RE: "Stupid HTML Tidy [A11y] warnings"

We wrote a bunch of patches for tidy, sent them to the lists and nothing
happened

I would not waist my time on this one.

Lisa

-----Original Message-----
From: w3c-wai-er-ig-request@w3.org [mailto:w3c-wai-er-ig-request@w3.org]
On Behalf Of Bjoern Hoehrmann
Sent: Tuesday, June 08, 2004 4:58 PM
To: chris.ridpath@utoronto.ca
Cc: w3c-wai-er-ig@w3.org
Subject: "Stupid HTML Tidy [A11y] warnings"



Hi Chris,
Hi WAI ER participants,

  HTML Tidy [1] incorporates A11y checks [2] like those in WAI AERT [3]
and OAC [4] and as this seems relevant for a new ERT WG charter I would
like to point you at today's end user feedback on these checks [5]. I am
not sure what to tell the reporter, other than to disable A11y checks
and ignore their presence, [6] summarizes my similar concerns on this
matter.

A number of these comments deal with the integration of A11y checks with
other conformance testing tools such as a Validator. A simple example is
a missing alt attribute for a img element. In a HTML 4.01 document the
alt attribute is mandatory and if the A11y component additionally checks
for it's presence you end up with two messages concerning the same issue
which is rather undesireable.

The same goes for target="_new", even though a DTD-only validator will
not catch such an error, a tool with more sophisticated conformance
checks will spot it, i.e., Tidy will emit an error for the attribute and
then continue to state "new windows require warning (_new)" which is
rather misleading.

There is more overlap. For example, the W3C I18N GEO Task Force has
http://www.w3.org/TR/i18n-html-tech-lang/#ri20030510.102829377 which
states "Always declare the language of the page as a whole in the html
tag", if we would add support for the technical report and check for it,
we would end up to suggest adding a lang/xml:lang attribute for I18N
reasons and again for A11y reasons.

There are also differences in the reliability of the checks. Some have
no reliability at all and all a tool could say is that some things
require manual testing. Which is rather annoying for some users. Testing
whether certain page elements have sufficient color contrast can be done
rather reliably, whether the content of a alt attribute makes sense is
much less reliable.

What I want at some point is a generic site quality tool that integrates
all sorts of conformance testing stuff, checking HTML syntax, embedded
style sheets, broken links, I18N issues, A11y, whether the SVG embedded
in the referenced Atom feeds meets the constraints of the specification,
etc. Such a tool would not be useful if it reports lots of duplicate
issues or confuses clear erros with possibly-maybe warnings. I would
thus like to suggest that a that the possibilty (or ease) of close
integration of ERT WG deliverables in or with other tools is listed as a
success criteria for the WG.

And if anyone has a good suggestion what to do in response to the bug
report other than messing with the A11y checks I would like to hear
those too :-)

[1] http://tidy.sourceforge.net/
[2] http://www.aprompt.ca/Tidy/accessibilitychecks.html
[3] http://www.w3.org/TR/AERT
[4] http://oac.atrc.utoronto.ca/
[5] http://tidy.sf.net/bug/968604
[6] http://sourceforge.net/mailarchive/message.php?msg_id=7539791

regards.

Received on Wednesday, 9 June 2004 02:48:36 UTC