W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 2001

Re: WCAG 1.0 CP 6.1 Considered Harmful [was: Re: who does what]

From: Sean B. Palmer <sean@mysterylights.com>
Date: Tue, 25 Sep 2001 20:18:16 +0100
Message-ID: <01cf01c145f6$f1051cc0$0bd993c3@y0r1d9>
To: <w3c-wai-gl@w3.org>, "William Loughborough" <love26@gorge.net>
Cc: "Wendy A Chisholm" <wendy@w3.org>
[...]
> > the author MUST have a way to make those additional
> > semantics known.
>
> Let's say for the sake of argument that I am persuaded that
> this is so. Where will that "way" be provided?

In the instance or linked to the instance. There are already techniques
that one can use to make the semantics of classes more well known, it's
just that WCAG/WAI doesn't note them. When I wrote "the author MUST have a
way"... the authors already do have ways available, but perhaps don't know
about them. WCAG techniques aim to make those available. WCAG aims to give
rationale to the idioms.

> [...] Would a checkpoint make it happen?

All checkpoints are pointless, we already know that. Techniques are what
should be investigated, and that requires background. It would be, IMO
useful in any case for GL to investigate this thoroughly. But one of my
side-arguements is that the checkpoints (6.1/4.4) as they are go beyond
being basically pointless, and onto being wrong, and even harmful. That was
exposed from the rationale that GL does not need to investigate class
semantics *because* of those checkpoints.

Don't forget, techniques for WCAG should to just be limited to HTML. I
expect that, whatever the merits of classes, they will be employed in other
languages. Perhaps this comes under the domain of XML GL in that we should
be getting language authors to think about classes, but in general, what if
they go ahead and use classes anyway? Do GL recommend hacking the languages
so that one can add semantics to classes?

HTML is a good case because it is such a popular format. Techniques for
exposing the semantics of classes need not be complex, the first one that
springs to mind is just putting the information in prose somewhere:-

   <p>Markup sections with the class attribute set to
   "example", and rendered visually with a thick grey
   border, are examples of code that one can use in
   applications.</p>

But it would be nice to have a more machine readable method of attaching
the semantics. Perhaps that would require changes to user agents, perhaps
that would require PF and the HTML WG to debate it out, perhaps it would
require other paradigm shifts. I'm just saying that it's important that we
investigate it, and that as an important side-issue the checkpoints that
are preventing us from investigating it are faulty.

[...]
> Let's just scrap all this "normative" delusional and start
> publishing tips/techniques/tricks for providing accessibility

Agreed. But the fact that we should start publishing techniques is a
normative "MUST", is it not? :-)

[...]
> There is no tradition/prospect of providing semantics.

Why should that prevent people from doing so if they want?

> solutions via pronouncement rather than myriad exemplification
> are exercises in Pedantics rather than assistances with Semantics.
> Show and Tell. Over and over.

Agreed.

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://purl.org/net/sbp/> .
Received on Tuesday, 25 September 2001 15:18:33 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:14 GMT