OpenSolaris Community Picks Board, Gets to Work

by Timothy Prickett Morgan

The five initial members of the advisory board of the OpenSolaris community held their first meeting this week, and Sun Microsystems, the company that is sponsoring the opening up of its Solaris Unix operating system variant, held a conference call to announce who the five members of the board were and to give those members a chance to outline their plans for OpenSolaris and the community.

As Sun announced earlier in the year when it divulged its plans for the OpenSolaris community, the initial community advisory board would have five members. Sun decided that to bootstrap the community, it would choose two members of the board from the pilot OpenSolaris community that has been running behind the scenes before Sun's OpenSolaris plans were made public; that community has about 140 members. Another two would be chosen from the 1,000-strong Solaris engineering group inside Sun, and a final fifth member would come from the open source community at large.

The luminary that the OpenSolaris community brought in is not, as some had suggested, Sun founder and open source BSD Unix creator Bill Joy (which might have been a little too close to Sun, on second thought), but an equally excellent choice: Roy Fielding.

Fielding was one of the founders of the World Wide Web project in 1993. He was also the primary architect of the HTTP 1.0 protocol that is the basis for Web serving and surfing, one of the co-founders of the open source Apache HTTP Web server project and the past chairman of the Apache Foundation, which governed that open source project. He is intimately aware of all of the governance and licensing issues involved with building a community and he has big-time credibility with the open source community at large. Fielding's day job is chief scientist at Day Software, a content management software provider based in Irvine, California. He said joining the OpenSolaris community was one way that he could pay Sun back for the Unix workstations he used when he was a graduate student to work on code for the Apache Web server.

The pilot OpenSolaris community chose Rich Teer, a Solaris consultant who has written one of the bibles of the Solaris market ("Solaris Systems Programming"), and Al Hopper, a software engineering consultant who is president of Logical Approach, a systems integrator based in Plano, Texas. Both have extensive experience with Solaris, and have been vocal about what they think Sun should do to promote it--especially when it comes to Solaris for X86. During the conference call, there was an uncomfortable moment when they were asked how they might interface with Open Source Development Labs, the consortium that is steering the development of the Linux operating system. After a long pause, and a politically correct and ambiguous comment from Tom Goguen, Sun's vice president of marketing for operating systems, it was Hopper who answered the question properly. "Our focus is OpenSolaris, and we are really independent of any other operating system organization," he explained. "We want to be leaders, not followers. We really feel that OpenSolaris is the leader, and we are not interested in looking over our shoulders."

It is safe to say that the OpenSolaris community will not have much official contact with OSDL, although it will monitor Linux and other open source projects that could bear on Solaris from a technology standpoint. Just as Sun would do if it had absolute control over Solaris, as it has had for two decades.

Sun chose Casper Dik, a software engineer from Holland who has 15 years of Solaris experience and who has spent the past five years working on security software for Solaris, to be its initial representative on the OpenSolaris community advisor board. Simon Phipps, the chief technology evangelist at Sun, rounds out the initial board.

Sun has committed to having the full OpenSolaris code base ready and launched by the end of the second quarter, and now the race is on to create a transparent community process and the bylaws of its governance that the members of the OpenSolaris community can ratify.

Sun will not launch the current Solaris 10 production version to the community, but rather the next dot release of the software, which is code-named "Nevada" and can be conceptually thought of as OpenSolaris 10.1. The OpenSolaris community will initially be comprised of those 1,000 Sun software engineers and anyone else who joins and contributes to the cause. Everyone on the call took great pains to explain that the OpenSolaris community would be a meritocracy comprised of contributing members (who just so happen to mostly work for Sun) and using a distributed, peer review process that Sun has been using internally for years. Ultimately, the OpenSolaris community itself will decide if it wants to work this way--not Sun. And if it changes course, Sun will live with it and not interfere. This is not, in short, the Java Community Process, which Sun dominates. Some might say stifles. And OpenSolaris is not just Sun "throwing code over the wall," as Phipps put it. Sun will participate in the community, abide by its governance and coding policies, and take snapshots of the code when it deems appropriate and roll it up as commercial Solaris releases, complete with Sun's traditional testing and qualification of applications and platforms.

It is unclear what compilers will be necessary to make a build of OpenSolaris. Right now, the 64-bit Opteron version of Solaris 10 for X86 is compiled using the open source GNU gcc compiler, while the Intel X86 version and the Sparc version are compiled using Sun's own compilers. Goguen said that the OpenSolaris code and all of the tools necessary for making a build of the operating system for all X86 and Sparc platforms would be available when the code launches this quarter.

When asked about the pitfalls the OpenSolaris community had to avoid, Fielding had some sound advice, which is going to be tough to take considering that Sun will utterly dominate the OpenSolaris community at the start. "The main thing we are trying to avoid is a common problem with vendor-sponsored projects," he explained. "That is essentially being loved to death by the original sponsor. The OpenSolaris community needs to act independently and to promote meritocracy."

The trick will be to get all of those Solaris engineers to really believe that even though they get their paychecks from Sun, they can now contribute in ways they think are best--and that they do so as individuals, not Sun employees.