Re: dropping longdesc attribute

Denis Boudreau (WebConforme) wrote:
> 
> Evening all,
> 
>>> Anyway, what rationale for dropping longdesc?
>>
>> IIRC, the reason for not including it was because very few authors 
>> ever use it and, when it is used, it's not used properly.
> 
> OMG, that argument again. Who cares if it's useful.

If a feature isn't useful, there is no need for it.  But, since I didn't 
say it wasn't useful, I'm going to assume you actually meant who cares 
if it's used properly.  In that case, users should care!  If it's not 
used properly, it doesn't benefit anyone.

> Quickly, let's trash h1 to h6 also - and let's not forget blockquote while 
> we're at it.

The difference is that it's easy to find significant evidence to justify 
the inclusion of h1 to h6 and blockquote.  It is not so easy to find 
evidence for longdesc.

> Longdesc are essential for screen readers to provide long descriptions 
> for graphics that otherwise could not be described because the nature of 
> their content is just too complicated for a simple alt attribute.

A description of how it benefits users when used properly doesn't 
constitute evidence of it being used properly.  Although I'm quite sure 
users would benefit when they stumble across a site that uses it 
properly, the question is does it get used properly on a significant 
number of sites?

> Is there a single accessibility feature in HTML that will survive this 
> group? Why is the left hand (html-wg) so bluntly ignoring what the right 
> one (wai-wg) does?

Contrary to what you may think, I am not against accessibility features 
in HTML.  We just have different opinions about the best way to achieve 
it.  A lot of people seem to be advocating the use of accessibility 
add-ons, like headers="" and longdesc="".  However, a much better 
approach to accessibility is to try and cater for it as part of the 
fundamental design, not include it as an afterthought.

For that reason, I would prefer if accessibility could be achived 
without things such as headers="" and longdesc="".  That is why, for 
example, I essentially argued that it would be better for accessibility 
if the headers attribute was redundant in most (if not all) cases, since 
it would achieve the same result with less effort from authors.  That is 
why so many arguments against headers were trying and show that the 
example tables provided did not need to use the headers attribute.  That 
is why so much evidence got question because the headers attribute was 
either redundant, or used wrongly.

>> More research on the issue is welcome.  In particular, evidence of 
>> significant real world usage that provides a practical benefit to 
>> users would be good.
> 
> And why should we bother? There has been a lot of efforts made 
> previously by John (Folliot) and others in order to save summary and 
> headers in tables. Still, the draft hasn't backed out one bit on the 
> subject.

Ok, let's put this in perspective, shall we?  Hixie has about 5000 
unanswered emails (if not more) relating to hundreds of unresolved 
issues, spanning over 3 years of development.  Rest assured that the 
headers issue will be dealt with in due course.  A final decision has 
not yet been made.

> Lachlan, why wouldn't you, for a change (as John had so eloquently
> asked previously), provide evidence that this attribute is useless?

I am not arguing that it's useless.  My argument is simply that there is 
not yet enough evidence to support its inclusion and that we should 
investigate better alternatives, if possible.

Anyway, the following are the results of a search containing about 100 
billion <img> elements [1].  About 80 million (0.08%) had a longdesc 
attribute specified.  Of those, about 33 million were blank.  From this, 
we can conclude that about 41% of all longdesc attributes are definitely 
used incorrectly.  That leaves an upper bound of about 47 million 
(0.047%) longdesc attributes in the sample that may be used correctly. 
However, it is quite likely that many of those don't contain a URI, or 
link to a description that isn't particularly useful.

But as I said, more research is welcome.  A final decision has not been 
made on this issue.

> Those of us who work with screen reader users on a regular basis
> know that this attribute is extremely relevant when it is used properly.

Then you should have plenty of evidence available to show where it has 
been used in useful ways.

I know wikipedia uses it, but the question is, is it used in a useful 
way?  This is an example taken from wikipedia:

<a href="/wiki/Image:Stargate_SG-1_Season_9_Title.jpg" class="image" 
title=""><img alt="" 
longdesc="/wiki/Image:Stargate_SG-1_Season_9_Title.jpg" 
src="http://upload.wikimedia.org/wikipedia/en/thumb/d/d5/Stargate_SG-1_Season_9_Title.jpg/250px-Stargate_SG-1_Season_9_Title.jpg" 
height="141" width="250"></a><br>
<small>Stargate SG-1 intertitle (Seasons 9-10)</small>

The longdesc attribute points to the same location as the link.  So, 
arguably, nothing would be lost if the longdesc was removed.  It's also 
questionable whether the page it links to actually qualifies as a long 
description.  The image pages on wikipedia generally don't describe the 
image in any useful way for assistive technology.  They usually just 
provide a larger version of the image, copyright info, file history, etc.

[1] (Source: Unpublished Google internal survey of several billion pages 
conducted in September 2006.)

-- 
Lachlan Hunt
http://lachy.id.au/

Received on Saturday, 23 June 2007 05:17:25 UTC