> Nathan Torkington@vuw.ac (303)
>
> > I'd like to be able to go
> > <OL FIRST=i>
> > and have the label of my first <LI> come out as (i)
>
> But Nat, that's presentation, and therefore up to the browser, not
> the author.
No its not, as someone (Terry Allen?) pointed out. It is absolutely
essential that some form of common reference be able to be used when
needed.
I read the article Bill cited, and I think that it is dealing with a
different situation than HTML is. The article only talks of
situations where the style sheet controlling the output can be tweaked
by humans, to be read by humans. In effect, all the descriptive
mark-up does in these situations is separate the semantics (the DTD)
from the presentation (the style sheet), and the authors fool with the
style-sheet until they get acceptable output.
This isn't what happens with HTML --- authors are stuck with what the
various browsers decide is an acceptable style sheet. Because there
is no ``standard style sheet'' to give authors an idea what the output
will look like, presentation becomes almost random. Now the
separation of semantics and presentation has taken control of
presentation away from authors who work in HTML.
The situations are not the same. For this reason, this article can't
be cited as a reason why presentation should not be introduced into
HTML.
However, I have come to agree with Bill that presentation shouldn't be
incorporated into the descriptive markup --- for precisely the reasons
mentioned in the article: ease of authoring, ease of maintenance,
standardisation and alternate views. However, I still maintain that
authors should be able to have some of the same control over
presentation that they have always had.
It seems to me that either
-- we move to make the markup language exceptionally precise
semantically --- that is, be able to tag every sense of emphasis
and every sense of cross-reference, leaving presentation entirely
up to the browser,
-- we ship semantic and presentation information separately, or
-- we create for ourselves a compromise language that encodes vague
semantics, and provides presentation hints as well.
I believe that O'Reilly had some ideas for style sheets as part of
HTML a while ago --- could someone from that august organisation mail
me a copy of their proposal?
Nat.