Re: ALT revisited

At 19:15 1997/06/26 -0400, Al Gilman wrote:
>A little data I just learned:
>
>In Netscape, setting "autoload images" off does not guarantee the
>delivery of the ALT text to the screen.  It merely suppresses net-gets
>of the image file.  If the image file is already in local cache,
>the image is displayed and the ALT text is nowhere on the screen.

Hackery.
>[Flushing the cache is more bother than a user should have to put
>up with, and on a workgroup server the individual user does not
>have a dedicated cache to flush once and forget.]
>
I find the MSIE use of a pop-up containing ALT text available to replace
a graphic hole a reasonable solution for a single anchor buried in the
graphic. For the sighted, a different one for each buried one in a graphic
has some appeal. For a pure-text user choice or user agent, all such should
appear, of whatever size needed (or if there isn't room, then arbitrarily
use wrapping/truncating rules).

>This leads Netscape+Screen_Reader users to ask for links in the
>body of the text which are redundant to buttons where the link
>content is an image.  Whereas Lynx users want link count
>minimized and meaningful ALT text making the image-link usable as
>read via the ALT text.  No wonder WebMasters have trouble
>understanding what people are screaming at them.

Venting the graphic space for buttons should be a user option,
if that space could more effectively be used for the ALT text.
>
>This problem could be fixed by guideline, i.e. if Netscape
>followed a "user control of styling" policy like what I included
>in my ACSS action item response, the image link with
>link-descriptive ALT text would serve all blind users alike.
>
>Still, reflecting on our earlier conversation, I would say that
>the use of ALT text in GUI browsers for sighted users favors
>putting a description of the image in the ALT text, whereas the
>user of ALT text in Lynx and pwWebSpeak favors putting a
>description of the link target in the ALT text.
>
If link target naming were terse and suggestive, that would be ok.
Often it is neither.

>This means that the present definition and use of ALT fails to
>isolate one semantic item consistently across browse modes.  Note
>that both semantic senses are worth supporting.  We should strive
>to refine the format and usage so that each gets a distinguished
>representation within the HTML.  [One of these could indeed
>possibly be TITLE.]
>
>ALT represents an example of an opportunity to clean up the
>accessibility of the Web by improving on the definition of the
>Web media.  We may think that the problem is that the authors are
>populating the standards wrong, but it is the standards that we
>have the greatest leverage over.  We should not neglect a chance
>to make the situation better by what we can change; relying on 
>what others have to do for us should be used sparingly.
>
I would like to have ALT be #REQUIRED. I prefer that its content be
descriptive. 
I would also support the second semantic sense of link target. Since
the HREF="..." material is needed, why cannot the Lynx/pwWebSpeak use
that value? I believe for any user agent, both have value. The sighted
user agent may have access to the URL. I'd encourage the other browsers
to make both available, and distinguished.
>

Regards/Harvey Bingham

Received on Saturday, 28 June 1997 20:55:38 UTC