The Debian leader has never been one to flaunt himself (only men have led the project so far). Zacchiroli has many different hats to wear as he shepherds an unruly group of more than 1000 developers. He is an accessible person and easy to deal with, his wisdom belying his youth.
An assistant professor at Paris Diderot University, he was kind enough to take time off his busy schedule of teaching and research to answer some queries from iTWire.
iTWire: An entity, be it a project, organisation or even a country, generally does not reward any leader with anything more than two terms in office. How confident were you of winning a third term as leader?
Stefano Zacchiroli: Country-like organisations rarely allow for more than two terms in a row indeed. But those kinds of terms have lengths in the five-year range, whereas DPL terms are one-year long. As the DPL is a volunteer position, like any other position in Debian, terms are that short to avoid asking candidates a time commitment that would sound scary to most. One-year terms also allow an easy way out if, say, an incumbent DPL burns out or get struck by real life circumstances.
I believe I've done a decent job in my two years as DPL and during this year's campaign I felt several Debian members were ready to acknowledge that. But at the same time it is natural to wish for a change and see how things could go with a different DPL "style". All in all I didn't indulge much in election forecasts myself, because both election outcomes would have had positive effects. Had I lost, I would've had a lot more free time to go back to hacking!
During your first two years as leader, Debian has become a member of the Open Source Initiative and also outlined a policy on patents. What are the ramifications of these two initiatives?
The two initiatives are quite different, but both fall in the realm of Free Software "politics". Debian is indeed not only a distro, but also a renewed "political" actor of the Free Software world. Many people care about our judgement on the free-ness of licences and more generally look at what Debian does on a whole lot of legal matters.
Regarding (software) patents, there is little disagreement that they pose a threat to Free Software. But there is also a lot of FUD surrounding them: for instance, we fear patents that are likely to be invalid just because their owners say so. It's like trusting the bartender saying their beer is the best in town. Patent FUD is bad because it increases the negative effects of patents on Free Software, pursuing the interests of patent-aggressive corporations and patent trolls.
With Debian, we are quite lucky in having access to legal resources such as the advice of Software Freedom Law Center and we try to use that access to benefit others who don't have it. In that spirit we approached SFLC and worked together to publish informative material meant to educate Free Software hackers, and in particular community distributions on actual patent risks. Good examples of our results are both the policy you mention and the Community Distribution Patent Policy FAQ.
Joining OSI has been a more general political matter. It descends from the realisation that while Debian has an important role in Free Software politics, our main vocation is to be a distro. We are not structured to do activities that are needed to fight political battles such as lobbying, petitioning, and the like. There are many organisations that do a good job at that, but not that many that allow other projects to affiliate. Considering that OSI is undergoing restructuring in that direction, and the past common history of Debian and OSI, it did make a lot of sense to step in and contribute to renewing the organisation. This way we can show Debian's support to relevant political battles for Free Software, including software patent abolition, and yet let the organisations which are suited to do so fight them.
Despite the fact that it has a reputation for being a major site of flame-fests, and also has a steep learning curve, the Linux kernel project still has no difficulty in attracting new contributors. How is Debian going on this front and have you contributed in any way towards making it easier for new people to join?
We are doing pretty well in terms of new developers. We got an average net increase of about 30 new Debian Developers (DDs) per year for the past two years. That is after discounting the turnover of people who leave (which is an entirely normal phenomenon in a project that is nearing its 20th birthday). And we have twice as much new Debian Maintainers (DMs) per year. Debian Maintainers are contributors who are not (yet) interested in becoming full DDs, but rather in maintaining specific Debian packages. Experience tells that most of them eventually decide to make the jump and become full DDs.
The DM role is not a recent innovation, but it got used a lot more recently and it has made way easier for people to join Debian.
Similarly, the process to join Debian has in recent years got based more on "track records" than Q&A. That is, we insist a lot on contributing to Debian *first* (with bug reports, patches, sponsored uploads, etc). Then, when applicants have a proven track record of good contributions, the membership process gets very expedited and boils down essentially to a few mail exchanges to ensure we're on the same page regarding Debian principles and objectives.
Debian seems (to me at least) to be something like the Free Software Foundation - responsible for a lot of the good aspects of FOSS but very rarely acknowledged. Over the last two years, I haven't seen any major moves towards garnering more media coverage for the project. Do you have any plans on this front? Or do you think it is unnecessary to solicit coverage? (This is a point I raised in 2010, during my first interview with you).
Well, our press/media activities have increased substantially during the period you mention. For instance, our press releases have doubled compared to past years. During the same period our periodic newsletter, Debian Project News, has been resurrected after a few years of inactivity. (By the way, all this has been possible thanks to the work of the Debian press and publicity teams, that are wonderful environments for those who have writing skills to contribute to Debian.)
