

> Source/WebCore/accessibility/AccessibilityRenderObject.cpp:1872
>                  return true;

this seems like it alters the informal standard that's pretty clear. if alt is empty, image should be ignored

> LayoutTests/accessibility/tree-contains-linked-images.html:24
> +     }

this layout test doesn't seem to test the altered behavior. this <img> was not ignored previously because it had an alt tag.

[Chris Guillory]

Comment on attachment 94712 [details]
LayoutTest now passes on Mac

View in context: https://bugs.webkit.org/attachment.cgi?id=94712&action=review

>> Source/WebCore/ChangeLog:10
>> +
> 
> I may be dense, but this comment doesn't make sense to me. 
> Are you saying... "don't expose the src of the link.... and to do that, you override the image src?"
> a) I don't understand why you don't want to expose the source of the link.
> b) I don't understand why you want to change what the URL is for the image. Right now it should be the the location of the image. That seems correct in my mind, because the src of the link is returned in the link

The comment could use some rewording.
a) The source for links remains unchanged.
b) The change includes images without alt text that are parented to linked in the accessibility tree. However, when a screen reader (JAWS, VoiceOver) focuses the link, the image's src is read. Previously the link's src was read. I think it's incorrect to override the images src to address this behavior. I'll look into an alternative.

>> Source/WebCore/accessibility/AccessibilityRenderObject.cpp:1872
>>                  return true;
> 
> this seems like it alters the informal standard that's pretty clear. if alt is empty, image should be ignored

Firefox and IE include images with alt text in the accessibility tree. JAWS reads them more conveniently as "link graphic" instead of just "link". I'll look more into the general update I'm requesting for the informal standard.

>> LayoutTests/accessibility/tree-contains-linked-images.html:24
>> +     }
> 
> this layout test doesn't seem to test the altered behavior. this <img> was not ignored previously because it had an alt tag.

Oops. I accidentally left that in.

[chris fleizach]

(In reply to comment #5)
> (From update of attachment 94712 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=94712&action=review
> 
> >> Source/WebCore/ChangeLog:10
> >> +
> > 
> > I may be dense, but this comment doesn't make sense to me. 
> > Are you saying... "don't expose the src of the link.... and to do that, you override the image src?"
> > a) I don't understand why you don't want to expose the source of the link.
> > b) I don't understand why you want to change what the URL is for the image. Right now it should be the the location of the image. That seems correct in my mind, because the src of the link is returned in the link
> 
> The comment could use some rewording.
> a) The source for links remains unchanged.
> b) The change includes images without alt text that are parented to linked in the accessibility tree. However, when a screen reader (JAWS, VoiceOver) focuses the link, the image's src is read. Previously the link's src was read. I think it's incorrect to override the images src to address this behavior. I'll look into an alternative.
> 
> >> Source/WebCore/accessibility/AccessibilityRenderObject.cpp:1872
> >>                  return true;
> > 
> > this seems like it alters the informal standard that's pretty clear. if alt is empty, image should be ignored
> 
> Firefox and IE include images with alt text in the accessibility tree. JAWS reads them more conveniently as "link graphic" instead of just "link". I'll look more into the general update I'm requesting for the informal standard.
> 

Do you mean that Firefox and IE include images "without" alt text if they are inside links?


> >> LayoutTests/accessibility/tree-contains-linked-images.html:24
> >> +     }
> > 
> > this layout test doesn't seem to test the altered behavior. this <img> was not ignored previously because it had an alt tag.
> 
> Oops. I accidentally left that in.

[James Craig]

Abandoned. Closing as invalid.