Feature freezes can be a relatively boring time to read the linux-kernel
mailing list. Discussions of cool new developments tend to get put to the
side in favor of benchmark results and bug fixes. But even people who wish
for more interesting discourse are likely to agree that when Richard
Stallman starts posting on linux-kernel, things have probably taken a wrong
turn. But, stuffed in between some classic Stallmanisms ("Just as some
people insist the Earth is flat, or that astrology makes valid predictions,
others believe that the whole system is Linux.
") is a discussion of
a fundamental disagreement over the nature of freedom and software.
The issue at hand, yet again, is that of binary-only kernel modules. But
the real, underlying issue has to do with where true freedom is to be
found. Would users of a Linux system that disallowed closed-source modules
be more or less free? In general, what effect does proprietary software
have on freedom?
The point of view championed by Mr. Stallman (and many others) is that
proprietary software is always bad for freedom. For example:
Making a program non-free is denying other people the freedom to
study, change and/or redistribute it. It is an act of domination.
To speak of the "freedom" to dominate others is to stretch the
concept of freedom into a Russell paradox.
According to this point of view, the best case scenario is that a
proprietary program weakens the motivation to develop free alternatives,
and is thus bad for freedom.
The other point of view says that true freedom means letting the author of
a program decide how that program is to be licensed, and letting users
choose which programs they wish to use. A binary-only kernel module gives
Linux users access to (say) more hardware and thus increases their
freedom. Proprietary software can help fund innovation and, even, the
creation of more free software. According to this viewpoint, restricting
proprietary software not only has an immediate (negative) effect on
freedom, it can also impact the availability of free software.
This argument highlights a fundamental division in the Linux community.
It can be swept under the rug much of the time - Linux offers much that is
good for everybody involved, and philosophical differences can be
overlooked most of the time. But the division remains, and it can surface
at inconvenient times.
Any vendor of proprietary kernel modules can not help but be nervous about
this issue. Kernel developers are, as a whole, more concerned with making
the kernel better than with making life difficult for proprietary software
vendors (though they are not always entirely concerned about making life
easy for those vendors). But the potential for lawsuits from a developer
holding copyrights on the kernel source exists. This concern led
developer Andre Hedrick to announce his
withdrawal from Linux development (though he later backed down from that position).
It is thus good that one thing that might actually come out of this long
linux-kernel flame war is a clearer statement of what sort of proprietary
kernel modules are permissible. There may even be an early, rough
consensus along these lines:
- Binary-only modules are acceptible as long as they stick to the
exported API. This is, essentially, the informal understanding which
has been in force for years.
- Kernel header files are considered to be a part of the exported API -
something which has never been clearly stated before. Even more to
the point, inline functions in header files (of which the kernel has
many) are also deemed to be part of the exported API.
This statement, if it holds, makes it clear that proprietary kernel modules
are generally acceptible.
So far, there have not been public objections to this position. If the
kernel developers can settle behind this sort of statement, vendors will
have a better idea of where they stand, and uncertainty in general will be
reduced. The difference over opinion on freedom will remain, but it need
not get in the way of people and companies actually trying to do things
Comments (34 posted)
Jon Johansen, one of the developers responsible for the creation and
distribution of the DeCSS code, has been found not guilty of all of the
charges which had been pressed against him in a Norwegian court. According
to the court, if you buy a film on DVD, you have the right to access that
film, even if you do not use the tools envisioned by the entertainment
industry. In one country, at least, the DeCSS code is legal.
