Portland, Ore. -- The first ever Ubuntu Live conference, running concurrently with the O'Reilly Open Source Convention (OSCON) at the Oregon Convention Center in Portland, ran for three days starting on Sunday, and crammed in as many keynotes, sessions, and tutorials as anyone could possibly want.
The first session I attended on Sunday was by Kees Cook, the Ubuntu security engineer who sends out most (if not all) of the distribution's security notices. Sessions are short -- about 30 minutes -- and they're supposed to include some time for Q&A, so speakers really have to power through a lot of information in a short time, which is exactly what Cook did. This didn't leave time for too much detail or in-depth discussion, but Cook gave a good, and occasionally funny, overview of how attendees could participate in Ubuntu testing by running development versions.
He talked about the different ways users can run testing versions, for instance by running development versions in VMware, using a live CD, or actually running a live system, and also talked briefly about ways to update running systems and how to fix problems when development packages break things.
In all, it was a good session, but a bit rushed -- which seemed to be the rule at Ubuntu Live. Most of the sessions were interesting, but the speakers had to rush to get everything out, leaving little time for much detail.
Unlike typical keynotes with longer talks by a single person, Ubuntu Live is doing triple-play keynotes with three speakers per, with only about 20 minutes each for talk and Q&A. The midday keynote on Sunday featured Eben Moglen, Mitch Kapor, and Jim Zemlin.
It's just plain cruel to ask anyone to follow Moglen as a speaker. Moglen first mentioned that the Software Freedom Law Center "couldn't work without Ubuntu," and then talked about what the GPLv3 means. He said that the GPLv3 process proves that it is possible to make public policy by group deliberation. He also said that the GPLv3 proves that there is "no excuse" for software companies to deliver software using licenses that are "my way or the highway" -- meaning with licenses that have not been devised with user input. Moglen said that users should insist on licenses they helped to create.
He also spent a fair amount of time talking about Microsoft's patent promises and the GPLv3, and says that Microsoft has "turned tail and run" in the face of GPLv3. As always, Moglen's talk was interesting and enjoyable, but far too short.
Kapor, president of the Open Source Applications Foundation and best known as the designer of Lotus 1-2-3, made an interesting analogy in his talk. With all the furor over the release of the seventh and final Harry Potter novel, Kapor says that if the open source movement were the Harry Potter series "we'd be at book three or four" right now.
He noted that less than a generation ago, open source was "a marginal idea," but it is now mainstream. He also reminisced about the fact that at the beginning of the movement, Richard M. Stallman actually picketed him due to the look and feel lawsuits he admits now to "clearly being on the wrong side of." Later, Kapor co-wrote an article with RMS about why patents are bad for software. It was interesting to hear an important figure in the software industry actually changing his opinion about a matter as weighty as software patents. Kapor's talk too was interesting but too short.
Finally, Zemlin, director of the Linux Foundation, talked about the "next stage" for Linux. Zemlin's talk was supplemented (rather heavily) with parodies of the Apple/PC ads. Zemlin talked briefly about some of the initiatives offered by the Linux Foundation, such as the travel fund for developers who want to attend conferences and the Linux Standard Base.
Monday, the schedule boasted another six keynotes, including Ubuntu CTO Matt Zimmerman, Canonical business director Chris Kenyon, MySQL AB CEO Mårten Mickos, Intel's Doug Fisher, Alfresco's Matt Asay, and Tim O'Reilly.
Zimmerman used his keynote to drop a few details about Ubuntu's roadmap, what's coming in the 7.10 (Gutsy Gibbon) release, and the date for the next Long Term Support (LTS) release. According to Zimmerman, the Gutsy release will be the first release to have 3-D effects out of the box, laptop power profiling, and multi-monitor configuration.
The first LTS release, 6.06 (Dapper Drake), came out last year, and many have wondered when (or whether) Ubuntu will release another. Zimmerman says that the first release of 2008, 8.04, will be an LTS release, with support on the desktop until 2011 and on the server until 2013.
On the server, Zimmerman says the Gutsy release should include "turnkey Web administration," additional one-step server recipes, and security protection in the form of AppArmour.
Asay's talk was on the "10 commandments of open source." Asay noted early on that when vendors and proprietary software advocates hammer on free software advocates for being "religious" about free software, what they're really saying is "don't be passionate" about the ideals behind the software. Asay noted that Ubuntu, like any FOSS venture that needs to make money, should be sure to serve the community as well as pursuing profit. Without both, the vendor or project is not as likely to be successful.
Mickos' started off his keynote by polling the audience for users of MySQL, and then asked "who doesn't use MySQL." A few hands were raised, and Mickos replied, "Good. Well, someday you'll need a fast database."
Mickos also touched on the fact that the Internet and open source had created a production force unparalled by any commercial company. He estimated that there may be as many as 30 million developers online, which means that "no matter how many developer you hire, there are 10 times as many outside your company."
He noted that this changes the way that users pay attention, or don't, to talks like his. "All of you, you're online, you're blogging, taking notes, you're barely listening to me ... better than a few years ago, when you would have had to listen to me or sleep." The point, of course, being that the majority of the audience was sitting with a computer, with a wireless connection, capable of being productive or sloughing off even in the middle of a keynote.
According to Mickos, to achieve commercial success with a FOSS product, companies need to serve two groups: Those who spend time to save money, and those who spend money to save time. Companies that go too far in either direction will find themselves without a sufficient customer base or sufficient user base.
The future of processors
But Ubuntu Live wasn't all about keynotes -- despite the large number of them. The schedule also featured a barrage of talks on a wide range of topics.
