W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 1998

Re: problem with class=dlink

From: Daniel Dardailler <danield@w3.org>
Date: Thu, 04 Jun 1998 14:39:34 +0200
Message-Id: <199806041239.OAA15820@www47.inria.fr>
To: nir dagan <dagan@upf.es>
cc: w3c-wai-gl@w3.org

> There is a major problem with identifying dlinks via the class
> attribute, for use by UAs.
> The classification of elements by classes has a meaning only 
> for author defined stylesheets or scripts. There are no 
> reserved names for classes. UAs should apply style to classes
> (e.g., display:none) only if the author specified it explicitely.
> (technically users can do it too, but how should they 
> know what a certain author defined class means. Thus a sensible 
> user will never write a stylesheet with classes as selectors)

Users will know the meaning of class name the same way they know the
meaning of 'none' in display:none: because W3C/WAI specifies it.

> Assume I write a website about international affairs 
> and have two classes of links "dlink" for domestic links, and 
> "flink" for a foreign link. (I want them to have different colors)
> Then comes some UA and doesn't show my domestic links because it came
> up with its own meaning of "dlink".

Yes, a typical conflict issue with public interfaces, no definitive
solution besides running central registry, which is not going to
happen at that point. One way to help is to add a prefix, ala
> In my view there is no general solution that will show the D-link
> in old browsers and a longdesc in new ones.

There is a way, but I'm not sure we need this complexity: adding a
unique ID for the image and a variant of that for the D anchor.

<IMG SRC=foo.gif ALT=foo ID=img001 LONGDESC=foo.htm>
 <A HREF=foo.htm CLASS=wai-dlink ID=img001-dlink>D</A>

But I think given the fact that most D links will be "close" to their
image, and the fact that they already share the same hyperlink target
(in the LONGDESC and the HREF), that should be enough for a UA to
identify one.
Received on Thursday, 4 June 1998 08:39:23 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:27 UTC