Fedora: LiveCD Roadmap
To produce a sustainable Live CD project for Fedora -- one with the resources and credibility it deserves.
In the summer of 2005, Red Hat sponsored several projects with Google's Summer of Code. One of these projects was named "Kadischi". Led by Darko Ilic, the Kadischi project was a set of scripts, based on Anaconda, that created a system image in a change root and then put that system into an ISO image. The Kadischi project has since acquired a number of users and a few developers, and has had some moderate success -- but its roadmap, and its leadership, have remained unclear.
In the summer of 2006, David Zeuthen of Red Hat revealed a Live CD based on his work from a project called Pilgrim. David is a developer for the One Laptop Per Child project, and the main goal of Pilgrim -- create system images that can easily run from a flash drive -- maps very closely to the goals for a proper Live CD. David spent a weekend retrofitting Pilgrim to produce a Live CD as well.
Thus, we have two codebases in Fedora that are designed to produce a Live CD.
We're now ready to roll the creation of an official Fedora Live CD into the Fedora release process. The Fedora Unity project has proven the viability of such an approach, and JesseKeating has begun work on pungi, a tool that will handle all of these distro release tasks for the Fedora Desktop and Fedora Server releases. The open question: how do we roll Live CDs? Which tool do we use, Pilgrim or Kadischi?
We've decided on Pilgrim, for a few key reasons:
- Pilgrim has a passionate leader in David Zeuthen (primary HAL developer and maintainer). An important project needs a strong lead, and we believe that DavidZ is that lead. Kadischi has meandered a bit, with no one really able to take the lead and drive significant progress.
- Pilgrim has multiple downstreams: both Fedora LiveCD and OLPC. As such, there's a much lower chance that resources will be taken away from the Pilgrim project; Pilgrim is vital to the success of OLPC. Therefore, Pilgrim is likely to be a stronger codebase over the long-term.
- Pilgrim does not rely upon Anaconda, which has been a key weakness for Kadischi; when Kadischi developers have asked for changes to the Anaconda codebase, it has been difficult to accommodate those changes because of the importance of the stability of Anaconda to Red Hat's business.
- Work out the public roadmap on fedora-livecd-list.
- Work with the Fedora Unity project on an official Live CD beta release based on Pilgrim.
- Integrate pilgrim livecd build process into pungi.
- Work towards making "install via dd" into a pluggable install mechanism for anaconda.