Compiz Considered in Peril

Originally developed by Novell in 2006, Compiz delivers window manager daemons for GNOME and KDE and provides modern graphics for the Linux desktop. The Beryl composition manager project had originally forked from Compiz, but the two were reunited in April 2007 as the Compiz Fusion product, which has conceptually remained an experimental branch where new functionality is tested and eventually merges into the window manager. It is from just this merge onward, asserts developer Kristian Lyngstol in mail on bohemians.org, that no real progress has been made in Compiz. "The reason for this, from my point of view, is a complete lack of direction and leadership," he writes, and he sees the future rather darkly: "...whether we call it an object framework, nomad or Compiz++, the reality is that all these branches are counterproductive, regardless of how fun or flashy they are."

The last few months have seen little influx of new developers to the project; in fact, many have left. "We MUST turn this trend around if Compiz is to survive," emphasizes Lyngstol. He sees three reasons for the vanishing developers. As the first, "the project has no goals, and essentially all development and design is done as a solo race," with the inherent risk that development work is lost when a remerge occurs. His second reason is “inconsistent organization”: "Two bugtrackers, one isn't really cared for. Two places to find code... Two development mail lists. Messy." The third reason is the lack of documentation. Lyngstol perceives Compiz foremost as a research project, with "very little work [done] to bring Compiz into a state where it can be considered truly stable. We need to stop using [the] Compiz master [branch] as an experiment."

As a solution to the problem Lyngstol suggests that authors and owners get together to decide what to do about the three product branches. If this doesn't happen, he writes, "we really have no other choice but to consider those branches forks of Compiz, and move ahead based on master." Lyngstol is quite anxious to get things going again on this basis: "I am ready to do the boring development work, but not until these management issues have been sorted out."

Lyngstol is not the only one concerned about the status of "his" project into the new year. Novell developer Michael Meeks recently feared for the future of the OpenOffice.org project on account of the same vanishing developers syndrome.