In his opening keynote at the Linux Plumbers Conference in Portland, Greg Kroah-Hartman did so succinctly, if bluntly. His metric? Kernel contributions. Simple.
Using this yardstick, Kroah-Hartman (employed by Novell (not a RedMonk customer), but not speaking on their behalf) summarily judged and found wanting Canonical (a RedMonk customer), the commercial backer of Ubuntu (the distribution that I use). We can debate the merits of his usage of the venue for that purpose, the manner in which he pursued the issue, his definition of what constitutes upstream, or even the numbers themselves, but really that’s all academic at this late date.
Underneath all the rhetoric and the broadsides lies a real question: is Canonical a member in good standing of the Linux ecosystem? The answer, in my view, is that it depends on the metric you use. As usual.
As Amanda notes, Kroah-Hartman clearly is passionate about the importance of upstream contributions – and possibly equally passionate about his belief that Canonical is not pulling its weight in that regard. Which is his right, as a kernel developer and an individual. Nor is he alone in this perception; I can’t count the number of times individuals and commercial entites alike have disparaged Canonical’s role as a consumer.
Nor would Canonical fundamentally dispute the assertion, if the scope were limited to the Linux kernel; Ubuntu CTO Matt Zimmerman is quite candid on the point, saying:
Canonical is primarily a consumer of the Linux kernel. It is one of the building blocks we need in order to fulfill our primary mission, which is to provide an operating system that end users want to use. It is, on the whole, a good piece of software which meets our needs well. We routinely backport patches from newer kernels, and fix bugs which are particularly relevant to us, but our kernel consists almost entirely of code we receive from upstream.
On that point, then, there is general consensus: we may take as a given that Canonical – whatever the numbers might actually be – not a significant contributor to the kernel, or a few closely associated projects. Especially relative to Novell and Red Hat (a RedMonk customer). The question then is what you make of that.
To paraphrase Old School‘s Beanie Campbell, the whole town knows what Kroah-Hartman makes of that: the presentation’s clear mission was to embarrass Canonical. We also, thanks to Matt’s response and a piece from Mark, know some of what Canonical thinks on the subject: not shockingly, they take exception. Amanda, employed by the Linux Foundation, tends to agree, further arguing that the method employed was “unfortunate.” Last, one of my analyst colleagues – Jay Lyman – allows that Kroah-Hartman seems “to be taking a bit of a confrontational approach to Canonical.” LWN, meanwhile, was sympathetic.
Personally, I think the attacks on Canonical are ill conceived. If you disagree, ask yourself this question:
Assuming that Canonical has finite resources, are they better directed towards integration, packaging and polish, or kernel development?
Looked at that way, to me, the answer is simple: the former. The kernel, at present, does not generally suffer from a lack of contributors – at least as far as I’m aware. A Linux Foundation whitepaper from April, in fact, listing Greg as a coauthor, wrote that the “rate of change in the kernel is high and increasing, with almost 10,000 patches going into recent kernel releases.” I understand that as a Novell or Red Hat employee, it probably is irritating to see downstream kernel consumers building on your work, but in the grand scheme of things, Canonical’s lack of kernel contributions means little to customers. Focusing on a single player for their relative lack of kernel contributions, then, seems to me as an attempt to solve a corporate, rather than community, problem.
More, it seems to ignore what Canonical is actually good at; what they have, so to speak, brought to the proverbial table. The Ubuntu team and their commercial sponsor have carved out a niche for themselves as the packaging experts; it’s my relatively educated opinion that they do it better than anyone at the moment. Even better, the competition they’ve triggered has led to rapid improvements in other, alternative distributions. It’s my firm belief that neither Fedora nor OpenSuSE would be remotely as usable as they are today absent Ubuntu. This style of contribution – or its relative importance – may not be immediately apparent to all audiences, but as an Ubuntu user I appreciate it. And judging by the decision of major IHVs like Dell to ship it, so do they.
If Novell and Red Hat are then the better plumber, to beat the housing analogy to death, Canonical, to me, is the better designer/architect. And much like I don’t want a house without running water, I’d prefer one that’s designed to be livable. It takes all kinds, as they say.
When I consider relative contributions, then, to the community, my metrics go beyond kernel commits. But to each their own, of course.
