Defending the flame of Linux freedom
23rd Nov 2008 | 11:00
Interview: Former Fedora Project Leader Max Spevack
Increasingly we're seeing big business jump on the Linux bandwagon, as companies wake up to the money that can be made out of a community of developers working for free.
Someone has to protect that community from being exploited, and if you're a Fedora contributor, that someone is Max Spevack.
Linux Format magazine caught up with Max to ask him about the way the Fedora world is turning.
Linux Format: Max, you were the Fedora Project leader for a while; now you're the manager of the community architecture team. How do those roles differ?
Max Spevack: More or less my entire job is Red Hat community management, and Fedora is the shining example of Red Hat's community work. Part of my job also is to make sure that Red Hat participates in non-Fedora communities, so I spend like 75–80% of my time on Fedora and the other 20% of my time on more general community business.
LXF: Was it the community side or the technical side that attracted you more?
MS: If I'm going to honest about myself I'd have to say that I was a decent programmer, not a superstar programmer. Even in college there were people who just loved to write code for hours and hours and hours, and those were the best programmers. I liked it and was good enough at it to get good grades, but it wasn't an all-consuming passion for me.
I always kind of thought that I would be in a technical company but in a managerial, a leadership type of role, so looking back it doesn't necessarily surprise me that this is the job I've ended up in. I live in awe of our brilliant engineers and the things they can do.
LXF: How well do you think Fedora/Red Hat is addressing the oft-repeated claim that Fedora is just a beta for RHEL?
MS: I hear a lot less of that than I did four years ago. I think the fact that we have built our community in an entirely transparent way says the most about that, because people see Fedora being an extreme innovator, they see SELinux get developed, they see Network Manager get developed, they see Red Hat participate in Xen and the work that's been done in virtualisation, and they see Red Hat doing stuff that's good for all distros, which I think gives Red Hat a lot of credibility as a technical innovator.
Fedora stands on its own as an operating system, and it just so happens that Fedora is upstream of Red Hat Enterprise Linux. No one is going to call Debian a beta of Ubuntu, but Debian is in many ways upstream for a lot of the Ubuntu packages in the same way that Fedora is upstream for a lot of the RHEL packages. That doesn't mean that one is a beta of the other.
LXF: Has there been any call for users to have, say, one in every three releases supported long-term, like Ubuntu, for those who don't want to buy RHEL or use CentOS?
MS: We have a Fedora long-term support: it's called either RHEL or one of the RHEL rebuilds that you can get for free. Ironically, when Fedora was started it was given a separate brand because when Red Hat Linux split into RHEL and Fedora, the Red Hat salespeople were worried that they wouldn't be able to sell certain products.
Fedora was given a separate brand to distinguish between things that Red Hat sold support for and those things that they didn't. Now there are people who use Fedora who have no idea that it's a part of Red Hat, which from my point of view as the Fedora guy is incredibly successful.
But it's a little ironic, because now when we try to explain to people that if you want longer-term support you can go to RHEL or one of the rebuilds of RHEL, not everyone knows that Fedora and Red Hat are part of the same team. Fedora longterm support is there – it just has a different brand name.
LXF: I thought it was fascinating when you said in your talk at LUGRadio Live that Red Hat costs would triple without Fedora.
MS: About two-thirds of the Fedora packages are maintained by community people, and if we didn't have that community, that chunk of work would either not get done, which would significantly harm Red Hat's entire value, or would have to made up by more [paid] engineers. The challenge on the flip side of that is to make sure that everyone in the Fedora community feels valued, that everyone who contributes can be proud of the way that Red Hat uses their code.
LXF: I always got the impression that the Fedora community regards Red hat as a separate distribution. How important do you think it is to them that the quality of what gets rolled in to Red Hat is maintained?
MS: A lot of Fedora contributors maybe use RHEL in their jobs, or work for companies that are using RHEL; there are also a lot of people who contribute through derivatives such as CentOS. We want them to feel that the work we're doing is good.
There's a continuum of distros with respect to how they view freedom, and I think we take a pretty hard line in terms of what we will or will not include in Fedora and on the repositories, and I think people appreciate that and want to see Red Hat defend free software as well. That's what I mean by keeping trust.
When a Fedora contributor is doing a lot of work for Fedora, indirectly down the line he's helping to generate revenue for Red Hat, so we want to be sure that we can show those people what Red Hat is investing back into Fedora, and also that when Red Hat makes business decisions it promotes open source in general.
LXF: When the development process kicks off for a new Fedora release, how does the project go about deciding its goals? Does Red Hat ever conflict with community developers?
MS: The last couple of years have gotten more and more community driven, and that's the way it should be. If you want to have a feature in Fedora and you're a community member or a Red Hat engineer who has a mandate from their boss because they want something in RHEL, the process is the same.
By the time of the feature freeze you've got to be more or less feature complete, and it can get a little subjective, but if you've got your code into Rawhide, which is what we call the Fedora development branch, and you've got it tested and you've got some actual records of having it test cases where it has performed well, then your feature is going to make it in.
That decision is made by the Fedora engineers' steering committee, which is comprised of nine people who are elected by the Fedora community at large.
LXF: Do you think other projects could learn from Fedora's strict six-month release cycle – Debian, for example, seems to have a lot of trouble getting a release out the door on time.
MS: I think in the end we're just very brutal, and our release engineers know that there's going to be a release in six months. If [a piece of software] is going to take six and a half months, it's not going in. It's pretty brutal but we choose to go at that pace.
First published in Linux Format, Issue 112