Thoughts on Debian Lenny
We spoke to some people involved in various ways in Debian to hear their take on Lenny and the release process.
Christian Perrier is a leading light of the Debian internationalisation team, as well as being a member of the Debian Installer team and maintaining several packages.
Elizabeth Krumbach is a Debian-using sysadmin and a package maintainer. Miriam Ruiz is a Debian developer who is maintaining or co-maintaining nearly 70 packages, as well as being one of the founders and leaders of the Debian Games Team, a member of Debian Women and about to take over Debian Jr.
Luk Claes is one of the general release managers of Debian and a board member of Software in the Public Interest, a US non-profit organisation that takes care of most of the legal and financial aspects of Debian (and other FLOSS projects).
Linux Format: What are you most excited about in Lenny?
Christian Perrier: Despite all the challenges we faced, we've still been able to release with what seems to be the best compromise between stability and relevance.
Elizabeth Krumbach: As with any release, the new versions of packages are a huge help – I've been dipping into the backports archives more and more often over the last few months with Etch.
I'm also delighted by the expanded package archive, which includes things like Drupal 6, a CMS that our clients have been clamouring for and we were reluctant to support, since it wasn't in Debian.
Miriam Ruiz: Being able to upgrade production machines to newer versions of some packages and get some of their new features, while keeping the Debian stability. It's not so important for people on the desktop, where testing or even unstable are more widely used.
Luk Claes: Personally, I'm most excited about the availability of live images and improved support for networking services and improved support for hardware.
LXF: How did the release process work from your point of view? Frustrations or successes?
CP: Some frustrations came from the fact that the installer release was one of the main blockers, mostly because the maintenance team were very short-staffed. Others were the everlasting debates about freeness of this and that, which aren't good for relations within the project.
EK: It's too long. One package I help maintain had a major release back in September. We had to wait until release to get it into Testing so a broader base of folks could use it, by which time I fear a lot of users had already deployed tarball installs to get the latest version with the new features.
The process also tends to be a bit hard on the community and causes a lot of bad press about the project, with discussions getting heated and personal.
MR: Debian is quite open and transparent about many of its inner processes, so I guess that most of the problems that have arisen in the release process are already known. I think that the Release Team, along with most other Debian teams and members, have done a really good job.
It's hard for different teams within Debian to keep doing their stuff. Packages that ordinarily would be uploaded to Sid need to be uploaded to Experimental instead. We're also stuck with versions released almost a year ago, which is not always a problem, but which can be really suboptimal.
It's mostly a time problem: what wouldn't be too disturbing in a two- or three-month freeze obviously is far worse for an eight-month one. That's one of the consequences of getting bigger. The most important problem I find in pre-release time is that people get too nervous.
If you read Planet Debian or the mailing lists, many people are more aggressive and things cannot be discussed rationally – I eventually left most of Debian's lists due to this. Let's hope that now the release is out people will calm down a bit.
LC: There were some bumps on the road we should avoid in the future, like the late discussions about firmware and Debian Free Software Guidelines issues in general. The decreased interest in the Debian Installer also led to some unneeded delays.
Thanks to the good work of everyone involved during the preparation period and the release weekend, the release itself went very smoothly, I think. But I'm probably too involved to have an objective view!
LXF: Is there anything you would have liked to see in the release that didn't make it?
CP: A few technical achievements, such as the simplification of the keyboard handling in both the installer and the console on the installed system. I've been saddened by seeing some localisation teams struggle to cope with updates.
The saddest was having to disable Estonian in the installer due to a lack of updates, even if I added five new languages for Lenny.
EK: No, I'm pretty happy with this release.
MR: I would have loved to have KDE 4, which is so cute! But I guess the KDE Team, and Ana Guerrero in particular, made the right decision to ship KDE 3 instead, favouring stability and feature-completeness over cuteness, and to provide KDE 4 through backports if needed.
With my own packages and the Games Team's ones, I think we've been able to include in the release all that we wanted. In particular, I'm very glad that the latest version of Gnash was included, which is far better than the one before.
LC: I think it's a bit unfortunate that Debian Edu (targeted at schools) was not able to release together with Debian.
LXF: How do you think Debian compares with, or stands out from, other distros?
CP: Debian is still the reference point, in my opinion, when it comes to 'cleanliness' and long-term stability. I don't really care about the complaints about our release cycles and the fact that the content of Debian stable is "always outdated". I'm concerned by the dispersion of resources brought on by Ubuntu.
Because of the immediate appeal it has, especially for newcomers, I feel that possible contributors are drawn away from Debian. This became particularly clear in the localisation area. I don't think this is a voluntary action by the Ubuntu environment, just a natural consequence – but it does require constant effort from Debian so that we stay in Ubuntu's upstream. I think that this problem is often a cause of our lack of human resources.
EK: I really value the commitment to stability and security. While the long freeze periods and delayed releases can get frustrating to developers and folks who want newer packages in the stable release, I think the trade-off is worth it for many production environments. If you want newer, less thoroughly tested products there are other distros to suit your needs.
MR: Debian is universal in the sense that it targets all different OS environments (including desktop, servers, embedded, clusters, etc); it targets many more architectures than any other Linux distribution; and it's also very concerned (although we're not alone in this) with worldwide applicability. With upgrades, there's a lot of effort put in so that users can switch to newer versions of packages gradually and without too much hassle.
Most of the people I know who use Debian have only had to install their system once, and after that only upgrade it, even over a period of years – which isn't true of all other distros. On the bad side, I know that Debian is not really acknowledged as the cutest distro out there.
Some other distros also devote more resources to a better out-of the-box system installation, and thus are more attractive to some people. I guess Debian is more concerned about upgrades, that you have to do throughout the system's life, than about installs, which you should only do once.
LC: Debian is the distribution of choice for stability and quality. I hope I can help to make Debian the universal distro.
LXF: What are you looking forward to getting stuck into for the next release?
CP: Further improving the localisation framework and the accessibility for collaboration with potential new contributors in that area. I also expect the issue of the console handling things to be finally solved.
MR: I want to release a newer version of GoPlay!, a gamesearching tool based on DebTags, and I will try hard to include my Open Rating project. Of course, one of the most important plans I have is taking leadership of what's currently the Debian Jr Project [a version of Debian aimed at kids]. I have some goals and plans for it that are quite ambitious.
LC: On the top of my agenda is finding an easier way to get transitions done, and having the main discussions taken care of at the beginning of the release cycle, so we can drastically shorten the freeze time. Improving transitions involves improving the software on the builds so packages are ready faster for migration, improving the sofware on release. debian.org so transitions don't get so complicated, and improving the stability of debian-installer so it has less chance of introducing large delays. LXF: Anything else you'd like to say on the subject?
CP: I dream of the day where Debian can become the reference distro for commercial vendors. I'm not sure this day will ever happen. If it comes, derived distros might be the motivation and, in that case, I expect them (mostly Ubuntu, of course) to pull us. I also really want to get my own company to adopt Debian for its desktop users. For this, we need official support from commercial apps vendors as well as some hardware vendors – not just HP!
MR: I'm really glad about the release, and also about things getting calmer in Debian now. As always, there will be harsh discussions, different ideological positions and all that stuff, but somehow we will be solving those issues, as we always have. I'm quite optimistic about that!