Here’s a quick rundown of what I learned at the Firefox 2008 summit in Whistler.
- Roughly 400 participants. As far as I know, almost everyone got out alive.
- Mozilla has more employees and community contributors than ever. Many newcomers from the past year who are now getting up to speed. There are more people working on the core and front end than I ever remember.
- Migrating most of the community to Mozilla 2 work hasn’t happened. The community is still focusing on Gecko 1.9, and still has plenty of opportunity to make improvements to the Gecko 1.9 codebase.
- Current schedule estimates (which of course will slip as usual) are Firefox 3.1 due in December, and Firefox 3.2 in Q2 2008
- An auto upgrade for Firefox 2 users may happen in Firefox 3.02 timeframe, but there is no commitment to any schedule afaik.
- Biggest emphasis is now mobile (Fennec). If anything, that was the theme of the summit. There’s another open source Mozilla project called Maemo for small devices, which seems to be Nokia’s solution. Not sure why there are 2 projects other than the fact that Fennec includes XUL whereas Maemo does not. It seems obvious that both projects can and will benefit from each other.
- Continuing work on using automated code transformations. I’ve always been extremely skeptical about this approach, but, listening to those working on it gives me hope. These are clearly smart people, and have me believing that the code transformations are provably accurate. It’s still unclear how much benefit will be soon from the transformations.
- <video> support has made it into current trunk builds, using an Ogg backend. There is no captioning support yet, but the developers definitely want to do it and are helping define the issues (which are much more complex than I first realized). The community is starting to work on a description of the technical problems we need to solve for captioning.
- Mozilla has made <canvas> now work in IE. Authors just need to add a single line of script. So all major browsers can use <canvas> now. There is a similar project called ScreamingMonkey to bring a treal JS engine to IE. It’s pretty clear that a lot can be done outside of Microsoft to update IE, but Mozilla would rather have people switch browsers than spend a lot of resources on that.
- Somewhat controversial: content creation and developer tools were discussed quite a bit. Many believe it’s crucial to have a much better authoring story in order to line up better against Silverlight and Flash.. Some managers seemed concerned about the potential distraction of such a large area that Mozilla does not already understand well, but agreed that the discussion was important.
- At the SVG session there was a discussion about the significant gaps remaining in the SVG implementation. SVG will need more resources, and there was some discussion about Mozilla Corp. diving into it more. There’s a chicken and egg problem with author adoption & IE support, etc.
- Upcoming feature lists are still vague. Most front end developers are just fixing bugs right now.
- Most new ideas start as add-ons or in Mozilla labs. Currently there are over 5900 extensions and 700 themes for Firefox 3 on addons.mozilla.org alone. Many more exist around the web but not on the official site. Those are big numbers, and there is little information about the a11y of the extensions. That’s a problematic area for a11y, and yet full of potentially compelling features for users with disabilities. There have been discussions on the approach for that, such as at least running Mark Finkle’s automated XUL checker through current and future extensions, and having humans review the top extensions for a11y. Mozilla probably doesn’t want any new requirements for extensions to be included in addons.mozilla.org, because it’s already impossible to keep up, but I wish at least keyboard navigability could be tested as part of the review process.
- Mozilla labs project Weave is about the ability for users to store and control permissions for personal data that can be shared across many websites. This relates to personalization but can go beyond into actual content such as calendars etc. (apologies if I am mixing up several projects in one here).
- Mozilla labs project Ubiquity is about the ability to type (or at some point speak) natural language commands to Mozilla. The model allows extensions or websites to define new commands. There are potential security issues here, which the developers tell me are being taken seriously.
- Mozilla Messaging has something like 6 full time developers now. They’re not sure what will make it (or rather not make it) into the next release of Thunderbird. The renewed energy around Thunderbird is pretty obvious, so it would appear that splitting it from Mozilla Coproation off was the right decision for everyone. There are also apparently a lot of other things going on this space, such as messaging inside Firefox and Spicebird. The Eudora port using Thunderbird code is apparently still ongoing as well. Finally, no one could tell me what mscott is working on, but it’s likely also related to this area. David Bienvenu is back working on Thunderbird again, IIRC. Bottom line is, I’ve now forgotten about the old “Thunderbird is in Trouble” headlines from last October. After talking with guys like Mark Banner, it’s clear there’s a lot happening.
- Open Office is looking at using Sunbird as the calendaring and contact solution. They’ve even looked at changing over to XUL for their UI toolkit, although this sounded somewhat tricky, technically.
- Mark Surman will very possibly become the new executive director for Mozilla. We had good discussion about moving a11y further. Frank Hecker will also still be involved, which is great in that he’s been a big supporter of a11y in Mozilla.
- Mozilla Corporation is working to hire another full time a11y developer
- Discussions underway about VoiceOver support. Mozilla’s contact at Apple is Mark Malone and they have a good relationship. They have been successfully necessary getting fixes for Mozilla into the OS X platform through him. They are suggesting that the a11y team try to work with him. The developer who coded Mozilla’s current VoiceOver support has had some more recent experiences with unfulfilled promises from Apple (in terms of dealing with bugs that affect us in a11y — VoiceOver effectively hangs with Mozilla). So, there are still strategy decisions for Mozilla to work out there.
- During the show, Google announced some WAI-ARIA support for GMail. Marco Zehe looked into it with JAWS 9 and was very impressed by the progress. Could GMail become a killer app for WAI-ARIA?
- Also during the show, Todd Kloots from Yahoo! posted an excellent tutorial on how the YUI tab control is enhanced with WAI-ARIA. With this level of detail, YUI is bound to achieve excellent accessibility.
