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

Re: introducing new semantics to HTML

From: Al Gilman <asgilman@access.digex.net>
Date: Tue, 17 Nov 1998 11:59:58 -0500 (EST)
Message-Id: <199811171659.LAA03802@access2.digex.net>
To: w3c-wai-gl@w3.org

to follow up on what Nir Dagan said:

> Al says:
> 
> "Where is the contradiction?  CLASS was put in HTML4.0
> specifically to be an extension mechanism, to support semantics
> not covered in the DTD.  This is just the tip of the iceberg.  In
> XML all semantics is after the fact, except for one very abstract
> inclusion tree and cross-reference via X-pointer."
> 
> Class is an extention mechanism for author defined semantics. 
> 
> Class is not a mechanism that allows the W3C to extend HTML
> in a way that breaks existing documents.
> 
> HTML spec. says authors can define any class for any purpose.

The HTML specification does not distinguish between authors
and stylesheets as the originators of class definitions.

> Since the class attribute exists in an experimental HTML version
> published by the W3C shortly after CSS1 became 
> recommendation, we can safely assume that every class name that 
> can you think of has been used by many people for completely 
> different author defined semantics.

I doubt this.  What is the incidence on the web today of
CLASS=<anything> much less CLASS="nav"?

> I may have used class=nav to indicate a poem.
> 
> If WAI decides that class=nav is defaulted by 
> browsers to {speak:none} it is exactly the same as 
> saying that from now on H1 should be defaulted 
> by browsers to {display:table}, as getting all my poems 
> not to be read is more or less as bad as displaying all 
> my headings as tables.

The proposal to reserve the "nav" value of CLASS for objects
which are navigation aids does not include with it any 
default value of the ACSS 'speak' property.  It is simply
an attempt to get a consistent spelling adopted for this
common page-structure notion.

And your "exactly the same" is only looking at the formalities
which capture a minority of what is really going on.  We should
look at what those structures are useful for, and actually used
for, before seeing if this default is comparable to that default
or not in its advisability.

> Any class name may have pre defined semantics 
> in *existing particular documents*.

Yes, I can see that there would be some actual damage to
incumbent uses of tokens for which we suggest a community 
semantic.

> Authors who believed the W3C, assumed that browsers 
> and users do not and never will set style rules for the author 
> defined classes.

In HTML4 and CSS2 the W3C made it clear that authors must write
documents in the knowledge that their styling can be overridden
by a user stylesheet.  The WAI should be making it clear that
authors should use community-based classes that will be already
understood and handled in user stylesheets.  Not making up new
tokens where something close enough is already recognized by the
community.

Capricious or merely idiosyncratic use of classes is one of the
major failure modes of the HTML+CSS architecture as an accessible
technology.

> On the other hand if we say that the new HTML element 
> NAVBAR may be defaulted to display:none in browsers, 
> it is a problem of the author if he extended W3C 
> recommendations to include this element to describe poems.
> This is because HTML states that authors should not 
> extend the DTD by declaring new elements.
> 
> In this sense if WAI says that browsers should safely assume that 
> class=nav is a navigation bar, and skip them when spoken, 
> WAI recommendations are backward imcompatible with HTML4.0, 
> which allows class=nav to be anything that the author wanted, 
> including a poem.

It makes more sense to deprecate using CLASS="nav" for poems
than to deprecate <I> and <B>, and the W3C has already taken
that step.

If you can show me three different authors who have used 
CLASS="nav" for something radically different from a navigation
aid, we can look for a different token.  Actually, if we had
the muscle we should search the web with a thesaurus in hand
and find out if anyone has already started using a CLASS mark
for any semantic like this and jump on the bandwagon.  Just
formally endorse the best token in use today.

Al
Received on Tuesday, 17 November 1998 11:59:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:46:58 GMT