Re: feedback requested on WAI CG Consensus Resolutions on Text alternatives in HTML 5 document

Hi Steven,

Here's my feedback based on reading your document and the spec.

On Aug 15, 2009, at 5:52 AM, Steven Faulkner wrote:

> Hi Ian,
>
> as part of my work on http://www.w3.org/html/wg/tracker/actions/131,  
> to progress towards consensus by the html wg on the contents of the  
> html 5 specification in regards to text alternatives, it would be  
> helpful to get feedback from you and other interested people on the   
> 'WAI CG Consensus Resolutions on Text alternatives in HTML 5'  
> document http://www.w3.org/2009/06/Text-Alternatives-in-HTML5

Your text does not seem that far off from what is currently in the  
spec. Your version allows the following to serve as the text  
alternative (or short description if you prefer) for an <img>:

(1) alt
(2) the <legend> of a <figure>
(3) aria-labelledby
(4) role="presentation" as a synonym for alt=""

The HTML5 spec allows the first two. The last two use ARIA markup, and  
will be allowed in due course.

In the exceptional case of images whose contents are unknown, the spec  
allows one of the following alternatives for giving the best available  
information about an image, in addition to some of the techniques above:

(5) the title attribute
(6) a section heading for a section containing nothing but the image

I believe this is to distinguish text *about* the image from a textual  
*equivalent* for the image, when text about the image is the best  
available.

The validator will currently warn if none of techniques (1), (2), (5)  
or (6) is used. I expect when ARIA is integrated, it would likely  
allow (3) and (4) as well. The image report feature of the validator  
will alert the author when they have only used techniques (5) or (6),  
reducing the risk this will happen inadvertently.

The spec also allows textual alternatives to be omitted entirely in  
private communication among known parties. The validator does not  
currently report an error for completely missing textual alternative,  
but it is supposed to according to the spec, so I expect this will be  
fixed: "Conformance checkers must report the lack of an alt attribute  
as an error unless the conditions listed above for images whose  
contents are not known or they have been configured to assume that the  
document is an e-mail or document intended for a specific person who  
is known to be able to view images." The conditions referred to are  
use of title, use of a figure legend, or the image being the sole  
image in a section with a heading.

Thus, I believe the spec more or less meets the goals of the text  
alternatives resolution you posted. It will flag inadvertantly missing  
alt, but it allows a few extra techniques for the case where image  
contents are unknown, and does not yet incorporate ARIA techniques.

For example, in your "Use Case 2" example, rather than suggesting alt  
text of "Photo 1 of 50 of album Paris 2009", HTML5 recommends that  
text should go in a title, a figure legend, or a section heading for a  
section containing only the photo, to make clear that the text is a  
description, not an equivalent. (Presumably, your proposal would also  
allow a figure legend.)

I invite you to read the current spec text on this and see if you  
think it is reasonable and roughly in line with the proposal you posted.
<http://dev.w3.org/html5/spec/Overview.html#alt>

Regards,
Maciej