W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 1999

RE: longdesc media type concern

From: Charles McCathieNevile <charles@w3.org>
Date: Mon, 6 Sep 1999 07:00:53 -0400 (EDT)
To: Gregg Vanderheiden <gv@trace.wisc.edu>
cc: "'Jason White'" <jasonw@ariel.ucs.unimelb.EDU.AU>, "'GL - WAI Guidelines WG (E-mail)'" <w3c-wai-gl@w3.org>, "'WAI Protocols and Formats'" <w3c-wai-pf@w3.org>
Message-ID: <Pine.LNX.4.10.9909060642060.27848-100000@tux.w3.org>
I think we are all clear on the primary goal of londesc - to provide a means
of associating further information, for the (broken) img element.

Having textual information is a primary concern to a person who is blind. But
for a person who is a non-reader, this does not necessarily apply. In the
case of someone who is deaf, text is often the second-least useful thing,
after audio. In the case of a person with a cognitive barrier to reading text
may be the single least useful possibility.

As it happens there is no very good mechanism currently for specifying what
the users particular needs are. The most obvious one would be to use RDF
information (or even PICS) to make assertions about the relationships of
various types of content. The closest we get at the moment is the object
element in html, content negotiation in HTTP, or the system-captions
attribute in SMIL. Te first two rely on being able to prioritise content-type
in a particular order, which doesn't refelct reality.

The last one, while it can be used to initiate content type negotiation based
on two or more variables (for example, system-captions="on" and
system-language="Auslan" could be used to offer a signed video accompanying
track, and further down the switch statement a single variable test of
system-captions="on" could be used to provide a textual description), the
scheme has been shown to be inadequate for providing descriptions of video.

In the case where people are using longdesc (why? support for it is slightly
less than support for the object element as far as I can tell) it is unlikely
that they will link to anything which does not contain text. It would seem to
me that best practice there would be to provide a hypertext description,
which might include links to information in all kinds of media.

In the SMIL context, in particular, where a longdesc element can be
associated with any media type, including a text stream (I believe), it makes
sense to allow any media type to be at the other end of the longdesc.

Charles McCN

On Sun, 5 Sep 1999, Gregg Vanderheiden wrote:
  -What is the purpose of LONGDESC?
  _- Of PRIMARY concern, is the ability to easily and automatically pull a
  text description from the LONGDESC link on the main page without having to
  go through multiple links or search another page to find it.  I'm not seeing
  that in the descriptions I've heard..
Received on Monday, 6 September 1999 07:01:33 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:30 UTC