The story of how Ian Murdock developed Debian while a student at Purdue and named the distribution for his girlfriend and himself has been told countless times. Many people, too, know that Debian is not only one of the largest distributions, but also the basis for countless other GNU/Linux distributions — including five of the top ten listed on Distrowatch — as well as a model for community based free and open source software (FOSS) projects.
Yet, as Debian celebrated its fifteenth anniversary on August 16th, the venerable distribution finds itself facing problems of scaling from within and competition from newer distributions without. What has the project done right, and what has it done wrong? And what is its future?
To find answers to these questions, I consulted past and present Debian Project Leaders. Their opinions (which are strictly personal, and represent neither Debian’s nor their employers’) paint a complex picture of one of the oldest, largest, and most influential FOSS projects today.
For Steve McIntyre, the present Debian Project Leader, Debian’s most obvious accomplishment is the combination of growth. “We started in 1993 with one lone developer, Ian Murdock,” McIntyre says. “Today, we have over 1,000 registered developers scattered all over the world, with thousands more in the community working on parts of the system: packages, documentation, and translation.”
In the next release, which is due this fall, he estimates that Debian will include 23,000 packages and provide support for ten different architectures, ranging from i386 and AMD64 to the PowerPC or SPARC — figures that no other distribution even comes close to.
Yet, despite Debian’s size and diversity, McIntyre adds, “We can also fairly claim to be one of the most stable and reliable distros, too. We don’t expect our users to ever have to re-install their systems.”
At the same time, McIntyre expresses pride in the project’s ideals, which are expressed in such documents as The Debian Constitution and The Debian Social Contract. “All of our development is open and free, with (of course) full source available for people to work with at every stage of our development. We don’t hide problems — our bug tracking system is open for people to search or browse without requiring logins. [And], as we have grown, many of those ideas have spread to the rest of the Linux community.”
Ian Murdock, the first project leader, expresses similar pride in Debian. However, in contrast to McIntyre, Murdock reserves his praise chiefly for the project’s development model and package management system.
Apart from the Linux kernel itself, “We were one of the first projects to understand that the real power of open source software is to include the community,” he says. “There’s been a lot about Debian that just happened, but that was very deliberate. I was a student, I had limited time, and I knew that I wasn’t going to be able to do all this myself, so we set out specifically to figure out how we could leverage a distributed workforce. I remember that, when the project started, people were saying that farming out all these pieces to developers would never work — but it actually worked out pretty well. I think that Debian’s influence in how open source development is done can’t be overstated.”
Murdock also singles out the package management system. “It was really the first Linux distribution to adopt the package system from Unix,” he says, referring to the now-universal habit in GNU/Linux of dividing software to be installed into separate pieces. At the time, he says, the package system was necessary due to slow Internet connections and the ability to squeeze software on to floppies, but he adds that “it turned out to be a really good metaphor for system management, too.”
The early inclusion of dependency resolution (having packages offer to install other software that they needed to run) was especially influential, and has now spread beyond Debian to the rival .RPM package system as well.
More recently, former project leader Martin Michlmayr emphasizes, Debian has a record of overcoming many longstanding problems.
“I think Debian has come a long way in the last few years,” Michlmayr says. “We have tackled several key problems that have troubled us for ages. From a technical point, we’ve made Debian much easier to install, configure, and use. Processes- and community-wise, we have made a lot of changes. You’ll find that the Debian community is quite pleasant and that, to a large extent, we no longer deserve the bad reputation of a hostile community.”
Talking of the project’s growing tendency to handle responsibilities in teams, Michlmayr continues, “Finally, there have been major improvements to our core teams. Many of our teams have had structural problems for years and showed little activity. But lately we’ve been able to add new contributors to a number of teams, such as the release, security, and FTPmaster teams, and there’s a lot of energy in those teams.”
On its fifteenth anniversary, members of the Debian project can look back at a steady stream of accomplishments. However, many internal problems remain.
“A question of increasing importance,” Michlmayr says, “is how Debian and other projects that mainly rely on volunteers can keep up with projects that have more contributions from companies. There are several areas where we’re trying to catch up.” On the other hand, he adds, “there are some features Debian had sooner than most (such as support in the installer for encrypted root) or where other distros don’t come close, such as support for Network Attached Storage devices.”
