W3C home > Mailing lists > Public > public-html@w3.org > June 2008

Re: Other aria representations [was: Re: ARIA as stop-gap]

From: Aaron M Leventhal <aleventh@us.ibm.com>
Date: Fri, 6 Jun 2008 08:52:36 +0200
To: "Jim Jewett" <jimjjewett@gmail.com>
Cc: "Leif Halvard Silli" <lhs@malform.no>, "HTML WG" <public-html@w3.org>, public-html-request@w3.org
Message-ID: <OF604C1560.2312DB43-ONC1257460.00259415-C1257460.0025EF3A@us.ibm.com>
Please forget stuffing several ARIA attributes into one. 
Some reasons:
1. An attribute value might have a space or any other char you use as a 
separator. You'd have to start escaping characters That makes this an 
ugly, expensive solution.
2. It's very important for ATs to be able to parse changes to individual 
attributes. That's the main point of ARIA -- it's not just a 
serialization. Parsing individual changes is very difficult to do under 
this system, you will have to compare old string to new string, which will 
not be performant nor would it likely be bug-free.
3. It's not better than the ARIA support we already have, which works 
along a whole tool chain for accessible content now. 

- Aaron



From:
"Jim Jewett" <jimjjewett@gmail.com>
To:
"HTML WG" <public-html@w3.org>
Cc:
"Leif Halvard Silli" <lhs@malform.no>
Date:
06/06/2008 02:09 AM
Subject:
Other aria representations [was: Re: ARIA as stop-gap]




Leif Halvard Silli  wrote:

> OTOH <element aria="prefix-value" > would work
> fine with CSS selectors

There may be several aria attributes on a single element.  They could
be space-separated, but that starts to look like a sub-element that
has to be parsed out using a microsyntax. And you need to worry about
line-wrap bugs.

That might be an acceptable cost, *iff* it were worth re-opening the
aria-integration with other working groups and about-to-ship
implemetors.

> If we really want it to be simple for authors to select
> elements based the element's aria values, then a
> *new* selector is needed.

That would be up to CSS selectors.

It might be worth defining some new pseudo-classes, similar to :lang
-- but I'm not convinced.

I suspect that most people will never use any aria-* properties except
for those defined as part of the standard elements' own definitions.
So they'll just style the elements directly.

I suspect that people who do create custom widgets will usually want
different styling for the custom widget than they want for the
standard elements.  (Or why bother to write a custom version?)  If so,
then you only want to capture explicitly declared aria properties, and
the CSS 2.1 attribute selectors are good enough.  I'm betting that CSS
2.1 bugs will be fixed faster than a new function gets implemented.

For those times when common styling really is desirable, a
comma-separated list of selectors is probably still easier than a new
function.

-jJ
Received on Friday, 6 June 2008 06:55:06 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 9 May 2012 00:16:18 GMT