Recently, we’ve been having some discussion around the use of non-free software and services in doing our Debian work. In judging consensus surrounding a discussion of Git packaging, I said that we do not have a consensus to forbid the use of non-free services like Github. I stand behind that consensus call. Ian Jackson, who initially thought that I misread the consensus later agreed with my call.
I have been debating whether it would be wise for me as project leader to say more on the issue. Ultimately I have decided to share my thoughts. Yes, some of this is my personal opinion. Yet I think my thoughts resonate with things said on the mailing list; by sharing my thoughts I may help facilitate the discussion.
We are bound together by the Social Contract. Anyone is welcome to contribute to Debian so long as they follow the Social Contract, the DFSG, and the rest of our community standards. The Social Contract talks about what we will build (a free operating system called Debian). Besides SC #3 (we will not hide problems), the contract says very little about how we will build Debian.
What matters is what you do, not what you believe. You don’t even need to believe in free software to be part of Debian, so long as you’re busy writing or contributing to free software. Whether it’s because you believe in user freedom or because your large company has chosen Debian for entirely pragmatic reasons, your free software contributions are welcome.
I think that is one of our core strengths. We’re an incredibly diverse community. When we try to tie something else to what it means to be Debian beyond the quality of that free operating system we produce, judged by how it meets the needs of our users, we risk diminishing Debian. Our diversity serves the free software community well. We have always balanced pragmatic concerns against freedom. We didn’t ignore binary blobs and non-free firmware in the kernel, but we took the time to make sure we balanced our users’ needs for functional systems against their needs for freedom. By being so diverse, we have helped build a product that is useful both to people who care about freedom and other issues. Debian has been pragmatic enough that our product is wildly popular. We care enough about freedom and do the hard work of finding workable solutions that many issues of software freedom have become mainstream concerns with viable solutions.
Debian has always taken a pragmatic approach to its own infrastructure and to how Debian is developed. The Social Contract requires that the resulting operating system be 100% free software. But that has never been true of the Debian Project nor of our developers.
- At the time the Social contract was adopted, uploading a package to Debian involved signing it with the non-free PGP version 2.6.3. It was years later that GnuPG became commonly used.
- Debian developers of the day didn’t use non-free tools to sign the Social Contract. They didn’t digitally sign it at all. Yet their discussions used the non-free Qmail because people running the Debian infrastructure decided that was the best solution for the project’s mailing lists.
“That was then,” you say.
- Today, some parts of security.debian.org redirect to security-cdn.debian.org, a non-free web service
- Our recommended mirror (deb.debian.org) is backed by multiple non-free CDN web services.
- Some day we may be using more non-free services. If trends in email handling continue, we may find that we need to use some non-free service to get the email we send accepted by major email providers. I know of no such plan in Debian today, but I know other organizations have faced similar choices.
Yet these choices to use non-free software and non-free services in the production of Debian have real costs. Many members of our community prefer to use free software. When we make these choices, we can make it harder for people to contribute to Debian. When we decline to use free software we may also be missing out on an opportunity to improve the free software community or to improve Debian itself. Ian eloquently describes the frustrations those who wish to use only free software face when faced with choices to use non-free services.
As alternatives to non-free software or services have become available, we as a project have consistently moved toward free options.
Normally, we let those doing the work within Debian choose whether non-free services or software are sufficiently better than the free alternatives that we will use them in our work. There is a strong desire to prefer free software and self-hosted infrastructure when that can meet our needs.
For individual maintainers, this generally means that you can choose the tools you want to do your Debian work. The resulting contributions to Debian must themselves be free. But if you want to go write all your Debian packaging in Visual Studio on Windows, we’re not going to stop you, although many of us will think your choices are unusual.
And my take is that if you want to store Debian packages on Github, you can do that too. But if you do that, you will be making it harder for many Debian contributors to contribute to your packages. As Ian discussed, even if you listen to the BTS, you will create two classes of contributors: those who are comfortable with your tools and those who are not. Perhaps you’ve considered this already. Perhaps you value making things easier for yourself or for interacting with an upstream community on Github over making it easier for contributors who want to use only free tools. Traditionally in Debian, we’ve decided that the people doing the work generally get to make that decision. Some day perhaps we’ll decide that all Debian packaging needs to be done in a VCS hosted on Debian infrastructure. And if we make that decision, we will almost certainly choose a free service to host. We’re not ready to make that change today.
So, what can you do if you want to use only free tools?
You could take Ian’s original approach and attempt to mandate project policy. Yet each time we mandate such policy, we will drive people and their contributions away. When the community as a whole evaluates such efforts we’ll need to ask ourselves whether the restriction is worth what we will lose. Sometimes it is. But unsurprisingly in my mind, Debian often finds a balance on these issues.
You could work to understand why people use Github or other non-free tools. As you take the time to understand and value the needs of those who use non-free services, you could ask them to understand and value your needs. If you identify gaps in what free software and services offer, work to fix those gaps.
Specifically in this instance, I think that setting up easy ways to bidirectionally mirror things between Github and services like Salsa could really help.
- We have come together to make a free operating system. Everything else is up for debate. When we shut down that debate—when we decide there is one right answer—we risk diluting our focus and diminishing ourselves.
- We and the entire free software community win through the Debian Project’s diversity.
- Freedom within the Debian Project has never been simple. Throughout our entire history we’ve used non-free bits in the sausage making, even though the result consists (and can be built from) entirely free bits.
- This complexity and diversity is part of what allows us to advocate for software freedom more successfully. Over time, we have replaced non-free software that we use with free alternatives, but those decisions are nuanced and ever-changing.