One of the more interesting talks at the conference was "Exploiting Parallelism with Multicore." Intel's James Reinders talked about the effect that multicore processors would have on programming, and why it was necessary to focus on multicore processors rather than trying to juice up clock speed every few months. While Moore's Law is still in effect, Reinders says that "the free lunch is over" in terms of getting benefits out of bumping up the clock speed, because of three factors: power concerns, hitting the memory wall, and the Instruction Level Parallelism (ILP) wall.
The power concerns are probably obvious: We've reached a point where the additional power consumption of faster processors is not worth the small performance bump we get. The second factor, the memory wall, is due to the fact that while CPUs keep getting faster, memory speed has not kept up, so we end up with processors waiting for something to do because system memory can't keep up. The ILP wall comes from the fact that the performance of single-threaded code has stalled with the current generation of processors. Instead, Reinders says that programmers are going to need to start thinking about how code will scale for multiple CPUs and "think parallel."
Reinders also says that CPU cores aren't going to remain identical. Right now, dual- and quad-core chips contain several identical cores -- but he says that in the future we will have CPUs with multiple cores where some of the cores are smaller and could be used for different functions. In addition, we may see desktop systems using different memory architectures, abandoning consistent memory access for Non-Uniform Memory Access (NUMA). He says that Intel has not officially announced any designs with heterogenous cores, but "it would be absurd if they didn't."
To meet the needs of the new CPUs, Reinders says that programming languages and practices will need to change, and we'll either need new languages or need to extend the languages that we are using now. Reinders suggests extending C++ using templates and generic programming practices, and he talked a bit about what that would look like with a few examples.
After the session, Reinders took questions from the audience. I asked whether Intel planned to assist the GNU Compiler Collection (GCC) folks in improving GCC to produce code optimized for multicore CPUs with heterogenous cores, and how would we be able to bridge the transition period between single-core and multi-core CPUs. Reinders hedged on the GCC part of the question. While Intel has contributed code to GCC and will continue to do so, Reinders indicated that optimizations for Intel's multicore CPUs would probably show up in Intel's proprietary compilers before we see the improvements in GCC. As for the transition between single and multi-core, Reinders didn't have any specific answers, but admitted that "it's never going to be tougher than now to try to produce code for both" types of CPUs. The future looks bright, but the path to the future is going to be a bit rocky.
Making Ubuntu run on laptops
Matthew Garrett, one of Ubuntu's developers who is not employed by Canonical, gave a talk on Tuesday about supporting laptops with Ubuntu. Garrett primarily discussed the reasons why laptops are harder to support than standard PCs. While many key components are more or less standard, Garrett listed the parts that tend to be non-standard, such as wireless chipsets, special function keys, sound codecs, rotating screens, resume/suspend, and hard drive protection, and said they can all give developers fits.
FOSS developers have to work hard to provide laptop support. For instance, Garrett talked about the hard drive protection features that most laptop vendors include now. Garrett says that every vendor on the market implements this differently -- meaning that developers have to figure out methods for each vendor, rather than being able to support one specification as they can do with "standard" hardware such as SATA chipsets.
During the Q&A, I asked Garrett if the next Ubuntu release would include better support for Apple MacBook and MacBook Pro laptops. Garrett's first answer was that users should avoid Apple laptops if possible, but he acknowledged that Gutsy should provide better support for the systems, and that there were only "two or three" bugs left for those systems to enable support out of the box.
Tuesday, Ubuntu Live was devoted to half-day tutorials. I sat in on the Bash and Dash tutorial by Carl Albing and JP Vossen -- co-authors of the Bash Cookbook. Participants spent a lot of time just discussing some basic features of bash that users could look up -- though, as Albing mentioned, users might have a bit of a hard time combing through the bash documentation to find said features. The pace of the tutorial could have been peppier, but the audience was enjoying it. In some ways, I felt like the tutorial was more of a BoF session than a tutorial, which isn't necessarily bad.
It's the community, stupid
One of the recurring themes that came up during Ubuntu Live was community. Asay, Mickos, and other keynoters hammered home the importance of community to Ubuntu's efforts, and on Tuesday I attended several community sessions.
The first session I attended was by Jono Bacon, Ubuntu's community manager. Bacon is always an engaging speaker, and Monday's talk was no exception, though not quite as silly as some of his previous talks. Bacon discussed the importance of the community, saying it was "fundamental to everything we do," and says that his goal is to "make Ubuntu the example of free software community done right."
Bacon talked about Ubuntu's governance structures, community teams that make Ubuntu work, and some of the challenges he's faced in his role as community manager. The hardest thing, says Bacon, is conflict between developers. He characterized the Ubuntu community as being relatively light on conflict, but admitted he'd had to deal with a few conflicts between developers -- though he said the developers "saw reason" in every case but one. However, he failed to offer specifics on how a FOSS project can deal with conflict between developers, or difficult community members. That would have been a worthy topic in and of itself.
I also sat in on Belinda Lopez's talk, "Why do we need an Ubuntu women's project?" According to Lopez, the project is not about separating project members, but about mentoring women interested in Ubuntu, providing short courses, offering general support, and holding membership drives. She also says that the women's project should be viewed somewhat like an "incubator" to help more women become interested and involved in the project.
Overall, "community" seemed to be the keyword for the entire conference. I can't recall any of the talks or sessions I attended that did not at least touch on the importance of community to Ubuntu.
For a first attempt at a conference, Ubuntu Live was well done. There was some room for improvement -- the organizers need to decide whether they're aiming at users, developers, or partners, and the session lengths need to be extended from 30 minutes -- but it was an interesting conference and the time I spent was well invested. Attendees I spoke to also seemed to find the conference to their liking, and looked forward to seeing what Ubuntu Live 2008 will be like.