Re: ACTION re: HTML 3: Too many tags!

Ian Graham (
Fri, 28 Jul 95 12:07:57 EDT

Murray Altheim <> wrote:
> I will preface this by saying that my suggested alternative is only that,
> an alternative. It would enable a method for stylesheets to be based
> entirely on semantic markup using a single attributed tag, leaving
> physical markup to the document curator. Although in retrospect I think
> it may still be a good idea to allow users to make their BOLD text
> bright fuschia if they really want to...
> > Ian Graham <> wrote:
> >> (Murray Altheim) wrote:
> [discussion of handling physical and semantic markup...]
> >> I would turn this entirely on its head, given the current discussion. If
> >> most source text comes from legacy documents, then physical markup can
> >> continue to be created by conversion routines.
> >>
> >> I would prefer instead that ALL logical/semantic/informational (depending
> >> on your language) markup be a single element, with attributes providing the
> >> semantic information (eg., VAR,DFN,EM,STRONG,ABBREV, etc.).
> >>
> >> This would allow for:
> >>
> >> 1. all sorts of subclassing (as per Benjamin, Paul et al's
> >> discussion on the subject) based on content rather than appearance
> >> 2. more consistent markup: all physical markup would be literally
> >> implemented (bold tags for bold text), while semantic information
> >> would be covered by one element.
> >> 3. allow informational formatting to proceed according to stylesheets,
> >> leaving physical markup solely to the browser as per the original
> >> source document's display characteristics, unless overridden.
> [...]
> >
> >Questions:
> > Do attributes necessarily make composite semantic markup easier
> > to understand? For example, consider:
> > <SEM CLASS="author strong">Antoine St. Exupery</SEM>
> > Does this mean that the strongly emphasized string is an
> > author's name, or should it mean that the string is an
> > author's name, who should be strongly emphasized? In this
> > case, ordering of the attributes seems important.
> >
> > On the other hand, the string:
> > <AU><STRONG>Antoine St. Exupery</STRONG></AU>
> > would mean "the strongly empasized string is an author",
> > while:
> > <STRONG><AU>Antoine St. Exupery</AU></STRONG>
> > should mean "the author is strongly emphasized".
> > These different meanings are , to my mind, more clearly
> > expressed through tag placement than through attribute
> > values.
> Ian,
> I've looked over your two examples and I really doubt we could *all* come to a
> consensus on their extended meaning. If we were to break down "the strongly
> emphasized string is an author" and "the author is strongly emphasized", it
> would still be difficult to assess the alternate meanings. And given that
> the final product of the two variants is theoretically the same in display,
> it may be moot; but I do see your point. It just seems that you could use
> the attribute ordering as easily as the tag placement, in fact it would be
> much simpler to modify if necessary: <SEM CLASS="strong author"> to <SEM
> CLASS="author strong">.

If things can be made to work that way, then I agree. However, with
REL/REV attribute values the consensus was the opposite -- that the
order of multiple values was not meaningful. So I would be concerned
about having order matter in some places, and not others.

[ I also couldn't recall if SGML allows for ordering in name tokens --
it probably does, but I loaned out my SGML bible..... ]

As for how we could define the meaning of the ordering.... you well
pointed out that my examples could be interpreted in several different
ways... I am unsure if in fact there would be a way to define an
unambiguous meaning to ordering.

> I do think <SEM> is about as good as anything, although I like Joe's
> suggestion of <EM> a bit better, as IMO "semantic" is a much more elusive
> concept than "emphasize", at least for those not as clever as all of us.
> The question is, do we want existing browsers to <EM
> class="foo">emphasize</EM> all semantically tagged 3.0 text, or ignore it
> by creating a new element (ie., <SEM>)? I think that having all instances
> of AUTHOR, ACRONYM, etc. hilighted would make things excessively difficult
> to read. Many of these elements would be used by search engines et al and
> not necessarily for presentation changes.

I too would prefer a new tag, so that everthing does not end up italicized!

> Murray