Re: ISSUE-30 (Longdesc) Change Proposal

Jonas Sicking On 09-10-30 16.07:

> On Fri, Oct 30, 2009 at 4:24 AM, Joshue O Connor:
>> [Please forgive any crosstalk/mis-attribution in my quotes below]
>>
>> Jonas Sicking wrote:
>>> This seems like something you should bring up with the WAI group. They
>>> were the ones choosing to design @aria-describedby by throwing away
>>> @longdesc rather than evolving it.
>> This is something that we have looked at. We can escalate it if needed
>> in PF. Please note that the development of @aria-describedby etc (in
>> fact any ARIA stuff) was a response to the limitations in current
>> declarative markup languages. So ARIA stuff is a semantic bridge. There
>> was therefore bound to be some overlap with existing elements in HTML
>> but then again ARIA is host agnostic and can be plugged into other
>> languages like SVG etc where there is no <longdesc>. So its kinda
>> disingenuous to suggest there was some causal link.
> 
> Sorry, I didn't mean that the WAI did a poor job in designing
> @aria-describedby by not following any particular syntax. What I
> intended to say was that comments regarding @aria-desribedbys syntax
> should be brought up with WAI, not with the HTML WG.

Well, thoughts along the line of "but now we can shuffle 
everything of this over to ARIA" have originated in this group. 
That said: We need to understand how HTML and ARIA relates, and it 
is good that we have a dialog about that. And it it is our task to 
understand what we have in @longdesc.

> I actually happen to like the fact that @aria-desribedby uses an idref
> rather than a URI, since I think in most cases its more appropriate to
> include the description in the same document as the <img> or whatever
> is begin described. If for no other reason because it makes it more
> likely that the two won't get separated when a file is moved or
> copied.


+1 to most of this. I think it is brilliant that aria-describedby 
and aria-labelledby are IDREFS. And perhaps it would only be 
confusing if they could also link to outside page resources.

But my reasoning *why* I think so is not so much that "it is 
logical that all the A11Y stuff should be kept inside the same 
page" as "all the context of the image normally is and will be 
kept within the page".

Because, as I see it aria-describedby/aria-labelledby is primarely 
meant to link up the context that is already present. That it may 
also be used to link up hidden texts that have been crafted/added 
especially for A11Y purposes, is important, but very often secondary.

I think we should retain the HTML 4 understanding of @longdesc as 
fallback for IMG. And I think that it is ARIA that must take this 
in. @longdesc, however, may link to both a in-the-page fallback 
text, as well as to an external fallback text.

 
> Note that being in the same resource doesn't mean that I think the
> description needs to be displayed by default. I continue to think that
> hiding the description using CSS or using the @hidden attribute and
> then pointing to it using @aria-describedby is a good solution.

It is a good solution, for UAs that support ARIA. It is not a good 
solution to make a fuzz about using "display:none" instead of 
"position:absolute; left:-999999999999999999px;" ...

>>>>> * Clearer message to authors for how to make their pages accessible
>>>> We can have a clear message on the proper implementation of @longdesc that
>>>> would be simple to understand and deliver upon.


If it was you Jonas who said this, then I am all for it: "A clear 
message on the proper implementation of @longdesc" etc.

>> If @longdesc does make the final cut, then yes.
> 
> I understand what is being said here at all. What I was saying was
> that i think it's important to have a clear and simple message to
> authors on how to include a description for an element.
> 
> The replying comments seem to be talking about something wholly different.
> 
>>>>> * Simpler AT tools
>>>> ??  AT *today* supports @longdesc - I personally do not think that they
>>>> are going to now remove this support in future versions. Why would they?
>>>> Just to replace it with aria-describedby? Really?
>> I doubt it. The two will probably be supported in tandem - for legacy
>> reasons - even if the use of @longdesc is very small. Its a case of
>> those who find it useful - well, find it useful.
> 
> If implementations want to keep implementing @longdesc that is fine.
> The big win to me is simplifying our message to authors regarding how
> to make pages accessible.


And I don't think we do that with *just* aria-describedby. And it 
should not be only about A11Y. Fallback is for everyone. While 
ARIA isn't. For authors it is simpler to think "ok, I provide a 
long description for this - this is the fallback". Than to use 
aria-describedby and possibly include several IDREFS. Which of 
those IDFREFS are *THE* fallback?

We don't want to UAs to implement @longdesc because they are more 
  kind than HTML 5 and wanted to be. We want them to implement it 
because it has a purpose. And ditto for HTML 5.

 
> I.e. being able to say "use @aria-describedby" is a win over saying
> "use @aria-describedby, except on an image where you can use
> @longdesc, except you probably don't want to use @longdesc if the
> description is in the same page since then if the base-uri is set
> using <base> since then the relative link in @longdesc no longer
> points inside the resource. And on tables you can also use @summary"

Honestly: You can say it much clearer than that. But it first 
requires that you set yourself in some kind of "be constructive" 
modus. (And please stop continuing to mix @summary into this.)

HTML 4 tells you how to use @longdesc without mixing in ARIA a 
single time. HTML 5 should continue to do the same. It is ARIA 
that needs to take in @longdesc, not the other way around. And if 
one has to mention ARIA in connection with @longdesc, then that is 
because we have gotten ARIA ... we did not have it before.

HTML 5 should say that @longdesc can be used to link to a single 
resource or element that is meant as a *direct* long description = 
fallback, of the IMG in question.

ARIA-describedby may link to further context (several IDFREFS, 
unlike @longdesc) that describes the IMG. If the longdesc URI 
points to an element inside the same page, then ARIA-describedby 
may include an IDREF to the very same long description element. 
However, the ARIA text equivalent computation algorithm must take 
care of this to avoid recursive references etc.
-- 
leif halvard silli

Received on Friday, 30 October 2009 16:08:56 UTC