Intel and XMir

This is a curious piece of historical revisionism, and since I was there I can't let it pass.

Scott was arguably Canonical's first full-time employee (things were a little fuzzy in the very early days and he and Robert Collins had a long-running debate over who was first), hired in early 2004. To start with Scott wasn't doing anything related to init systems, and indeed wasn't initially on the "distro team" that was focusing solely on Ubuntu, but was mainly working on problems related to revision control.

In late 2004 / early 2005, Scott started working on the boot process (recall that that was the time when everyone was moving over to the various components of Project Utopia, including udev), and over a year or so of working on that he came to the conclusion that a number of problems were fundamentally unfixable without replacing the init system. At a Canonical distro team sprint in early 2006 he presented a design for a new event-based init daemon to us, and I remember going over it in considerable detail and giving him feedback on the proposed state machine (I think that was roughly when he settled on the name "Upstart" too). At UDS in June 2006 he presented this as a spec for the Ubuntu 6.10 cycle (and you can see that https://wiki.ubuntu.com/ReplacementInit?action=info lines up with this), it was one of his four specced projects for that release cycle, and we switched Ubuntu over to the new Upstart init daemon in September 2006.

Now, it's certainly fair to say that Upstart was Scott's brainchild rather than something that Canonical management asked him to do or whatever, and he did a lot of the work for it in his free time because it was something he cared about. He also did a lot of it on work time, he had many discussions with me and other Canonical employees about its design and implementation, and he felt strongly about making Ubuntu's boot process reliable as well as running it as an upstream project that other distributions could use.

Upstart would not have happened without Scott, just as many projects (corporate-backed or not) wouldn't have happened without their initial primary developers. But it is quite false to say that Canonical "just hired the main developer", as if we noticed this promising Upstart project and decided we should hire the person who'd been developing it. That's not how it happened.

C'mon Olav.. you of all people know that there is no such thing as a company contributing code to an open source project; open source is contributed to by people, and Canonical's engineers certainly do -- regularly -- contribute code to open source projects in general. It's a small company, though, with 600 employees and about 2/3 that of engineers. If you compare that population to the size of the RHEL engineering contingent, it's no wonder their contributions upstream have less impact.

If you were saying that a) Canonical prioritizes Ubuntu as its own product and b) focuses on integration and polish as opposed to upstream features, there is absolutely no contest there. You're right -- last-mile polish is the area where the biggest gap existed in open source when it was founded, and I think Ubuntu' s success and adoption confirm it was an effective goal to pursue. I truly wish the company had more resource to invest in upstream features -- I myself miss many -- but when you are a small and yet-to-be-profitable company, focus can be a good way to avoid failure.

I know Canonical has its share of policy problems. But as a key influencer, your criticism could help us fix them, as opposed to blending them all into a big Canonical Does Nothing For Open Source meme.

Go and learn about it, otherwise don't claim anything about Red Hat.