[Irl-dean] Opinions: HTML Transitional and WCAG Double-A

[Barry McMullin]

On Tue, 9 Jan 2007, Eoin Campbell wrote:

> In my view it is perfectly reasonable to claim even AAA conformance while
> declaring the HTML 4 Transitional DTD in your page DOCTYPE.
> It is often the case that many pages on a website actually conform to
> HTML 4 Strict, but some (for various reasons) do not, and rather than
> individually set the most appropriate DOCTYPE for specific pages,
> the webmaster specifies all pages as Transitional.

Um ... seems a bit strained to me, I have to say.  If the site is
small enough for "manual" maintenance of individual pages, then I
think this could reasonably extend to choosing the right doctype
on a case-by-case basis; and if it is too large for that, then I
would expect it is big enough to justify automated setting of the
correct doctype on a case-by-case basis (using "HTML tidy"
statically, or mod_tidy with Apache, dynamically, or any of a
variety of other tools) ... but, OK, as Eoin suggests, let's park
the "degenerate" case of a page with a transitional doctype, but
which would actually still validate to a strict DTD, and look at
the more interesting case:

> What you may have meant to ask was: if a page _uses_ Transitional
> constructs that are not
> permitted in Strict (e.g. the U element, or the A/@TARGET attribute),
> is it valid to AA?
>
> I think the answer must be: it depends on whether those constructs are
> permitted
> for AA conformance or not.

Yes; and then I'm back to my response to Gez' comment; assuming
such constructs do indeed exist (i.e., constructs which are
permitted in Transitional but not in Strict, and which do not
intrinsically imply violation of any WCAG P1 or P2 checkpoint)
then I'd actually quite like to have a list of these!  Gez
suggested the target attribute; I'm not quite so sure, and
neither apparently is Eoin, though for different reasons:

> For example, Checkpoint 10.1 is Priority 2, and recommends not
> opening new windows, so using A/@target would not be compliant.

On this, I think I agree with Gez - 10.1 allows new window
opening (or, at least, *requesting* new window opening) as long
as the user is somehow "informed" of this in advance. I also
agree with Gez (and probably Eoin) that that still doesn't make
it *desirable* to open new windows in general; but it does mean
that requesting a new window, in itself, is not *necessarily* a
violation of 10.1.

Mind you, I'll note that the clear implication of 10.1 is that
designers should always assume that a user *can* refuse a request
to open a new window; so I would argue that no functionality
should *rely* on opening new windows.  That is, if I found a site
that was unusable simply and only because the user's browser
refuses to open new windows, then I would be pretty tempted to
say the site fails 10.1.  I admit that is a matter of some fine
interpretation! But if you do buy this line, then it is perfectly
legitimate to use javascript (rather than target) to request new
windows, safe in the knowledge that the site is still usable
without javascript; which, in turns, allows you to use HTML
strict rather than transitional. That is, presumably checkpoint
11.12 "Avoid deprecated features of W3C technologies" (P2) then
trumps any desire to use Transitional?

But in any case, other offers of candidate "Transitional-only"
features, that don't intrinsically imply violation of any WCAG P1
or P2 checkpoint, will be greatfully appreciated!

Thanks,

- Barry.