W3C home > Mailing lists > Public > wai-xtech@w3.org > April 2008

Re: Another summary of alt="" issues and why the spec says what it says

From: Leif Halvard Silli <lhs@malform.no>
Date: Sun, 20 Apr 2008 09:29:43 +0200
Message-ID: <480AF0E7.1020308@malform.no>
To: Steven Faulkner <faulkner.steve@gmail.com>
CC: joshue.oconnor@cfit.ie, Ian Hickson <ian@hixie.ch>, Jim Jewett <jimjjewett@gmail.com>, david.dailey@sru.edu, John Foliot <foliot@wats.ca>, HTML4All <list@html4all.org>, Dan Connolly <connolly@w3.org>, HTML WG <public-html@w3.org>, "Michael(tm) Smith" <mike@w3.org>, wai-xtech@w3.org, Al Gilman <Alfred.S.Gilman@ieee.org>

Steven Faulkner in reply to Joshue's question 2008-04-18 11.41:
> question - how does using aria-describedby help in this example (or
> rather how could it help as the expected AT behaviuor is not yet
> documented)
>
> <h2>Photograph 2</h2>
>   <img src=321098412.jpeg>
>   <p>View North East from 23 High Street.</p>
>
> AT [...] cannot be certain of the relationship [...]
>   

Indeed. <img> and <p> *could* be unrelated to each others.

> same thing with describedby added:
>
> <h2>Photograph 2</h2>
>   <img src=321098412.jpeg arai-describedby="photo2">
>   <p id="photo2">View North East from 23 High Street.</p>
>   

[Snipped the aria-describedby=idref programmatic association benefits.]

The aria-describedby approach adds new, hidden, meta information whic 
must be kept in order. Instead, I'd like to propose these solution:

1. Reserved keywords acting as CSS selectors:
Elements with an ALT attribute or fallback content could have reserved 
keywords which would point to the element containing its description: A 
"_prev" keyword could point to previous element, a "_next" to next 
element, a "_parent" to the parent element. For FIGURE, keywords are not 
needed and should not be taken account of, as long as FIGURE only 
contains LEGEND plus one single, embedding element.  But if there are 
more elements in FIGURE, then the keyword could link the IMG to a 
particular other element inside FIGURE. Or, if there are less elements 
(= only one IMG, and thus no LEGEND), then the keyword could link to 
outside the FIGURE element.

Option: The _parent and _next keywords could possibly be joined into 
just one "_auto" keyword.

Alternative syntax: _ for _auto, _* for _parent, _+ for _next and _- for 
_prev.

Pros:  1) If one adds lots of images in a repeated pattern of 
<IMG><DESCRIPTION> etc, then this method could be effective. 2) It is 
also possibly useful for hand coders. 3) It supports the (current) 
requirement of obligatory alt text. 4) It doesn't use a new IDREF based 
meta information attribute which must be kept in order.

Example:

    <h2>Photograph 2</h2>
      <img src=321098412.jpeg alt=_next >
      <p>View North East from 23 High Street.</p>
      

2. Automatic IMG description via TITLE or ALT matching:
Looking at how HTML 5's Automatic cross-referencing works, and leaning 
towards what web authors are used to when it comes to IMG - and the 
errors they commit (duplicate ALT texts), we could define an automatic 
feature where the programmatic association is achieved via a match 
between a describing element's content or TITLE attribute, and the TITLE 
or ALT content of the described IMG (the *nearest* match). And just as 
for the auto cross-ref feature, we must define which elements can be 
used as captions (I propsoe only P and H1-H6). Just like 
aria-describedby, this approach would work for all embedding elements.

Yes, I know that for the IMG element, I turn the issue of "duplicate alt 
text" on its head. AT must learn to tacle such duplicate alt text. Or, 
one could argue that the best thing would be to use the TITLE attribute 
of the embedding element.

For users of graphic User Agents, such a rule should not be much of a 
problem. For instance, getting the same TITLE for e.g. both e.g. a VIDEO 
element and a captioning element, could help draw the conclusion that 
the elements belong together, and perhaps give the impression that they 
are one and the same element.

This feature could work in tandem with the reserved keyword selectors. A 
reserved keyword selector would take presendence over this method.

Examples:

1)

    <h2>Photograph 2</h2>
      <img src=321098412.jpeg 
           alt="View North East from 23 High Street.">
      <p>View North East from 23 High Street.</p>
      

2)

    <h2>Photograph 2</h2>
      <img src=321098412.jpeg alt="Photograph 2">
      <p>View North East from 23 High Street.</p>

3) Here, could argue that it would be wrong using alt="", as it gives a 
conflicting message:

    <h2>Photograph 2</h2>
      <img src=321098412.jpeg alt="" title="photo 2">
      <p title="photo 2">View North East from 23 High Street.</p>


4)

    <h2>Photograph 2</h2>
      <img src=321098412.jpeg alt="photo 2">
      <p title="photo 2">View North East from 23 High Street.</p>

-- 
leif halvard silli
Received on Sunday, 20 April 2008 07:30:40 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 13:15:45 GMT