Re: on Why= with keyword REV= et al

David - Morris (dwm@shell.portal.com)
Tue, 16 May 95 18:31:34 EDT

On Tue, 16 May 1995, Michael J Hannah wrote:

> ++ From: Peter Flynn <pflynn@curia.ucc.ie>
> ++ But I don't see what <why> would do that <note role="reason"> wouldn't.
> I also do not agree with the proposal to make WHY a keyword. I agree
> with Peter that NOTE would be better, but let's keep with the existing
> attribute CLASS. I don't see what role= would do that class= doesn't,
> since the real point of either is presentation guidance.

I guess I completely misread Craig's original proposal for the WHY=
attribute which I read to be a differentiation between giving the
browser some guidance (REL/REV) and giving the browser something to
tell the user in the author's reason for the hotlink (why=). My
fundamental proposal was to move the author reason to be an element
with richer content rules. REL/REV would continue to have rigid
rules as to value so as to enhance their ability to be interpreted
by mechanical processes.

The only reason I would see for a 'new' element name would boil down
to structural differentiation and my admitted novice understanding of
SGML. I believe it would be hard to define the DTD such that
<note role=reason> might have different content rules from a different
role. If this is not an issue then the <note> element would be fine.

As I understand CLASS, it represents an arbitrary author classification
scheme and its values wouldn't have an apriori set of meaning as far
as WWW browsers was concerned. It seems to be heavily entwined with
the notion of using a style sheet so to attempt to use it for another
similar additional purpose will add much more confusion.

On the other hand, a style sheet might be able to provide a value for
an attribute which was keyed from the class which would argue for
keeping the author comment as an attribute. Could a style sheet
provide the whole value for an element which was omitted. That is
if our why/note.reason= element didn't exist withing the content of
and anchor, could the class of the anchor provide a value for the
element?

Dave Morris