Re: Identifying D-links with "rel"

There is a strong dependency between what we say in GL (and AU of
course) and what we say in UA.

For GL, we have several options:

 0- "use LONGDESC" and "use (loosely defined) dlink"
 1- "use LONGDESC"
 2- "use LONGDESC" and "use <A rel=dlink>D</A>"
 3- "use LONGDESC" and "use <A class=waidlink>D</A>"

0 is current GL.

We can assume that if we go with 1, people who really want to provide
dlink will include them anyway, with no rel or class: markup free
style, which is not much different than 0. So to me 0 == 1 really.

If GL goes with 0 or 1, there is nothing we can say in UA about dlink.

Nir convinced me that authors who wants to hide their dlinks can use
any class they want (e.g. class=hide).

As Jon reminded us, there is a potential use for a better speficied
dlink for OBJECT, where we don't have a LONGDESC.

e.g.

 <OBJECT data=foo.gif>
   This is a short desc for a foo (<A href=foo.htm rel=dlink>long
   description</A> available) 
 </OBJECT>

This way the UI can use the same shortcut to access IMG LONGDESC and
these (otherwise guessed upon) dlinks.

So for reason of consistency, I favor option 2 for GL and a mention in
UA that UA should make their best effort to not present twice the
access to LONGDESC and rel=dlink and to use the same user interface to
accessing the OBJECT rel=dlink and the IMG long description.