Re: Accessibility: Re: Path to Last Call (was closing various issues)

On Mon, Aug 31, 2009 at 12:48 PM, Jim Jewett<jimjjewett@gmail.com> wrote:
> My admittedly biased perspective is that the programmers have taken
> some first steps, but the "meeting halfway" part was a missed
> connection.

After a bomb like this, I really am obligated to be a bit more
specific, in case someone is suddenly motivated.

ARIA/HTML integration:
---------------------------

HSivonen's draft aria-html integration is at
http://hsivonen.iki.fi/aria-html5-bis/
It isn't the easiest to understand without code, but it is currently
in production use in his validator, and I believe it got the ball
rolling.

Ian Hixie's current draft does include an attempt to integrate ARIA
and html.  I believe it could use much improvement (and I have
submitted a bug report already), but it is concrete enough that it can
be reviewed, and bugs can be fixed.
http://www.whatwg.org/specs/web-apps/current-work/#annotations-for-assistive-technology-products

I'm also aware of Steven Faulkner's draft
http://www.paciellogroup.com/blog/misc/ARIA/html5-elements1.html
but am leery of the status.  The last I saw was
http://lists.w3.org/Archives/Public/public-html/2009Aug/1103.html
where he basically said not to look at it.  Since the even the
contents looked plausible to me, that had the effect of making me
think I should bow out of the discussion and wait for things to be
simplified.

Table Summary auto-generation:
--------------------------------------

[Earlier brainstorming had been led by Simon Pieters, which gave cause
for optimism that at least Opera would be willing to implement, if
there were agreement.]

Request for feedback on summary autogeneration:
http://lists.w3.org/Archives/Public/public-html/2009Jul/0330.html

Stephen Stewart said +1 to the idea, but he spoke as an author, rather
than an accessibility expert, and didn't evaluate the specific
proposals.

Joshue O'Connor said that auto-generated isn't as good as
human-generated, but didn't say where the failures were, or whether
the auto-generated were worth doing when the human did neglect his or
her duty.

Janina Sajka said that good headers aren't enough, because examining
them is a pain.  But that sounds like a User Interface problem --
exactly what autogenerated summaries would solve.  I didn't see any
response to the specific suggested summaries.

Gez Lemon seemed to agree with Janina, but didn't respond about the
specific examples, so I'm still not sure what is sufficient to be
useful.  He also posted a "good example" that was autogenerated by the
authoring tool, but took advantage of knowledge that wouldn't normally
be available except on an intranet.  So it couldn't be copied by
browsers, and it wasn't clear whether that information was needed to
provide something that was worthwhile (as opposed to out-and-out
great).

-jJ

Received on Monday, 31 August 2009 18:07:49 UTC