A face-to-face conversation with Max Spevack of Fedora
Max Spevack, Chairman, Fedora Project
We had the opportunity to sit down face-to-face with Max Spevack, chairman of the Fedora project, at the Red Hat Summit in San Diego to talk about all things Fedora – the merger of Fedora Core and Extras, Fedora 7, and the road ahead. Here are Max's responses to our questions.
Q: What is the vision for Fedora today?
Max: One of the things that we've been doing very actively for the last year and a half, certainly since I've been the Fedora leader and even earlier during Greg DeKoenigsberg's leadership, has been to transition the governance and decision making process in Fedora to be more community focused. We have tried to build a technical leadership in Fedora that can be owned and run by the community. If those community members happen to be from Red Hat, that's great and if they are non Red Hat contributors, that's fine too. So the Fedora Extras project started as a way to do this. The leadership group of that effort, for the vast majority, were non Red Hat folks. It took about two years or so but Fedora Extras was very successful. We saw that the packages being built by community volunteers in a lot of ways were better than the packages that were built purely within Red Hat. The guidelines for building those packages were better, the end results were better. And so the guidelines and decisions that the Fedora Extras community made were slowly adopted into Fedora Core. Now we've thrown away the idea of Core completely and merged everything into one repository that basically follows all the rules that Fedora Extras had set out. So now the build system that all the packages of Fedora get built on is all completely external, all completely open source. Everything relies on the community and there are a lot of interesting things that means.
Q: What is the vision for Fedora tomorrow?
Max: The two biggest goals for Fedora 7 were the live CD/live DVD stuff which was something Fedora needed in order to be able to compete at the same level as other distributions. So we've got that.
In Fedora 8 and 9, over the next year let's say, we want to make the technology even better. For example, envision you take your whole computer around on your USB key and you stick it into any laptop and your desktop just appears. If you lose it, your data is encrypted, so it's not as much of a catastrophe as it could be – at least your bank account is not all over the world.
A big goal for Fedora 7 was the merge of Core and Extras and the result was anyone being able to generate their own custom version of Fedora as easily as possible. What we've got there are some tools. The basic ones that were built include "Pungi", which is the distribution compose tool. It takes a pile of RPMs and turns them into an installation tree. And there's also "livecd-creator", whose name makes it pretty obvious what it does. Livecd-creator takes a bunch of packages and turns them into a live CD. Then there's another tool that takes that live CD and copies it onto a USB key called the "livecd-iso-to-disk" script. These three tools are command line tools and for someone to use them takes a little bit of actual engineering skill. We want to lower that barrier as much as possible. So another project that's underway is called "Revisor". Revisor is cool for a few reasons. It was completely developed by non Red Hat members who understood the vision that the rest of the Fedora project had and kinda went with it. It takes those three tools I just mentioned and puts a really nice UI on top of it so that it's basically like a wizard work flow where you go through a graphical interface and select the packages you want. You can select if you want a live CD or an installable CD and there's other options they're working on. You just select what you want and it will do it for you.
So all of a sudden the ability to customize Fedora doesn't hinge on you being a release engineer or a programmer at all. Not everyone will want to do that but I think that there are a lot of people who will. And there are some really interesting possibilities. One is businesses, say ISPs or small businesses that run Fedora right now who have their own application software that they would add to the distro. Now they could add it in at build time as opposed to after install, and distribute it however they like, which is kind of cool. Another interesting possibility, especially in the non-US use case, is "I don't want all this stuff in English. I want it in whatever my local language is. And maybe I'm a country that doesn't have super amounts of bandwidth and so I want something that is localized and a pretty small download size." Now people can build that themselves. And so I think that's a pretty useful case. And then the third interesting possibility is that it fosters a sense of innovation and competition within the community where you can have different people working on putting together the best package set for desktop, or the best package set for whatever server they want to use. Or people can compete to see how small of an install size they can make something that still can connect to the internet, with YUM to go get more.
Q: Virtualization is emerging in a big way. What are Fedora's strategies for supporting Xen, KVM, QEMU, etc.?
Max: Xen is obviously a big part of what's in Fedora, but QEMU and KVM are also in there. I think Fedora has a responsibility to offer everything. Just because Red Hat puts a lot of resources and money behind Xen doesn't mean that Fedora should ignore the other available options. There are people who are going to want them, so we make them all available.
Our Fedora infrastructure team uses Xen in pretty much everything. It's the ultimate in eating our own dog food. We are using Fedora to build Fedora and we were using Xen in production long before Red Hat was saying well we'll sell you Xen in RHEL. That's something that's ready for production.
But I like the idea of making everything easy enough to use that you don't have to be an expert to use it. I think of myself as a good use case for a lot of the Fedora stuff. I have an engineering background but I'm not a super duper programmer. I couldn't write the stuff myself. But I know what it's supposed to do and how it's supposed to work. If I can use it, it's probably OK to give away. But if it's too complicated for me, it's certainly going to be too complicated for a lot of other people. I'm not going to insist that my mother be able to use it, but I want to be able to use it.
Q: OLPC's desktop sets a different paradigm. Services centric desktop vs. something like Sugar. Do you see Fedora taking OLPC's ideas on the desktop such as Sugar.
Max: OLPC is built on top of Fedora so Fedora in a lot of ways serves as an upstream for OLPC. But we know we want to work in a lot of the ideas that have been talked about from the Online Desktop, the future of the Mugshot project. I don't think Fedora is ever going to look like Sugar does, but I think it's going to look like what the vision of the Online Desktop is.
Q: Given the different initiatives that Red Hat has right now for the desktop, do you think that there would be different versions of Fedora for all those different desktops? Or will there just be components that people could use or play around with?
Max: That's a good question. What's cool about the changes we made for Fedora 7 has been infrastructure change, changing the way we actually build Fedora from the ground up. And so Red Hat can do an even better job of competing with itself because Fedora serves as the upstream for RHEL and also serves as the upstream for other Red Hat products. So Fedora is set up in a way that Fedora can kind of keep doing what it does, pushing at the forefront, and the rest of Red Hat can take Fedora and build a classic RHEL that everyone knows. Or produce some other Red Hat Shadowman branded products that look like a totally different desktop. And Fedora can serve both of those downstream components just as well. The ability to do that is one of the requirements of Fedora because it's part of Red Hat and so we have to serve the community on the one hand as well as Red Hat on the other.
Q: Why is Fedora in a holding pattern at only number 4 at DistroWatch. Why isn't Fedora number 1?
Max: You know, I don't like the competition to see who has loudest fanboys. I have wanted to make Fedora cool, to make it work and make it good. This aligns with Red Hat's larger marketing belief that we're just going to make this software and eventually people will notice it's the best out there on merit. I don't spend a lot of time worrying about where we rank on DistroWatch or whatever. We collect and publish the statistics of how many IP addresses come back to yum looking for updates. We have a page that talks about those results and how reliable they are and where potential fudge factors are. I don't see any other big distribution trying to be that transparent. I see a lot of people say "Well I've got 8 million users". And I don't hear anyone asking the follow up question of "Where did that number come from? How do you justify that?" At Red Hat, we're proactively trying to say here's our numbers and then, before you ask, here's where we got those numbers from. That's on a link called statistics on the very front page of the Fedora Wiki.
Max, we appreciate your taking the time to provide us with a great overview on Fedora's progress and the road ahead.
© Technetra. First published July 2007 in LinuxForYou magazine (www.linuxforu.com).