- The director of Mozilla platform development Damon Sicore has agreed to look into the longstanding bugs related to keyboard navigation and plugins. Even, he has some ideas of who could work on them.
- Mobile: the a11y strategy needs to be determined. Most mobile phones for which screen readers are available run on the Symbian Series 60 platform, which is not supported by Fennec yet.
- Windows Mobile, although currently accessible via two screen readers, is supported by Fennec, but MSAA/IA2 is not available on the platform. The community will need to look into more how we can interface with those screen readers.
- Linux accessibility on the mobile platform is waiting for Codethink’s CORBA to DBus migration project to be completed.
- Interestingly, Levelstar has a technology which makes Mozilla accessible on their mobile Linux device called the Icon. It’s probably a proprietary XPCOM component that talks directly to their screen reader, and probably supports only HTML (not XUL or WAI-ARIA).
- The summit was an overwhelming success. I enjoyed seeing both old timers and all the young blood. I’m so concerned about the nitty gritty details anymore, but according to Zak Greant this is a well-known natural progresion.
- It was great meeting Alexander Surkov (surkov on IRC), whom I’ve been working with for 2 years but never met. Surkov came along when I really need someone to help push in mozilla/accessible, and we did a ton of work in Firefox 3. On IRC I saw him as more of a serious bug hunter and as someone who might drift away when he finishes his Ph.D. in mathematics. I learned that we have a lot in common. He laughs quite a lot, is as forgetful as me, and we both really enjoy Mozilla work.
- I had a couple of nice chats with Mitchell Baker and other leaders in Mozilla, who all made me feel very good about the a11y work we’ve done so far in Mozilla
- There are a lot of German speaking people in the community! Believe it or not, they are not as well organized as you’d imagine … so what kind of German engineers did we get in Mozilla?
- Since I moved to Germany in April I was interested in who is working on Mozilla near me. So, I started a list of Mozilla contributors in Germany, Austria and Switzerland.
A humble suggestion
- The list of Germanic contributors is a good start, but Mozilla could really use a wordwide online directory of contributors (with pictures). It was really tough to find people I wanted to talk to at the Summit. Not to mention, if I’m on a trip, it might be good to ping people and let them know I’m around (or vice versa), in case they’d like to meet for a beer. We don’t always have to have huge meetups where you can’t have time for everyone. Wouldn’t it be great if we could see where contributors are located on a map?
Hi Aaron, great overview. Just one clarification- Maemo is Nokia’s Linux distribution for the N7xx/N8xx series of devices. The browser on Maemo Linux is MicroB, an early 1.9 alpha of Gecko.
Thanks Gen, that makes sense. What are MicroB and Fennec ultimately going to be differentiated by? If I understand correctly, it’s that Fennec comes with both XUL support and a browser UI driven by XUL. MicroB does not need XUL because its UI is based GTK, correct?
Other people (timeless?) are much more familiar with the details of the differences between MicroB and Fennec, but I’m pretty sure that MicroB took out a bunch of support for some key technologies. Plus, they used I think the 1.9alpha2, so MicroB does not have all of the improvements that were made to 1.9 since that time (I assume they put some in but not all.) How they will differentiate in the future is something I think Nokia has to answer.
Ari Jaaksi, who runs the Maemo team at Nokia, has blogged about his desire to bring MicroB back to the mozilla-central trunk. I’ve transcribed some of Jaaksi’s comments from OSiM USA 2008 on my blog:
Question from the audience (@ 16:20): “Excuse me, another question. If I remember correctly, it was 3 years ago when you [Nokia] implemented Webkit in to the Series 60 devices, you had to make a lot changes, for example in memory management. Did you use the ‘upstream model’ in that case? I mean, did you feed back to the community the changes you had made for your devices?”
Answer from Ari Jaaksi: “Not the way we [Nokia] should have done it. Let me be very honest about that. Also with our Internet tablets we have horror stories where we didn’t do it [share patches back with the trunk]. Just today, or yesterday I discussed this with the Mozilla guy, the name escapes me at the moment [Jaaksi is referencing Jay Sullivan, Mozilla’s VP of mobile], I don’t know if he is here today, about our Mozilla browser here. It is really that, what we did was last summer when we started to ship with the Mozilla browser we made a couple of mistakes. We are kind of working upstream there [with Mozilla] but we are not doing as much as I would like to do and we sort of need to go back. We almost forked the code [from Mozilla] but we need to go back [to sync up with the main Gecko 1.9 trunk].”
Jaaksi also talks about these same goals in a broader brush on his own blog:
Once upon a time there was an effort to make a Mozilla contributors directory via some sort of mash-up site; I’m not sure what happened to it. I thought the mash-up site’s UI was pretty terrible, and I couldn’t figure out any way to create useful subsets of the data (i.e., people who worked on certain things), rather than just geographic/map-based relationships. Possibly good as a travel guide, but not so useful as a more comprehensive directory.
Ah, it was Frappr; here’s the post from Gerv about it: http://weblogs.mozillazine.org/gerv/archives/2005/12/put_yourself_on_frappr_1.html
Fennec is the browser with a XUL-based UI (extensions!), like Firefox, and MicroB just embeds Gecko (with XUL support turned off at build time).
Great summary thanks thanks Aaron
[…] que pudimos viajar. Mientras tanto, puede leerse un resumen interesante (en inglés) en este blog. Pueden verse fotos del encuentro buscando la etiqueta moz08 en flickr (servicio social de […]