W3C home > Mailing lists > Public > public-html@w3.org > October 2009

Re: ISSUE-30 (Longdesc) Change Proposal

From: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
Date: Tue, 27 Oct 2009 15:31:25 +0100
Message-ID: <4AE7043D.5000803@xn--mlform-iua.no>
To: Lachlan Hunt <lachlan.hunt@lachy.id.au>
CC: "public-html@w3.org" <public-html@w3.org>
Lachlan Hunt On 09-10-27 13.10:

> Leif Halvard Silli wrote:
>>> On Mon, Oct 26, 2009 at 11:26 AM, Leif Halvard Silli
>>>> Another argument for this feature is, I think (as have been mentioned
>>>> earlier) that aria-describedby="" can be used for the same thing.
>> I found Lachlan's comment in September: [1]
>>
>>> That would just be reinventing longdesc with a different name without
>>> solving any of the problems that longdesc has.
>> However, upon rereading, it seems like Lachlan was actually expressing
>> satisfaction that describedby and longdesc has not been defined the same
>> way. There are other comments in that same thread expressing similar
>> things.
> 
> My take on longdesc is that it's badly designed for the problem it's 
> trying to solve, and even aria-describedby is not ideal for addressing 
> the cases where the description is in an external document.


Ok. Well, then my first recollection of what your reply meant was 
more correct than the second ...

 
> I would rather see the problem of how to provide access to a long 
> description in an external document addressed in a way that can utilise 
> ordinary links, without introducing too much complexity.  There are 
> several ways this could be achieved.
> 
> 1. Making the image itself a link to its description:
> 
> <a href="desc.html" rel="longdesc"><img src="image" alt="..."></a>

(1) Then the old truth is that then you can't discern between a 
longdesc link and a link, as this isn't possible/allowed:

<a href="a"><a href="b" rel="longdesc"><img></a></a>

(2) Further, it doesn't (in practice) mean the same thing. E.g in 
my previous message I gave this example:

<p>The <i aria-describedby="#image">diagram</i>
<p><img id="image" longdesc="diagram-description.htm" src=img
alt="Diagram of something". >

With your proposal here, it would have to be written:

<p>The <i aria-describedby="#image">diagram</i>
<p><a rel="longdesc" href="diagram-description.htm"><img 
id="image" src=img  alt="Diagram of something". ></a>

(And it doesn't matter whether you make the <i> into an <a>.)

When a sighted user discovers there is a link on the IMG, he/she 
is unlikely to understand that it is only meant to point to a long 
description of the IMG.

> The rel=longdesc attribute should be enough for ATs to identify the link 
> as being a long description, and the fact that the image is inside the 
> link should be enough to make the association.
> 
> It's also significantly easier than having to add an id and matching 
> aria-describedby attributes,


I don't understand why you think your proposal requires less use 
of @id. I can't see it does.

> and is far more resistant to things like 
> copy/paste errors, commonly seen with, e.g., the label/for attribute, 
> and which is likely to occur with aria-describedby too.  (In fact, it's 
> likely to occur more often with aria-describedby since, unlike 
> label/for, there's no easy way to detect mismatched IDs/IDREFs).

It seems you are arguing for rel="longdesc" as - in general - 
better than aria-describedby="". But that position seems like a 
lost cause, from the beginning.

I don't think that hidden anchors is any good, if it can be 
avoided. Firstly it requires that you use CSS to hide it. I think 
that is called indirection. And the link between CSS and mark-up 
easily gets lost. It is also not clear what happens if you copy a 
hidden link ... Will it get copied?

The semantics should be where they belong: in the document.

> 2. If the image has a caption, including the link within that:
> 
> <figure>
>    <img src="image" alt="...">
>    <legend>Some caption here.  <a href="desc.html" rel="longdesc">More 
> information</a>.</legend>
> </figure>
> 
> (Using <legend> for now cause the dt/dd idea needs to be dropped and we 
> don't have another alternative in the spec yet)

Using a caption is certainly a good idea, in general. But I don't 
feel that it is directly related to the issue at hand.

> 3. Using an adjacent link that shares the same parent as the image 
> itself, and implying the association:
> 
> <p><img src="image" alt="..."> <a href="desc.html" rel="longdesc">More 
> information</a></p>
> 
> This would need to have an implicit association algorithm defined to 
> handle cases where there are more than one image and/or description links.

As opposed to @longdesc, which "just works".
-- 
leif halvard silli
Received on Tuesday, 27 October 2009 14:37:00 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:39:09 UTC