Willie Walker's Weblog

As I mentioned in a previous blog entry, this year's summit was to be held Columbus Day Long Weekend (October 11-13) in Boston, MA. Well, it happened and it was awesome.

GNOME Accessibility community volunteer extraordinaire, Joanie Diggs, and I commuted down from New Hampster each day in my 12 year old, 33MPG, green Corolla wagon, doing our best to conserve energy and be nice to the environment. Plus, it also meant Joanie didn't drive, meaning that the roads were that much safer. So what if I did a three lane dive to make a right turn in the middle of heavy traffic...I used my turn signal after all. In fact, I'm almost positive it blinked more than once.

As a reminder -- this was a summit, not a conference. As such, it was geared towards informal face-to-face discussion and hacking, of which we did plenty. The purposeful reduction of organization was unsettling at first (i.e., step 1: find the MIT building we're meeting in), but this actually helped reduce the pressure of needing to be somewhere at a certain time, allowing us to be somewhat free.

OK, enough blathering. Specific things:

• WebKit: some WebKit folks were there (e.g., Pierre-Luc Beaudoin), and we went over the AT-SPI and what an assisitve technology expects to get from a good implementation. Joanie then built Alp Toker's special accessibility branch of WebKit and went over it with the WebKit folks in accerciser. This was extremely beneficial because it helped the WebKit folks understand the scope of the problem much better: there's still a lot of work to do. In addition, Vincent Untz was also present, so he too got a better understanding of where WebKit is with respect to accessibility. This is one of those things where a face-to-face discussion saved us all lots and lots of time and helped get the WebKit folks going in the right direction. Many many thanks to the WebKit folks for being so receptive and willing to work on accessibility, and many thanks to Joanie for going over WebKit with accerciser.

• AT-SPI/D-Bus: Mark Doffman, Mike Gorse, and I hacked on the D-Bus implementation and actually managed to get Orca to kind-of-sort-of speak with it. It was a very good exercise and helped give me a great understanding of where the implementation stands. From what I can see, the overall object hierarchy is in place and AT-SPI object events are being delivered somewhat reliably. There's still some work to be done on the registry to allow an assistive technology to discover applications and also on the input device event mechanism to allow an assistive technology to intercept and consume input device events. Both Mark and Mike are great guys and I suspect they will have these features worked out soon so that we can get on to debugging and performance analysis. Like the WebKit discussion, being able to sit down face-to-face helped us speed many things along. If we had to do this via the phone, chat, or e-mail, it would have taken a very long time. If you want to try it out (Mark and Mike want you to), check out their web page. Please also give them feedback on the web page, too, to help make it easy for others to experiment with the work.

• XEvIE: while we didn't have a decent discussion about XEvIE, I was able to make a great connection with an awesomely smart 17-year-old sophomore from RPI: Rob Carr. Rob is an ex-Compiz developer who also understands a fair amount about the input side of X. He said he'd be willing to take a look at XEvIE and try to bring it up to date so it doesn't get kicked out of X altogether (or to at least bring it back from the ashes). I'm going to see if we can turn this into a GOPA task -- the work will be valuable and I think Rob should be compensated if he can do the work.

• gnome-session: Ray Strode and I sat down and talked about the various gnome-session and gdm changes that occurred in GNOME 2.24. This was a very valuable thing for me because it helped me understand why we are seeing some quirks and potential ways to work around them. In addition, Ray and I also talked about migrating Orca to use gconf as part of our plans to refactor Orca. We went over a few examples in gnome-panel which kind of paralleled what we might do in Orca.

  We also talked about a few ways to manage the user experience of people selecting the presentation they want from the desktop (e.g., "give me speech", "give me braille", "give me magnification" as opposed to "give me Orca") while allowing Orca to remain a singleton instance. Two viable options include doing this via gconf settings and via D-Bus. When we get further with the Orca refactor, I'll revisit these ideas.

  Ray and I also talked about better integrating Orca into the desktop via an autostart desktop file that would automatically restart Orca if it crashed. This may or may not be quite what we want, but it is still worth investigating further.

  Finally, Ray and I talked about some specific ugliness we were doing in the Orca shell script. I ended up ripping a lot of it out as a result of our conversation, helping fix a couple issues that came up as a result of the new gnome-session 2.24 work. Many thanks to Ray for his time.

• Network Manager Applet: I spoke with Dan Williams, who works on the Network Manager Applet which presents a status icon in the panel. We talked briefly about ideas to make it more accessible, and Dan was very receptive to the idea. I think we need to move forward with trying to make GtkStatusIcon more accessible (I'll ask Li about this when I'm in Beijing). Like everyone at the summit, the interaction with Dan was great. He was receptive, positive, and willing to work.

• Testing: we had a wonderful discussion on testing at the summit. Both Vincent Untz and Brad Taylor were involved, and we talked about ways to move our dream of more global testing forward. One of the biggest goals is to make this a community effort so that many people can contribute and to also make a "make test" target for each application. We discussed Dogtail, LDTP, and StrongWind, with our general consensus seeming to be that StrongWind is the better starting point. The next steps are to try to cheerlead this effort some more in the community and see if we can find someone to drive it. I'll try to do more of that when I visit Beijing later this week. If there's one thing I took away from the summit, though, everyone seems to be in full support of something to take us forward with testing.

• Mono Accessibility: Brad Taylor described the Mono accessibility work being done both for Windows and GTK+. Looks like it should hopefully dovetail nicely into AT-SPI. It was a nice introduction to the work, and many thanks to Brad for showing it off.

• Orca Preferences GUI: yes, the Orca preferences GUI is cluttered and offers a lot of knobs and dials. I sat down with a very nice gentleman from Ubuntu (I'm very sorry that I lost his name) to go over the various things Orca users are looking for and want to configure. He told me he'd try to think about ways to improve the UI to make it more efficient and intuitive for users. This would be very nice to achieve.

• Audio: Janina Sajka and Jonathan Blandford had a good discussion about trying to do something about Linux audio in general. I don't know the details, but I'm hoping Janina will write them up.

• Accessibility Web Presence: Joanie and I set aside a good chunk of Monday to focus on the web presence of GNOME Accessibility. In a nutshell, it's stale, convoluted, and very out of date. We also chatted a bit with Vincent Untz to get a better view of the thoughts and practice around the general way GNOME presents itself to the world. We captured our thoughts and developed what seems to be a pretty good plan.

For some closing thoughts on all of this -- the summit reminded me of what a great community the GNOME community is. Everyone is committed to the effort and believes in the work immensely. In addition, this year I really sensed that we are starting to reach a tipping point with accessibility. Over the past couple decades, I've often felt resistance and resentment from mainstream developers whenever it came to including accessible solutions in the desktop. It was too big, too slow, too whatever. A trend I've seen developing in the past year or so, however, is that mainstream developers are becoming not only more accepting of accessibility, but they are starting to bring it into their daily practice. This is way cool, and I'd like to believe that GNOME and the GNOME accessibility work is the one helping make this tipping point happen.