This is a Q&A session as usual when Linus appears on stage.
We’ve used the process with merge windows now for about 7 years, and it’s working pretty well now. Right now, 2.12 is about 1-1.5 weeks away.
Q: Aren’t we focusing a bit too much on speed? Shouldn’t we take a step back and focus on stability for a bit? A: Linus is very happy that we don’t (have to) worry about features: we don’t try to cram in new things that aren’t ready, but we also don’t stop adding features for stability. This avoids building up a backlog of unmerged things. It also makes sure that people don’t hurry their code: if it doesn’t make it in in 3.12, you just have to wait less than 3 months before 3.13.
Q: So what about not opening the merge windows for about a week after a release, to give things time to settle down? A: Linus has done this a couple of times, but it doesn’t work very well. Developers tend to be focused on what they’re doing and won’t be distracted into this stability thing.
Q: What is needed to be a good maintainer? A: Being reliable. Being responsible and responsive. That doesn’t necessarily mean answering each and every question. Part of the work is also to say no (and that can be impolite).
Q: What keeps you up at night? A: Nothing much, and certainly never anything technical because that can be fixed. Sometimes their are community issues (i.e. people not getting alone) that cause stress.
What Linus wishes for, is that companies learn to work better with the community (though we’re not always easy to work with). He doesn’t wish for any new features, since Linux has done everything he needs for 20 years now.
Q: How do you educate companies to push things upstream instead of keeping things secret? A: We don’t have to convince everybody to do the right thing. We do this because it’s fun and because it works. If companies don’t work with the community, they’ll end up wasting resources so they’ll darwinistically disappear. Dirk confirms that he sees other companies mimicking Intel’s behaviour.
Q: What are your hopes for the Linux ecosystem in about 5-10 years? A: Linus started with Linux because he wanted Unix on the desktop… He’s disappointed that the Linux desktop is a morass of infighting and people doing their own thing. He hopes that the desktop people will collaborate more on the technology rather than making the login screen look really nice. But overall he has a good feeling about the future of open source in general.
Q: Is there going to be a Linux conference in China? A: About five years ago, we were thinking about how to reach people in Japan, because the community was dominated by American and Western-European developers. Now we’re already much stronger in Japan and Korea – there’s an event hosted in both countries by the Linux Foundation. There are also events popping up in other SE-Asia countries.
Q: Apple just announced free upgrades, what does Linus think about that? A: “free” doesn’t mean anything, it’s still not open source.
Linus doesn’t see anything new and exciting happening anymore in the core of the kernel. It happens in the periphery (drivers, arch, net), and in userspace. The excitement for Linus has always been doing things well, not so much about exciting new features. There have been many occasions where Linux has created great new interfaces, but it turns out that they’re hardly used or not in a good way. The kernel’s job is to provide base support for the real work, which is done outside of the kernel. That’s also why stable ABI is so important.
Linus loves the Steam announcement. That may be the tipping point for the Linux desktop. It will not just create traction with the GPU hardware guys, but it also gives distribution the hint that this is the way to go. Standardization should be done this way: not in a smoky room pushing out a document, but rather making things that work and others follow.
Q: Is there a chance that the license will change? A: No. Linus loves the GPLv2, the quid quo pro appeals to Linus. But of course it’s also practically impossible because it is impossible to find back all the copyright owners. Even if you can find all of them, it will be hard to convince each and all of them to change the license in the same way.
Linus avoided being paid for doing Linux for 10 years because he didn’t want to be seen as biased. Now he’s at Linux Foundation, which is seen as a neutral entity by everyone.
Q: Is there anything that we can do to get more women in the community? A: It’s clearly something we want to solve. It’s partly cultural, which is something we can’t solve. Linus is not pessimistic, cfr. 10 years ago people were worried that we don’t have many Japanese contributors but this has improved quite a lot already. Dirk adds that we’re already doing good things, and the fact that we’re talking about it is already a good step.
Q: What needs to happen for Linus to retire and how will he able to tell the difference? A: It would need to stop being interesting. What Linus does has changed a lot over the years, but it’s still not boring. Also when he gets the feeling that he can’t cope. He’s sure they’ll be able to find something else. The kernel is unique in having 1000s of people involved, and many of them for 10-20 years.
Q: What is the plan? A: It’s evolution: what works is what survives. Linus doesn’t know what direction it is going in, but he knows that it will improve.