Re: ACTION-1710: Rich Get with Freedom Scientific on our solution for listitem roles

I thought nested links weren’t allowed? HTML 4 states that “Nested links are illegal <http://www.w3.org/TR/REC-html40/struct/links.html#idx-link-6>”, although I’m unable to find that sentence in HTML 5 spec <http://www.w3.org/TR/html5/links.html#links-created-by-a-and-area-elements>  Was this changed? Nu HTML Checker gives an error when I test <a href="#some">Some <a href="#link">Link</a></a>, but doesn’t specify why.

—Michiel

> On 01 Sep 2015, at 01:27, Richard Schwerdtfeger <schwer@us.ibm.com> wrote:
> 
> Hi Bryan, 
> 
> Let me explain. In HTML an anchor tag can span lots of content. It is essentially a container. You could in fact have paragraphs of text and other links inside it. So, essentially it is a container. 
> 
> So, where this all becomes a problem is knowing how to convey the label for the link and the contents for the link in the virtual buffer. Do you stop at just the label or do you include the content within the link. Many developers are being told they must apply a label to a listitems and apparently that is happening on links. If we require a name and the author has not applied a label then we have an accessible name on the link computed using name from content. Then you have the content within the link. 
> 
> If instead we say we don't require a name and name from content is not supported this then allows the author to apply a label to the link as is the link was a container, and the name is derived from that, if the author so chooses. At this point the content of the link (could be paragraphs of text) could then be accessed as content within the link. We also, never have a name that is computed from the content and then the content of the link in the screen reader view right after it. 
> 
> Does this help? I think the issue is that role="link" and anchor tags really can span large amounts of content the same way as a list item.
> 
> Cheers,
> Rich
> 
> 
> Rich Schwerdtfeger
> 
> <graycol.gif>Bryan Garaventa ---08/31/2015 06:12:14 PM---I understand the logic about not requiring labels on role=listitem, but I don't understand the logic
> 
> From: Bryan Garaventa <bryan.garaventa@ssbbartgroup.com>
> To: Richard Schwerdtfeger/Austin/IBM@IBMUS, PF <public-pfwg@w3.org>
> Cc: "cooper@w3.org" <cooper@w3.org>
> Date: 08/31/2015 06:12 PM
> Subject: RE: ACTION-1710: Rich Get with Freedom Scientific on our solution for listitem  roles
> 
> 
> 
> I understand the logic about not requiring labels on role=listitem, but I don’t understand the logic about role=link.
>  
> A link is an interactive widget, and like all others it requires a label, otherwise it will be an inaccessible link that has no apparent purpose to the AT user when it receives focus.
>  
> I may be misunderstanding the bug though, so please correct me if I am wrong.
>  
>  
>  
> From: Richard Schwerdtfeger [mailto:schwer@us.ibm.com <mailto:schwer@us.ibm.com>] 
> Sent: Monday, August 31, 2015 4:04 PM
> To: PF <public-pfwg@w3.org>
> Cc: cooper@w3.org
> Subject: ACTION-1710: Rich Get with Freedom Scientific on our solution for listitem roles
>  
> I met with Freedom Scientific and they support our solution regarding a label not being required on listitem roles and that the name not being computed from content. 
> 
> What they asked is that we do the same thing for role="link". I have created a new issue (741) for this: https://www.w3.org/WAI/PF/Group/track/issues/741 <https://www.w3.org/WAI/PF/Group/track/issues/741>
> 
> We still need to wait on an answser from Actions 1711 and 1712. 
> 
> Michael, please close action 1710 on the call Thursday as being completed. I am leaving it open so that the group can discuss it for the call. 
> 
> Rich
> 
> Rich Schwerdtfeger
> 
> 
>