Re: Comments/Quibble on "Making Classes Accessible"

At 05:58 PM 2000-12-11 +0000, Sean B. Palmer wrote:
>> Is there any part of w3c space you haven't
>> perused <grin/>?
>"If there is, I don't wanna know about it" :-)
>> We want to give this rapid scan capability to users who are
>> blind.  Hence, it would be useful for the browser to enable
>> the user to choose any class and then step through elements
>> of that class.
>That's a very good idea. I always thought we needed more forms and
>definitions of "class". We can use it for style, transformations, and
>sectioning, and that's a lot for one attribute to handle. It's potentially
>the most semantic attribute in XHTML, and yet it can also be used to infer
>presentation...highly irregular!
>I like the idea of binding classes to ID's, as per Dan Connolly's HyperRDF:
><> which extends
the class mechanism, but a lot
>of browsers don't allow for escaped characters in CSS:-
>     .rel\:class { color: #000000; }
>Which is very frustrating.
>> and the opinion that all XHTML is presentation, and that true
>> content would be XML
>True content would be RDF/XML. I created an example for a fake meeting where
>all of the meeting details are stored in RDF [1] and transformed [2] on the
>fly to produce an XHTML [3] output. The short decription is at
><> I urge you
to take a look!
>     [1]
>     [2]
>     [3]
>The XSLT files are a bit scrappy for now, but they work!
>> what I'm suggesting here is that when authors use classes
>> (e.g. stainless-steel) to add semantics to web pages, let's
>> create a  way so that people with disabilities get the equivalent
>> benefits.
>Any form of semantic equals a benefit to disabled users, because they don't
>miss out on anything :-) That's the whole point of Semantics, everything is
>raw information, and then from that you can convert it into the medium of
>your preference. I could write a stylesheet to transform that meeting
>example of mine into content for Aural Browsers, flashy IE5 types, WAP, or
>> The simplest way is to squeeze it into CSS, but that
>> runs counter to the philosophy that CSS has only "presentation".
>> RDF is another way, and Sean has corrected my initial amateur
>> attempts to do that.  We could also do an XML schema.
>Why would we be writing an XML Schema? Unless you mean we could write an XML
>Schema for XHTML ("don't even go there!"), Schemas are the third layer: 1)
>Data, 2) Metadata, 3) Schemes for metadata (i.e. metadata about metadata).
>Like A.S. says, we are struggling with how best to attack data, and only a
>handful are working on metadata. Schemas are way ahead of us! I challenge
>you to find some RDF that has a non 1 dimensional structure as dictated by a
>Schema. It isn't even possible yet...
>Schemas are great, but only in the right places!
>> Having said all that,  if we want to continue this, the question is
>> what list to put it on?
>Your choice. As the original document was an ERT production, I thought it
>best to direct my comments here. It might move to PF...


Somehow we need to a) address the general W3C language architecture through PF
while still b) working the specfic case of capturing evaluations in ER.

Note: There are some architectural ideas expressed above, such as the idea
semantics does not include presentation, that metadata can be semantically
distinguished from data in the absense of a schema, etc. that bear further
investigation.  Ideally the way to conduct such examination is in the light of
more concrete access-related scenarios (situations and the functions you need
in those situation).

Related point:

With regard to a metadata module, Sean, have you reviewed the SMIL 2 metadata
module?  Is that module something we could use in EDL?  It would seem to be
"RDF in XHTML modularization" precedent that is furthest down the
Recommendation track.


>Kindest Regards,
>Sean B. Palmer
>"Perhaps, but let's not get bogged down in semantics."
>   - Homer J. Simpson, BABF07.

Received on Monday, 11 December 2000 14:05:18 UTC