W3C home > Mailing lists > Public > public-html-a11y@w3.org > November 2012

Re: longdesc using #id

From: Charles McCathie Nevile <chaals@yandex-team.ru>
Date: Sun, 25 Nov 2012 23:00:02 +0400
To: "'Steve Faulkner'" <faulkner.steve@gmail.com>, "'HTML Accessibility Task Force'" <public-html-a11y@w3.org>, "David MacDonald" <david100@sympatico.ca>
Message-ID: <op.woce2ckzy3oazb@chaals.local>
Chair hat off.

Summary: It seems like we are on the same page with regards to what we
think the technology does.

We have disagreements about history. I don't think they are so important
here, but if you want to know what they, read on...

On Fri, 23 Nov 2012 14:58:46 +0100, David MacDonald
<david100@sympatico.ca> wrote:

> I also concur with Chaals that this is a use case that would be good for
> Longdesc, and that HTML4 allows for it, but that the spec provided no
> example, and did not mention it ...
>  I support its inclusion in our extension... I just think we're being a  
> bit strident to say its non-implementation for over 15 years is a "bug".

It is in the specification, but doesn't work in a user agent. That's what
I think a bug is. (This isn't a particularly controversial statement, or a
moral judgement, so I don't think it is a priori "strident").

> 	>> ... there have been implementation reports that some screen
> reader / browser combinations don't handle this well.
> Is there one implementation (Browser/AT combination) that *does* handle  
> it well? Or was there *ever* a combination that did?

Apart from pwWebSpeak, Opera with VoiceOver handles this case fine. Try
the demo at
(note that Opera has not managed to make good screen reader compatibility
work, so the standard way to open the longdesc is using the context menu)

Equally, add the extension from
http://www.splintered.co.uk/extensions/longdesc_0.6.xpi to Firefox and use
it with your favourite screen reader. The extension is explicitly designed
to work with internal links as well as external pages which is why there
is an example like that in the test. (This work was done between 2005 and

iCab also implements this on Mac. It has been doing it for many years -
since around 2000 or 2001 if I recall correctly.

> I think we should say it like it is... that HTML4 does allow for a #id to
> point to the same page, but no one ever did...

I don't think so, because that statement is not correct.

> and at the time, if I remember correctly, the team was looking for an
> alternative to "D" link, and the thinking was a description separate
> page, because they thought sighted people didn't need the description…

This doesn't match my recollection. But different people involved were
thinking along different lines. Some people may well have thought this
way, others didn't.

I think the key is that the resulting specification and implementations
allow this, and it has been implemented and used, although not

> (there was not much understanding at the time of the needs of the  
> cognitive community)
> There is room to interpret HTML4 this way, but there has never been an
> implementation of it in 15 years, and therefore it is effectively a new
> proposed behaviour,

I don't think that this is historically correct.

> which I wholeheartedly support.

That's the important thing :)



> Cheers
> David MacDonald
> CanAdapt Solutions Inc.
>   Adapting the web to all users
>             Including those with disabilities
> www.Can-Adapt.com
> -----Original Message-----
> From: Steve Faulkner [mailto:faulkner.steve@gmail.com]
> Sent: November-23-12 6:36 AM
> To: HTML Accessibility Task Force
> Subject: Re: longdesc using #id
> Chair hat off.
> I concur with chaals on this.
> Regards
> Steve
> On 23 Nov 2012, at 14:21, "Charles McCathie Nevile"  
> <chaals@yandex-team.ru>
> wrote:
>> (chair hat off)
>> Hi,
>> there has been some discussion about longdescs whose URL is a relative
> reference to somewhere in the page. And there have been implementation
> reports that some screen reader / browser combinations don't handle this
> well.
>> I think that is a bug in the relevant software (maybe screenreader,  
>> maybe
> the underlying browser). There is a demo of longdesc that Patrick Lauke
> wrote to go with his Firefox extension, last updated in 2008, which uses
> internal page links, and the extension as well as the browser-native
> implementations of iCab and Opera all handle the case of an internal
> reference just as they would an external one.
>> In the original HTML specification of longdesc, the value of the  
>> attribute
> is given as URI, and that can be an internal or an absolute link (the  
> same
> as the href attribute for a link). I don't know of any evidence that
> longdesc was intended to only work on external pages, despite some
> implementations failing if you don't do that.
>> During the seemingly endless discussions of ISSUE-30 in the HTML WG, one
> argument raised against longdesc was that it forced the description to  
> be on
> a separate page. This assertion is not backed up by the spec, but by some
> implementations.
>> It seems to me that there are valid use cases for having the description
> on the same page as an image, and pointing to it. So I think the HTML 4  
> spec
> was right, and we should re-affirm that, and file bugs against the  
> software
> that doesn't implement it properly. (I have already done this for the  
> Yandex
> browser...)
>> cheers
>> Chaals
>> --
>> Charles McCathie Nevile - Consultant (web standards) CTO Office, Yandex
>>      chaals@yandex-team.ru         Find more at http://yandex.com

Charles McCathie Nevile - Consultant (web standards) CTO Office, Yandex
         chaals@yandex-team.ru         Find more at http://yandex.com
Received on Sunday, 25 November 2012 19:00:33 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:05:32 UTC