Reader's Questions to the Debian Project Leader
Question: How long will Debian support new releases after their first release and how long will this support be continued after its successor has come out in the future?
Anthony: Our security team is currently committed to supporting each release for a year after its successor is released; which is why woody support is just now being discontinued. The testing security team is currently working on making sure our security support infrastructure can be used by two teams concurrently though; which may mean that in the future we can support old releases for as long as there are people willing to work on supporting them.
Steve: Our standard policy at this point is quite simple. For each major version N of Debian, we will support it either until 12 months after version N+1 is released, or until version N+2 has happened. That gives our users a reasonable amount of time in which to test and upgrade to each stable release.
Question: Are there newsworthy and tangible plans to integrate stabilizing Hurd into the Debian distribution? What kind of micro kernel Debian Hurd will be based on: Mach or L4?
Anthony: I'll leave these for Steve too.
Steve: Hurd is not currently a candidate for releasing with Etch, as there is still a lot of work needed in order to make it ready. We're still providing archive space for the Hurd porters, and quite a large proportion of Debian packages have been built on Hurd too. However, there are still some important blockers that need to be fixed before Hurd will be released, and I'd expect these will take quite some be completed this summer - we have a student project under the Google Summer of Code to port the Debian installer to Hurd. In terms of the kernel under Hurd, I understand that things have moved from Mach to L4, but I must admit I've not been following this too closely.
Question: Will your users see a BSD kernel based Debian in a foreseeable future?
Steve: Again, this will depend on how quickly the work is done. Patches have been applied to many packages where needed, but there is still much more work outstanding.
Question: Important competitors of Debian, i.e. Red Hat, have spent a lot of energy to integrate long awaited features like virtualization (Xen) or security technologies (SELinux) recently. Will Debian Etch bring comparable features to its users?
Anthony: SELinux support has recently been approved as a release goal by the release managers, so improved support for that should be available for etch users when it's released; Xen support is also being actively worked on.
Steve: Hopefully we should have a comparable feature set, yes. Many newer features in Linux have enthusiastic Debian developers tracking them and working on integrating them into the system. Both of the examples you mention are major projects which have generated a good deal of interest - we have several DDs working on SELinux policy issues, and I know personally of a project to add Xen support into the installer along with preseeding to make large virtualization setups easier to manage.
Question: Please, now have a look into your crystal ball and tell us: Where will the Debian project be moving to five years ahead? Will there be fewer supported architectures and will Debian break the 50,000 limit of distributed packages? How will Debian improve its capabilities to cope with developer conflicts? Finally, do you think that the DPL will have less power and influence, or will this job vanish completely in five years?
Anthony: We'll certainly have dropped support for some architectures - but whether that takes the form of no longer supporting them at all, providing a different level of support such as only supporting embedded systems for some architectures, or updating an architecture definition to exclude some older systems and better support some newer ones isn't clear. 50,000 packages isn't a particular limit - it seems certain we'll continue supporting new bits of software up to and beyond 50,000 packages. More importantly we'll be developing new technologies to make it easier - or even possible - to find the bits of software that are actually interesting amongst the thousands of programs that just aren't relevant for any given user.
That's a leading question - it could well have much more power, influence and significance in five years! Personally, I hope leadership activities will broaden, with the DPL taking a more active role in leadership, but also other developers throughout the project taking on leadership roles too.
Steve: *grin* Fortune telling has never been my strong point, but I'll try...
Five years from now, we should have had several more stable releases. If current trends continue, we will end up with a huge number of binary packages that will fill a ridiculous number of CDs and DVDs. My own quick calculation suggests that the number of binary packages is growing year-on-year by ~20%. In five years, that puts us somewhere between 40,000 and 50,000... In terms of architectures, I expect we will probably have about the same number as today - some may be added as others are removed.
Developer conflicts are a growing problem as time goes on. We're working on ways to help reduce their impact, and to help people get along better. There is an ongoing effort to define some guidelines, while at the same time a recognised part of the DPL's job is to act as a mediator when disagreements require it.
And that leads nicely on to the last part of the question - I don't see the DPL's job vanishing any time soon. But it won't carry much power - in fact, even now Debian's constitution is quite carefully worded to make sure that the DPL has very little direct power! The places where the project leader can make a difference are more acting as support for others in the project: helping to guide technical discussions to conclusions, acting as a mediator in disagreements when needed, being an external point of contact for Debian. I don't see those jobs going away any time soon...
Question: Finally, we would like to see you reflecting on a question of your own (maybe something our readers might have completely missed). Could you please articulate this question and your answer?
Anthony: It's a bit self-interested, but I'll go with "What ways are there for your readers to help Debian out?" :)
That's probably the most important question in free software, because without people who use free software spontaneously doing things to help it out, there wouldn't be any free software development at all. And it's an interesting question because while it's easy to think the only thing that matters is hardcore hackers who know perl and C churning out code in the dead of night, there are many other activities that are also important but don't come to mind so easily. That includes things like writing documentation for people who don't know the difference between fprintf and sprintf, or finding and analysing bugs so developers have enough information to fix them, or helping promote free software and encouraging other people how to use it so there are more developers and users who can contribute back, or raising awareness amongst politicians or executives who might make laws or policies that make it harder to use or create free software, or just supporting your local free software developers with an encouraging word or by shouting some of your local LUG members a free beer or pizza.
Steve: There's a question that I've been asked by lots of people recently, and I'm surprised wasn't mentioned here: when will Etch be released? Our release managers are aiming for a release by the end of 2006. That's still looking likely - we need to start freezing bits of the distribution quite soon to allow us to stabilise. Then we need to get working on reducing the number of release-critical bugs. It'll be hard work for all of us, but I have confidence in the RMs that they can keep in control and meet the schedule. Oops, I'm predicting the future again! :-)