March 20, 2003 9:55 PM PST
Schism hits key open-source group
The six-member Core Team of the XFree86 project announced Thursday that it had ejected member Keith Packard for trying to create a parallel XFree86 project and refusing to discuss reasons for the radical move with the rest of the core team. The core team disclosed the ouster in conjunction with an announcement of a new mailing list to discuss XFree86's future.
In an interview, Packard didn't comment on specific actions but indicated that he was trying to make it easier for interested and qualified programmers to contribute to the XFree86 project. "XFree86 is not currently a friendly place to play," he said.
XFree86, which sends graphics commands for tasks such as drawing windows to video cards, is a crucial part of Linux and several other operating systems. Packard has been leading improvements to font technology.
The stakes in the issue are high. Open-source advocates often boast of the advantages of the collaborative programming model, in which the source code underlying a particular program may be freely shared and changed. One drawback of this freedom, however, is that people with different ideas can "fork" the software into different projects that are incompatible, that dilute programmers' energy by forcing them to duplicate the same features or that require outside programmers to work with more than one group.
Forking isn't all bad, though, advocates say. It can act as a safety valve to prevent a misguided group from gaining too much control, and it can be used to provide competition to assess which of two approaches is best. Indeed, forks of the heart of Linux itself, the kernel, frequently branch off as programmers work on pet projects they hope eventually will be accepted by leader Linus Torvalds into the mainstream.
The XFree86 raised the forking issue as evidence of the seriousness of Packard's actions.
"It has been brought to the attention of the XFree86 Core Team that one of its members, Keith Packard, has been actively (but privately) seeking out support for a fork of XFree86 that would be led by himself," a message from the Core Team and the XFree86 board of directors said.
Packard also is forming a group of people with "vested interests" to discuss his concerns about XFree86 but has refused to disclose those concerns, the group said. "As a consequence, Keith Packard is no longer a member of the XFree86 Core Team," the group said.
Packard defended his actions, saying he was gathering information before making his case. "I was trying to get some help in framing the discussion from members of the community before making incoherent statements to the XFree86 management," he told CNET News.com. "I'm still trying to do that."
But Packard abused his privileges as one of the handful of people authorized to "commit" changes to the XFree86 code base, said XFree86 project leader David Dawes in remarks to the new forum mailing list. Specifically, Packard added a new "Xfixes" section to the software "without any prior discussion, let alone public review," Dawes said.
David Wexelblat, one of the XFree86 board members who signed the original notice of Packard's expulsion, directed reporter queries about the group's actions to his own posting. "While still a member of the XFree86 Core Team, (Packard) has explicitly attempted to subvert XFree86 by soliciting individuals and corporations to create an alternative to XFree86," Wexelblat said.
But some changes are afoot, Wexelblat added. For example, the project needs more people who are authorized to lock in software changes, "mostly to reduce load on the people who are currently responsible. I believe this is already in progress," he said.
One source familiar with the debate said Packard's actions are aligned with the views of top Linux seller Red Hat and server maker Hewlett-Packard, which employs Packard. Red Hat and HP didn't immediately respond to requests for comment.
The split left some distressed.
"Bad mistake," said Alan Cox, a top deputy to Torvalds and an employee of Red Hat, in a response to the XFree86 Core Team's move to oust Packard. It would have been better to let him try an experimental project within the XFree86 community then evaluate that project on its merits, he said.
"XFree86 is hard to get involved with usefully, resistant to cool ideas," Cox said. The project suffers from "plodding progress," a reluctance by leaders to delegate decisions to lower-ranking programmers and preference for infrequent releases rather than the frequent updates used for the Linux kernel itself.
And in January, another Red Hat programmer, Mike Harris, said XFree86's actions damage relations with outside companies such as graphics card maker ATI Technologies, which has had to wait "months and months" for software updates to be accepted.
"How long is ATI going to continue submitting patches to XFree86.org that take nine months to get (accepted), and then perhaps another four to six months to be available in an operating system," Harris asked. "Quite frankly, if I were ATI, and submitting patches as frequently as they do, and the patches just sat there, I might start thinking twice about bothering in the future."
Others urged an amicable resolution.
"I think I'm going to cry," lamented one commenter at the Slashdot "news for nerds" discussion site. "Please try to work this out with Keith...Don't end up hating each other and refusing to share your code with each other. Either side of such a fork would have a much weaker team."