W3C home > Mailing lists > Public > public-html-a11y@w3.org > March 2010

Re: summarization information delivery options: attribute or element

From: Shelley Powers <shelleypowers@burningbird.net>
Date: Tue, 02 Mar 2010 10:56:54 -0600
Message-ID: <4B8D4356.8020908@burningbird.net>
To: Leif Halvard Silli <xn--mlform-iua@xn--mlform-iua.no>
CC: public-html-a11y@w3.org
Leif Halvard Silli wrote:
> Shelley Powers, Tue, 02 Mar 2010 07:59:42 -0600:
>> Leif Halvard Silli wrote:
>>>> Though technically not a _new_ element [1], using two captions on a 
>>>> table is not, I believe, an enhancement that will degrade 
>>>> gracefully with older user agents.
>>>> Firefox doesn't print out the second caption, as Leif states in his 
>>>> change proposal, but that's not a behavior we can count on.
>>> Right. We need support in more than one UA.
>> Unfortunately, the second caption is not backwards compatible. It 
>> breaks with existing and older browsers.
> I think you are rushing to conclusions. On this page, I have both 
> hidden and shown second captions, and it works for Webkit, IE, Opera 
> and Firefox:
> http://målform.no/html5/caption+role
> You could test it in NVDA if you want, If NVDA doesn't read even the 
> *visible* second <caption>s, then we for sure have a problem. As long 
> as NVDA reads the visible captions, then trick to serve a hidden 
> caption to NVDA is some variant of: 
> 	a) make second caption visible via CSS, and then use
> 	b) caption+caption{position:absolute;left:-9999cm} 
> 	  (or something similar) to visually hide it. (I will soon publish a 
> demo of this.)
Let me ask you something: do you think that a web designer or author 
will do this? Let's forget UAs and parsing, and what NVDA will speak or 
not: do you think a web designer or developer will do this? 

What a lot of this comes down to is one sentence you wrote:

"We already have the summary attribute, which has as drawback that it is 
an attribute, and attributes are, by definition, not as accessible 
whether for those in need of AT software nor for the average Web author. "

In my opinion, the summary attribute is a lot more palatable than a 
second caption with some not very attractive CSS. I can't speak for all 
web authors and designers and developers, but I wouldn't use it. Sorry, 
I just wouldn't.

Now, you mention @summary is not accessible for AT software. How so? As 
far as I know of, summary does work with AT software. In fact, as far as 
I know of, that's never been an issue with @summary.

> I am well aware that some UAs hide a second caption while others reveal 
> it. So a second caption *must* be followed by advice about how to use 
> CSS to hide/reveal it. Isn't that acceptable? How is that different 
> from introducing a new element, of which HTML5 has quite a few, and 
> which also needs advice to authors about how to style it?
> Some of us have debated a <summary> element directly as child of 
> <table>. The issues with that are also related to visibility (I think), 
> but even more is it related to whether the UA spits the element out of 
> <table> or not - i.e. it removes it from what it is supposed to be 
> related to.

Yes, UAs split out summary, div, etc elements. Not caption, true.
>>>> Opera does print out the second caption. Safari and Chrome do not. 
>>>> IE does. NVDA says one caption when the page is loaded in Firefox. 
>>>> NVDA says both captions, when the page is loaded in IE.
>>>> The results we will get for a second caption are unreliable.
>>> My CP specifically says that we must investigate what the best CSS 
>>> is. It sounds from your description like NVDA behaves very much like 
>>> VoiceOver does: It only reads that what is somehow visible. 
>>> VoiceOver e.g. doesn't read the content of @summary. Thus it reads 
>>> the second caption, but only when it is somehow visible.
>>> So, can you also tell us how NVDA behaves with regard to @summary?
>> It worked on mouse over. It might work in other ways, but that was 
>> the only one I tried. I'm still pretty new to NVDA.
> OK. Thanks for mentioning NVDA though. I will also try to test it 
> myself as soon as I get time.

Received on Tuesday, 2 March 2010 16:57:33 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:55:33 UTC