The position is a significant--though not entirely unexpected--rejection of the update, the first to the seminal license in 15 years. Linux, the kernel at the heart of an operating system that clones much of generally proprietary Unix, is considered the best-known and most successful example of open-source software.
"Conversion isn't going to happen," Torvalds said in a posting to the Linux kernel mailing list. "I don't think the GPL v3 conversion is going to happen for the kernel, since I personally don't want to convert any of my code."
Torvalds specifically objected to one new provision in the GPL 3 draft that opposes digital rights management, which is technology that uses encryption to control the use of content and running of software. "I think it's insane to require people to make their private signing keys available, for example. I wouldn't do it," he said.
The GPL is a legal document and manifesto of the free software and open-source movements. It outlines several freedoms for collaborative software development, stipulating that a program's underlying source code may be seen, copied, modified and distributed.
The Linux-GPL issue highlights a long-running philosophical split in the collaborative programming movements. Torvalds represents a pragmatic approach that accommodates computer industry prevailing practices. For example, Torvalds worked for years on proprietary software at chip designer Transmeta, and he permits proprietary video card drivers to be loaded as modules into the Linux kernel.
On the other side of the divide is Richard Stallman, founder and president of the Free Software Foundation. His goals are explicitly ethical and social, and his principles are unbending. "The foundation believes that free software--that is, software that can be freely studied, copied, modified, reused, redistributed and shared by its users--is the only ethically satisfactory form of software development, as free and open scientific research is the only ethically satisfactory context for the conduct of mathematics, physics or biology," Stallman and FSF attorney Eben Moglen wrote in a GPL 3 background article.
GPL 3 draft released
The Free Software Foundation released the first public draft of GPL 3 earlier in January. The move began what's expected to be about a year's worth of discussion and revision.
The GPL 3 draft contains new words opposing digital rights management, which Stallman and Moglen regard as technology that restricts freedoms users must have.
"As a free software license, this license intrinsically disfavors technical attempts to restrict users' freedom to copy, modify and share copyrighted works," the draft license states. "No permission is given...for modes of distribution that deny users that run covered works the full exercise of the legal rights granted by this license."
In other words, some form of locking of GPL code to prevent changes from an authorized version is forbidden.
Torvalds' position is not a surprise. In a 2003 posting to the kernel mailing list, the Linux founder explicitly opened the door to DRM.
"I also don't necessarily like DRM myself," Torvalds wrote. "But...I'm an 'Oppenheimer,' and I refuse to play politics with Linux, and I think you can use Linux for whatever you want to--which very much includes things I don't necessarily personally approve of."
Torvalds founded the Linux project in 1991, the same year the current GPL version 2 was released, and is still its leader. His kernel project dovetailed with work Stallman had already began to create a free clone of Unix, called Gnu's Not Unix (GNU). Because of that combination, the Free Software Foundation prefers the entire operating system be called GNU/Linux--though it has other important components, such as the Xorg graphics system, that come from other groups.
In a 2004 interview, Torvalds indicated he wants the GPL to serve nothing beyond the single function of keeping source code open.
"I really want a license to do just two things: Make the code available to others, and make sure that improvements stay that way. That's really it. Nothing more, nothing less. Everything else is fluff."
Because of that cautious stance, Torvalds specifically didn't follow with Linux the Free Software Foundation's recommendation to describe a software project as governed by version 2 or "any later version."
The issue of moving to GPL 3 is grounded in copyrights. Many open-source projects, such as MySQL or OpenSolaris, require that programmers turn over copyrights to a central organization. That organization then grants the programmers a license of their own to the software source code in question. But with Linux, the copyrights are held by a large number of individuals and companies that contributed the code.
To convert Linux to GPL 3, it's likely more than just Torvalds' approval would be required. For example, when the SpamAssassin project converted to the Apache License so it could become part of the Apache Software Foundation, project organizers spent months getting explicit permission for the change from about 100 copyright holders. Even then, not all contributors could be found, and some software had to be rewritten.
The Free Software Foundation also has lodged objections about Torvalds. In an interview after the GPL 3 draft was released, Moglen said Torvalds doesn't use a "pure GPL" and that practices such as permitting proprietary video drivers violate the license.
Missing out
Keeping Linux with GPL 2 means the project won't be able to take advantage of some changes. And some experts believe GPL 3 is better.
"I think it's a definite improvement. It clarifies where there is ambiguity and deals with issues that have come up over time," said Mark Radcliffe, an intellectual property attorney with DLA Piper Rudnick Gray Cary who represents the Open Source Initiative and who is overseeing some gathering of commentary for the GPL 3.
Regarding rights management, Radcliffe said Stallman "views DRM as potentially evil. He wants to make it very clear that DRM is not permitted, and you cannot implement DRM systems using GPL code."
But Radcliffe also believes those fears could be overstated, judging by the commercial failures of attempts to control software in the past--such as with hardware "dongles" that must be attached to a computer before a particular program will run. "The practical risk of it being applied to software is lower than it being applied to content," he said.
