W3C home > Mailing lists > Public > w3c-wai-er-ig@w3.org > December 2000

Re: Comments on AERT (longish)

From: Al Gilman <asgilman@iamdigex.net>
Date: Sat, 30 Dec 2000 18:48:48 -0500
Message-Id: <200012302350.SAA611810@smtp1.mail.iamworld.net>
To: "Sean B. Palmer" <sean@mysterylights.com>, "Nick Kew" <nick@webthing.com>, <w3c-wai-er-ig@w3.org>
At 09:33 PM 2000-12-30 +0000, Sean B. Palmer wrote:
>Thanks to Nick for pointing this out...
>
>> <http://www.w3.org/WAI/ER/WD-AERT/>http://www.w3.org/WAI/ER/WD-AERT/ [...]
>> Checkpoint 3.2 - Create documents that validate to published formal
>> grammars
>
>[Note that this should include XML Schemas as a means of publishing
>syntactic constraints on a grammar.]

AG:: Is a comment against the quote from WCAG 1.0.  Hence should be raised
as a
potential clarification of WCAG 1.0 with the GL group.  In principle the
guideline should be read broadly to include schemas, but in practice the GL
group may not wish to go there until Schema-aware User Agents are more widely
available.  There are phased roll-in of capability issues, here.

>
>> 3.2.1 [priority 2] Check document for public text identifier
>
>I fully agree with Nick's qualms here: there is no value whatsoever to be
>gained from checking FPIs. AERTs could dereference the "system identifier"
>by all means, but *no* Web system should base *anything* upon FPIs as they
>are not unique (although they are defined as being so), and are nothing of
>value. URIs are unique and form the basic foundations of the WWW; use them
>instead. XHTML Families can have any mixture of FPI but there is no way of
>validating its uniqueness. However, a SysID has only one single unique URI:
>the most important part of the doctypedec because it points to the DTD.
>
>Summary:-
>FPIs don't say anything about a document structure
>FPIs are dangerously undefined
>SysIDs are unique URIs
>SysIDs are dereferancable
>SysIDs are the most important part of a doctypedec
>
>Please, please, please consider changing this part of the WD!
>

AG::


I think we blew it on this one, but not at the level you are addressing.

There is one technique for WCAG 1.0, Checkpoint 3.2 and that is "Validate the
content (in accordance with whatever indication is available as to the rules
the content claims to be following)."  Actually, validation to a pure-guess
DTD
is an appropriate AERT technique, and inserting a DOCTYPE declaration if such
validation succeeds is a corresponding repair."

The rest is implementation details.

Note:  If an FPI is on the w3c validator list, then the WAI chould reasonably
trust that it is used for a specific DTD and the AERT could validate to an
assumed DTD on finding an FPI and no URI, unless there are articulable warning
signs or exceptional cases.  FPIs should not be rejected out of hand if known.

On the other hand, the idea that "the author should identify a recovery path
for the grammar in question" is a content guidelines issue and neither ER nor
AU will likely touch it unless agreed to by GL.  I will say that this appears
to be related to the lack of consensus on the use of URIs in identifying
language profiles (little-t, little-d document type definitions) as evident at
<<http://lists.w3.org/Archives/Public/xml-uri/>http://lists.w3.org/Archives
/Public/xml-uri/>.  This is by Sean's discussion a Web-wide architectural
issue
and will therefore not be put to bed until the W3C Architecture Group
steers us
through a resolution.

Note that there is a current thread in GL dealing with "make your pages
easy to
check for accessiblity" so there is interest there in this issue.

 
<http://lists.w3.org/Archives/Public/w3c-wai-gl/2000OctDec/thread.html#889>h
ttp://lists.w3.org/Archives/Public/w3c-wai-gl/2000OctDec/thread.html#889

PF would like to be aware of all architectural issues that are candidates for
Web Architecture Group action that have a clearly articulable access impact. 
Until the WAI/CG directs differently, please direct issues where access is
injured by gaps or inadequacies in Web architecture to the PF group for
collection and presentation to the Architecture Group of the W3C.

But the technique for WCAG 1.0, Checkpoint 3.2, is obviously "validate the
sucker!"  That doesn't have to be controversial, just carried forward to the
authoring techniques.

Al

>Kindest Regards,
>Sean B. Palmer
><http://infomesh.net/sbp/>http://infomesh.net/sbp/
>"Perhaps, but let's not get bogged down in semantics."
>   - Homer J. Simpson, BABF07.
>  
Received on Saturday, 30 December 2000 18:44:32 GMT

This archive was generated by hypermail 2.2.0 + w3c-0.30 : Thursday, 9 June 2005 12:10:38 GMT