Interview: Gentoo's Berkholz and Gaffney Look Ahead
I recently got the chance to talk with Gentoo Catalyst/Genkernel Release Engineering Lead Andrew Gaffney and developer/council member Donnie Berkholz about Gentoo Linux, the challenges it is facing, and the work that is being done by the community -- from veteran developers to new users -- to overcome them.
OStatic: On the gentoo.org site, it was announced that Gentoo would be taking a more "back-to-basics" approach to releases, with an emphasis on making the most of Gentoo's human resources. No doubt the stress of the 2007.1 release cancellation and unavoidable (and not necessarily Linux/Gentoo related) delays of 2008.0 took a toll on developers and engineers. What are the plans (even tentatively) to get Gentoo back on track for a regular release cycle?
Andrew Gaffney: At this point, there are no plans to get back on a track for a regular release cycle. There are tentative plans to perhaps do some regular (bi-annually) LiveCD releases down the road, but that's not something we care about immediately. The plan is just to do the automated weekly builds of stages and the minimal CDs.
OStatic: In an article Donnie wrote for LWN, a few problems contributing to the previous release delay were detailed. Two issues cited were that release work isn't really anyone's favorite task, and that reproducing installation bugs is difficult, due to an almost endless array of hardware combinations and bug reports that don't always give complete or relevant information. How are these problems being addressed presently, and are they approached differently now than prior to the 2008.0 release? Do you think these will be addressed differently as the approach to the release cycle/patterns of LiveCD versus minimal CD and stage tarball releases changes?
Andrew Gaffney: Part of the problem previously is that it was a very small group of people working on a very specific tree snapshot. Now, we'll be working with the normal tree, so any bugs with the stages can be pushed off to the relevant teams instead of having to deal with them ourselves within releng [OStatic note: Gentoo's release engineering team]. There will always be bug reports with little to no useful information, but at least they will be spread out a bit, so we don't have to deal with a bunch of them all at once.
OStatic: The third problem was the tendency of people to want to offer opinions on how things should work without seeing the big picture, or perhaps understanding how the project works in reality.
This seems like it would be the most difficult problem to approach, and one that could, if handled badly, stand to really damage the project. It seems as though it would come down to communication -- how do you encourage ideas from all segments of the Gentoo user and developer base, and how do you show the user/developer base how the project actually fits together in the end? How are ideas/suggestions handled currently? How much time can the development team devote to working with ideas that may be "diamonds in the rough" -- a great idea at the core, but the method presented for realizing it might not fit within the realities of the Gentoo universe right now?
Donnie Berkholz: We want to encourage communications at the right level for each group -- developers, potential contributors, or other users. That level is wherever they have enough context and background knowledge to contribute most effectively. This varies from participating in places like the web forums -- beginning at asking questions, then proceeding to answering them -- to ways closer to development, like filing bugs, and later supplying patches with them.
Another way we try to address this is by providing transparency into how our various projects and teams work. We do this by documenting it on our web site and through communication on our mailing lists and IRC channels.
As far as working with rough ideas, the Gentoo project is just a big group of independent people. If any of them see promise in an idea, they will pick it up and develop it further. Ideas show up everywhere Gentoo users and contributors are, from blogs across the web to IRC, Bugzilla and mailing lists.
OStatic: Certainly Gentoo has at least a somewhat rough road ahead. What do you feel are Gentoo's strongest points? What components of Gentoo Linux make the bumps a little less jarring, and make you feel that the rough road is well worth traveling?
Donnie Berkholz: Gentoo's strongest points are the developers committed to a positive environment where we can all have fun while making a great distribution. Any time I begin to waver in my faith in Gentoo, I think of them and it renews my energy and desire to make Gentoo great.
OStatic: Thanks, Andrew and Donnie, for your time.