RE: Proposal: ARIA-ROLE & CSS definition integration

Andrew -

We can all go 'round the block all afternoon talking about the precise
differences between "Semantic Web", HTML support for "semantics", and so on.
We all have a pretty congruent understanding of the idea, and it is quite
clearly enough to base discussions on, judging from my inbox.

>From my perspective, the dichotomy that I present is false only as far as it
assumes that ARIA is the only way to get semantics back on track. If the
current HTML 5 spec gets adopted as-is, it will be nearly impossible for
anyone (or anything) other than a human user without special needs operating
a standard Web browser to productively consume HTML. Search engines will
flounder on code that requires the parsing, interpretation, and running of
JavaScript to render the screen. Forget about people with special needs, the
tools gag on the garbage code that HTML 5 endorses. I cannot imagine trying
to write an application which collects, collates, or aggregates information
from pages that leverage the full HTML 5 spec.

It is news to me, and probably a great many people on this list that
"semantic Web" (regardless of the capitalization, or the inclusion of the
word "Web") is *not* an explicit goal of this group's efforts. Every time I
mention semantics in anyway, everyone on this list acts the same way that I
do, that it is a major goal and that we should all work towards it. So that
being said, it may be time for us to update the charter.

So, again, we can choose to stick our heads in the sand and pass the buck on
this issue. Or we can own up to the fact that the HTML 5 spec as it
currently stands is a dog, and find ways to improve it inline with the
historic goals like separation of presentation and content.


-----Original Message-----
From: Andrew Sidwell [] 
Sent: Tuesday, June 10, 2008 10:29 AM
To: Justin James
Cc: 'Pat Hayes'; 'Al Gilman'; 'James Craig';; 'W3C
Subject: Re: Proposal: ARIA-ROLE & CSS definition integration

Justin James wrote:
>> Interesting, but nothing at all to do with the SWeb project/goals. You
> still talking > about human/human communication here. The SWeb goal is to
> perform work without human
>> intervention or communication being necessary. Perhaps 'inference web'
> would have been a > better term, but we are stuck with 'semantic web' now.
> This is how I have always view "Semantic Web". A non-special needs user of
> standard Web browser doesn't need HTML to figure out when red, bold text
> means "important" and when it means "this field on this form is required."
> That Web browser application *does* need semantic tags to act differently
> red, bold text that should be "important" text as opposed to indicated a
> required form field; so does a search engine, and so do a lot of other
> "interesting" applications.

> Other than the ARIA spec, which should be rolled 100% into HTML 5, I have
> seen little to believe that we are honestly working toward the Semantic
> goal. In fact, I see an awful lot that is contrary to it. All of these
> RIA-related features need to be *removed* from HTML 5 unless ARIA is a
> mandatory part of HTML 5 (made part of the spec, and the tags are
> mandatory). To do otherwise is to violate one of our stated goals.

The stated goals where?  The HTML WG charter makes one mention of 
"semantics" (little "s"): "[The deliverable is] A language evolved from 
HTML4 for describing the semantics of documents and applications on the 
World Wide Web."

> So, which bullet do we prefer to bite? Removing the "Web 2.0" stuff from
> HTML 5? Or making ARIA a mandatory set of attributes/tags/whatever in HTML
> 5, and (hopefully) finding a way to make a direct connection between CSS
> definitions and ARIA?

Sounds like a false dichotomy to me, especially given that charter 
specifies the deliverable must contain DOM interfaces, form/UI widgets, 
and various other APIs.  Depends what you mean by "mandatory", of course.

Received on Tuesday, 10 June 2008 20:52:23 UTC