Mark Doffman gave a briefing of the state of GNOME accessibility towards 3.0. If I could recall the few slides from memory they included:

AT-SPI

This is probably the largest body of work that has been done recently on the GNOME a11y front. AT-SPI currently is the biggest bonobo interface we have, and it’s migration to D-Bus, mostly by Mark and Mike Gorse has been a major undertaking. The first release is due shortly.

gnome-mag

Gnome-mag did not have a maintainer until very recently. Now we do. It must be ported to D-Bus too.

gnome-speech

Text to speech is a hot topic. Mark didn’t go into detail regarding this, but we are dealing with multiple issues. First, of course, is it’s use of CORBA. Second, the fact that it leaves the sound device output up to the synth engine makes results inconsistent with different synthesizers. This is especially noticeable when major sound subsystem changes are occuring, specifically around PulseAudio.
Luke “TheMuso” Yelavich is currently scrambling to get Speech Dispatcher into shape. Speech Dispatcher uses raw sockets and it’s own protocol, it does not use D-Bus, as Mark suggested. It would be really nice if it did, and if it were D-Bus activated, but it isn’t. On a non-accessibility related note, I think we would all benefit from a desktop speech service.

GNOME-Shell

GNOME-Shell and Mutter currently use Clutter for a lot of their graphics, this makes AT-SPI support tricky and non-trivial. This will obviously need to be resolved before GNOME 3.0 since as the name suggests, this will be GNOME’s shell!
I am going to eat ice cream.
Ok, I am back. I forgot what else I had to write, so I might get back to this in another post. Or not.

Early Thursday morning I get on a plane, and start the long and winding journey to Las Palmas de Gran Canaria. Two weeks ago I could not not imagine boarding another plane, luckily I have recovered from my previous travel fatigue, and I am excited to go! This would not have been possible without the generous sponsorship from the GNOME foundation. It has to be the greatest community around. I also look forward to meeting the second greatest community – KDE. It will also be nice to meet new friends and get re-acquainted with old ones.

I am up and traveling more than usual this summer. Which is great, although I really would like to spend as much time as possible up in the North West before it gets cold and dark again.

I am now at UDS Karmic (come say hello!), and it takes a while to explain to folks where I live, and what is up with me. I moved out of my Tel-Aviv apartment last week, and I am couch surfing (thanks Emily!) in Seattle next week, looking for a home. Oh, and once I land in Seattle, I will go right back to the airport and spend the weekend in the Bay Area for a JVP conference.

Thanks to the GUADEC travel committee, I will be attending Gran Canaria! I didn’t really plan to attend this year, but now that I am, I need to take advantage of the time there to the max. I’ll be cheering Ara when she presents the desktop testing project, and I will be unofficially unvieling LDTP2. I also want to catch up with folks on the a11y front, there has been some excellent work lateley: D-Bus AT-SPI is taking shape, UIA, Orca is becoming slick, and MouseTrap is neat. There is also plenty of chllanges ahead, namely WebKit and the audio/speech stack. I’ll also try to make myself as useful as possible to the event organizers.

Recently some brave individuals started working on an upstream automated testing project for GNOME. This is exciting.
I say brave because we had a discussion regarding this a couple of years back, and if I recall correctly we stalled when it came to choosing an automation library. Anyway, it was decided that we will be going ahead with LDTP, big step.
LDTP is a mature library, and being such it had accrued sets of requirements from the different utilizers. The core is threaded, and written in cspi, which is unmaintained today, and uses a raw socket to communicate to the client end.
With Nagappan‘s blessing, I started work on a Python rewrite of the core. The main goal being to simplify the code-base and prepare LDTP for the eventual migration of AT-SPI to D-Bus. We also chose to use higher level protocols for communication: XML-RPC. Also, this new version does away with threading in favor of an event loop that plays nicely with AT-SPI events.
After a week of work, I became fairly pleased with the shape it has taken. I used introspection wherever possible as not to duplicate the API both on the client and on the server. The introspection also helps in generating ooldtp’s API, the object oriented interface to LDTP.
The code could be checked out from my github tree.