This particular case may not be done yet, since Norwegian law apparently
allows the prosecution to appeal an acquittal. It is, however, a major
victory; the court looked at the fundamental issues and ruled in favor of
Mr. Johansen's acquittal, along with the ElcomSoft acquittal, gives rise to
hope that 2003 may be the year in which the intellectual property takeover
tide is turned. People (and courts) are seeing beyond the piracy rhetoric
and looking at the real costs of increasing power over information. Maybe,
just maybe, this particular power grab can be stopped before it's too
That outcome is far from assured, however. Proposed legislation worldwide
threatens to impose DMCA-like anti-circumvention measures, and the CBDTPA
will certainly return to the U.S. Senate. The entertainment industry is
still flush with money and lawyers, and has shown no signs of changing its
approach; Jack Valenti is still calling
for "speed bumps to keep people honest." A couple of important - if small
- battles have been won, but the real fight is just beginning. As
beginnings go, however, this is a nice one.
Comments (1 posted)
[This article was contributed by LWN reader Joe
There's been a lot of media attention focused on Prentice Hall's plan to
publish books published under the Open Publication License
branded as the "Bruce Perens Open Source
." There's nothing wrong
with that, of course, but what most of the media is failing to mention
is that publishing books under open licenses isn't exactly a
Books published under open licenses of one sort or another have been
around almost as long as Linux. The Linux Documentation Project was
published in a number of forms very early on, including the Linux Bible by
Yggdrasil and the Linux Encyclopedia published by WorkGroup
Solutions. For a while, that was just about the only printed
documentation available for Linux. Other open source titles started to
follow in 1999 and 2000 after Linux started to be viewed as a commercial
opportunity by publishers.
The list of titles available under open source licenses these days is
pretty hefty. This is a list of just a few titles that are relatively
That's hardly a definitive list, there are many more out there. Nearly
every publisher that has dabbled in Linux titles has released a few
books under open licenses. Some publishers have tried to make a fast
buck by compiling open source documentation, others have agreed to
publish original works under open licenses. Some titles have sold well,
and others not so well but the sales figures are more likely a
reflection of the topic or content of the title than the license that
the book is published under. In fact, Prentice Hall has published other
books under open licenses, but with much less fanfare.
The unique thing about Prentice Hall's approach is that it specifically
trying to create a brand centered around books under open licenses.
Bruce Perens told us that Prentice Hall decided to brand the
books with his name because they "felt that anyone could do an Open
Source series, and they needed an additional differentiator. That
differentiator is my leadership of the series, they feel I have
credibility in this space." He says that he's very happy for the
publicity. "I definitely want it. All the publicity that I could get
because it definitely helps the Free Software community for people
outside the community to see that more stuff is being done in the Free
Perens says that the book will be published electronically about three
months after the print versions hit store shelves. The reason for the
lag is to give Prentice Hall time to "saturate the market" with the
print version, to reduce the incentive for other publishers to republish
the same content in print form.
Right now, Prentice Hall has three books available and several more in
the works. Perens says that the company is not putting an upper limit on the
number of titles that they will publish in this series. Authors writing
for the series will be getting the same kind of publishing agreements
from Prentice Hall, including comparable advances and royalties. Perens
has received about twenty or thirty proposals since the series was
announced, and he says he's game for more.
He also noted that the company does not intend to invoke any of the
non-free optional clauses of the OPL, and that they may very well
publish titles under other free licenses like the GNU Free Documentation
With any luck, if Prentice Hall is seen to be successful, other
publishers will follow suit and commit more resources to publishing
titles under free licenses. There are a number of advantages to having
documentation freely available, aside from being able to get the title
for free. Computer publishers are notorious for letting titles go out of
print if the sales aren't up to par, making many good technology titles
unavailable for all intents and purposes.
Publication under a free license
also opens the door for translations of titles that might not
otherwise be produced, and updated versions when the author and/or
publisher has lost interest in a title.
Free software benefits greatly from free documentation. This move by
Prentice Hall is a welcome development in that it should produce more free
documentation for our community. The community must keep in mind, however,
that this sort of experiment will be short-lived if the market for books
collapses. If we want free (as in speech) documentation, we need to put
our money where our eyeballs are.
Comments (5 posted)
Page editor: Jonathan Corbet
Next page: Security>>