But you are right in saying that one rarely sees Debian mentioned in major media (newspapers, consumer shows, etc.), whereas corporate-backed distributions are starting to show up there. My feeling is that to get there you need the typical network of press contacts that commercial entities are better at maintaining, given their incomes depend on it, than volunteer entities.
When I interviewed you before you began your first term, I asked whether Debian would start a trend of transparency as far as finances went. Has there been any movement on this front?
We want to publish comprehensive quarterly reports of all Debian transactions, but we aren't there yet. The blocker turned out to be that SPI, one of the organisations holding Debian money, was not yet ready to provide us with a periodic parse-able log of our transactions. That is the price to pay for going the ecosystem way instead of reinventing wheels: we depend on the functioning of others. But on the other hand that is also the strength of the ecosystem: once the problem will be solved for us, and we are almost there, all other Free Software projects hosted by SPI will benefit from the improvement.
You made a statement during your first campaign for leader, namely "I will fight strong package ownership when it conflicts with quality". How far have you succeeded with this?
My interest in that is something that predates my role as DPL. It all started with my RCBW initiative, which was meant to lower the barrier for contributions within the distribution and to make the release process more widely participated. Key to that initiative is the usage of Non-Maintainer Uploads (NMUs), i.e. the act of making changes in a package you don't maintain yourself.
NMUs have had bad publicity in the past, as they could upset legitimate maintainers, but they are fundamental to implement in Debian the agile mantra of "collective code ownership". Since then, many people have joined the initiative and I hope more will do in the future, as it's particularly important for release awareness during freezes, and the Wheezy freeze is now just around the corner.
I don't have numbers to claim I "succeeded" in that, let's say, "cultural" effort. But I've surely been a loud advocate of NMU for the past two years as DPL. I have a feeling they are nowadays way more accepted than they were years ago, which is good, and I hope the trend will continue.
To what extent have communication channels between Debian and Ubuntu improved since you became DPL?
Thanks to the interest of many people from both camps, many cross-distro initiatives ensued, like the Debian Derivatives front desk and the Derivatives Census, to name just a few. Two years later, as I reported at a recent UDS, the amount of patches forwarded from Ubuntu to Debian is at its maximum and we see a steady flow of new Debian contributors coming from an Ubuntu background.
But it's not all roses: while communication between Debian and Ubuntu (as a project) seems to be on the rise, communication between Debian and Canonical (as a company) seems to stagnate at times. For one thing, I regret the lack in Debian of valuable Free Software components developed by Canonical. They probably wish for more interest by Debian in packaging them, while on the Debian side people wish for that software to be more easily portable to distributions other than Ubuntu. Not to mention specific political choices that various hackers consider blockers, such as CLA/CAAs. So, as you see, there are still margins for improvement, and we are working on them.
Your first platform back in 2010 said you would provide more gradual and rewarding access paths to Debian. What has happened on this front during your two years as leader?
For packager contributors, the bigger change has been the introduction of DMs, which predates my DPL years. All in all, it seems to be a sufficient intermediate step which is both popular and synergistic with the process to become full DDs. We haven't seen the need to propose significant changes on that front.
But it is for other kinds of contributors that a radical change has happened a couple of years ago: we have ruled that Debian welcomes as Project Members (AKA "Debian Developers", but more contribution-neutral) contributors active in any project area, both technical (packaging, development, porting, sysadming, etc) and non-technical (writing, accounting, translation, publicity, legal, etc). We now have the first half a dozen developers who became so by working on tasks other than packaging... and we welcome more!
How much progress has Debian made towards being accepted at the corporate level?
"Increasing adoption at the corporate level" - it's another of those areas where corporate-backed distributions tend to be better than community-based distros. To enter that arena one needs to take care of stuff like: sealing deals with hardware manufacturers, building and maintaining a capillary support network for those who want to pay for it, ensuring that important hardware and software parts are "certified" for your distro, etc. All those activities are generally not great fun, and for sure they are not fun for the average Debian Developer.
But as I said in a recent interview this is no excuse. There is a market based on those activities and it plays an important role in the success of Free Software.
So what we have done is to turn the problem around, asking companies that have a strategic interest in Debian to form an interest group where one can discuss how to make it easier to make business based on Debian. At the moment we are inviting companies that are already very committed to Debian, by offering Debian services to their clients and by employing Debian Developers.
Finally, you have stated openly that this will be your last term as leader. What happens if you find out towards the end of the year that you still need another six months or so to bring all your plans to fruition?
I'll just say "too bad" and move on.
One of the main reasons I've decided to run again is to prepare a smooth transition to the next DPL and ensure I've not made myself irreplaceable. I think that's a duty for anyone who end up being in some position of responsibility. That's essentially the only item on my TODO list that should be done by me. Everything else can be handed over to others. So a TODO list that's not empty at the end of the term won't be a big deal. And if at that point I will really want to be working on something which is still pending, nothing prevent me from volunteering with future DPLs to work on that. In my experience DPLs welcome volunteers and are happy to delegate tasks (hint, hint).