More seriously, the question arises of how long Debian can continue to grow. Murdock remembers several occasions when the project seemed to have reached its natural limit, but, at its present size, Debian is already larger than most companies, and the continued ability to grow can no longer be assumed by anybody.
“We work very well already, despite that large size, due to the way in which we can devolve most of the work down to individuals or small teams,” McIntyre says. “As the project continues to grow, however, the communications overhead could become more of an issue. And, as we use more and more resources and spawn more and more teams to support our system, the amount of work needed to coordinate those will of course grow, too.”
In fact, Murdock suggests that Debian is already significantly hampered by its size. “As it’s grown, it’s taken on the characteristics of any large organization,” Murdock says. “There’s bureaucracy, and there’s lots of red tape.” As an example, he points to the often long waiting list for people undergoing the lengthy process of becoming official developers.
More specifically, Murdock says that Debian’s “single biggest weakness” is the fact that, as the project struggles to manage growth and preserve its ideals, it no longer attracts what he calls “the impassioned leader figure,” the benevolent dictator who can make decisions efficiently and keep the project to a definite timetable while keeping the group united.
Murdock suggests that such a leader “is indispensable for any organization, whether it’s a company or an open source project.” However, it is hard to see how the project could keep its preference for openness and consensus on major policy with such a leader. Nor, perhaps, as a relentlessly non-commercial project, does Debian need efficient decision-making to the same degree that a for-profit corporation does.
All the same, Murdock may have a point when he suggests that the benevolent dictatorship of Mark Shuttleworth explains why Ubuntu, which is based on Debian, has surpassed its ancestor in popularity. Regular releases and a commercial arm in the form of Canonical might very well be more reassuring to many software users than the participatory democracy of Debian.
The popularity of Ubuntu, Murdock suggests (as well as, he might have added, the popularity of specialized Debian-derived distributions such as Knoppix and Damn Small Linux) may very well mean that Debian’s role is changing. Instead of being the distribution of choice for many users, the project may be evolving into an upstream supplier for other, more user-focused distributions. The reliability of its packages, as well as the fact that its package format has not fragmented to anything like the extent that the .RPM format, could make Debian well-suited to this role.
Alternatively, perhaps Michlmayr is correct in calling Debian and Ubuntu “complementary.” Those who share Debian’s non-commercial values will continue to be attracted to it, while those who prefer the latest releases and an emphasis on user-friendliness will be more likely to gravitate towards Ubuntu.
Yet even Debian’s reputation as the freest of free distributions is also endangered today by the rise of distributions like GNewSense, another Debian derivative, and Blag. A major rationale for both these newer distributions is that they have removed the proprietary blobs used by some kernel drivers, and are therefore “truly” free in a way that Debian, Ubuntu, and most other major distributions are not.
To keep its reputation, Debian may very well have to deal with these proprietary blobs itself — perhaps by moving them to the non-free repository, where users must make a deliberate effort to install them. But, true to its past form, Debian seems in no hurry to reach a decision.
“I’m not convinced that being the freest in town is necessarily a good goal on its own,” Michlmayr says. “You always have to find a good balance between your philosophy and the practicality of it. Whenever we discuss freedom, someone always asks how the fourth point of our Social Contract — ‘our priorities are our users and free software’ — is to be interpreted. Debian certainly aims to be 100% free, but the Debian and free software communities need much more discussion about what freedom mean when it comes to images and sound files, documentation, certain binary data and other forms of information. In this sense, GNewSense is very valuable because it creates more awareness of the problem.”
Such challenges have already had their effect on Debian. “It’s probably fair to say that Debian’s direct growth has slowed a little over the last few years,” McIntyre admits. But, “we still have a steady stream of new developers joining the project, and, as we continue growing and picking up ever-larger numbers of packages, we’re naturally forming teams to maintain these new packages.”
If these varied opinions mean anything, they seem to indicate that Debian’s role is changing, but will continue to be an influential one. Summarizing his experiences in meeting new users and developers at DebCon, the annual Debian conference, McIntyre expresses an optimism that seems common to many of those involved in the project when he says, “Talking to some of these new folks, I can see that there’s likely to be more than enough life in the project to take us through another 15 years. I wouldn’t miss it for the world.”
Bruce Byfield is a regular contributor to Datamation, where this article first appeared.