The Ubuntu 5.10 release is out, and the initial reviews are good. The
Ubuntu team, however, is not taking time out to drink beer and relax before
pondering its next release. Well, OK, maybe they are taking a little
time. But, when the hangovers wear off, they are still putting some
thought into their next release, which will break some new ground.
Meanwhile, the Debian Project is looking forward to its next release as
well. In both cases, the planning process gives us a hint of what to
expect from these distributions in the near future.
Ubuntu's approach has been to crank out a distribution every six months,
integrating a great deal of bleeding-edge software each time. This process
has been through three cycles now, with obvious success. The next release
(6.04, or "Dapper Drake") will be different, however: Ubuntu has stated
that 6.04 will be supported for three years on desktops, and five years on
server systems. That is quite a promise for such a young company to make,
but, if Ubuntu can live up to it, the popularity of this distribution could
grow. Thus far, five-year support has come with a hefty price tag; the
prospect of free updates from Ubuntu for that long could make a number of
companies wonder just what they are paying for. The fact that Ubuntu's
security response time tends to be excellent can only help in that regard.
All this depends on Ubuntu being able to make a credible promise of
long-term support. This week, Ubuntu's Jeff Waugh took some steps in that
direction with these thoughts on the Dapper
release process. If this proposal becomes policy,
Dapper will, indeed, be a different
sort of release.
The core of the proposed Dapper process is this: the upstream version freeze which
was imposed for the 5.10 release will remain in place. Essentially, the
distribution will be frozen for the next six months, with the bulk of
development effort going into ensuring that it is the most stable,
supportable release possible. Another way of looking at it is that all of
those users happily downloading the Breezy release now get to be the beta
testers for 6.04. This is a major change for Ubuntu, but, as Jeff put it:
We can't just follow the same release process and expect to be able
to ship a long term supportable system. 6.04 will be different, so
we need to think about it differently.
Of course, too much stability would be contrary to the Ubuntu spirit, so
the developers are leaving themselves a bit of room to toss in some newer
packages. So 6.04 will have a few, small upgrades, including:
- GNOME 2.14 (and whatever is the current KDE)
- Firefox 1.5
- The modular X.org 7 release
- OpenOffice.org 2.0
- A newer kernel, probably 2.6.14
The list of exceptions is expected to be discussed at the upcoming UbuntuBelowZero
gathering. The picture coming into focus now suggests that 6.04 will
include some major upgrades, but much of the infrastructural code,
especially that used on server systems, will remain at the version shipped
The Debian Project got its Sarge release out the door last June. By normal
Debian timelines, it is thus quite early to be thinking about pulling
things together for another release. Instead, Debian developers should be
busily testing the patience of sid users by filling it with unstable,
incompatible, major package updates. Well, the developers have indeed been
on top of that task, but release manager Steve Langasek is trying to ruin
the fun with this plan for the next Debian
release, called "etch."
That release will be put together by Steve, along with new co-release
manager Andreas Barth. They have a timeline, which involves a toolchain
freeze at the end of next July, a general freeze in October 2006, and the
etch release is planned, with great precision, for December 4, 2006.
July seems like a distant prospect, but Steve notes that
this deadline does not leave a whole lot of time for big changes:
What's not spelled out in the above timeline is that this basically
leaves people until around the end of the year to to implement any
dastardly plans they have that require sweeping changes to the
archive, followed by another half a year of comparatively minor
changes (you know, the kind that *don't* render half the libraries
RC-buggy in a single upload...)
If this timeline holds, we should see the shape of the etch release by the
beginning of next year. Looking at the current plan, it seems that etch
will have made the switch to gcc 4.0 and (finally) X.org. Another
long-delayed advance will be support for the amd64 architecture as an
official Debian port. Then there is the crucial business of purging
the distribution of non-free documentation, and non-free firmware as
well. Tasks on the wishlist include full SELinux support, a default UTF-8
locale, multiarch support, and more.
The following eleven months of stabilization seem glacial by Ubuntu
standards, but it is an optimistic timeline for Debian. One interesting
change that the project is considering is to continue to allow
non-maintainer updates to all packages throughout the etch cycle. Debian
developers have historically been the lords of their particular bits of
package turf, so non-maintainer updates have always been a sensitive
issue. The release managers believe, however, that non-maintainer updates
speed the release process - and make Debian a better distribution as well.
Both distributions have a lot to gain if they can make their plans stick.
Ubuntu will have produced a stable distribution which it can credibly
promise to support for five years, all while keeping its six-month release
cycle. Debian, meanwhile, will be able to get a stable distribution out in
a timely manner without compromising its high quality standards. In both
cases, the end result can only be good for Linux users.
[Update: Ubuntu patron Mark Shuttleworth has posted his position on freezing for 6.04; he is
inclined to be more permissive - for a while at least - on what gets into
Comments (19 posted)
One problem with governments is that, unsurprisingly, powerful interests
try to direct governmental power toward their own ends. Those who
would fight power grabs quickly learn a hard lesson: those pushing for more
power usually need only win once, while those who oppose them must win over
and over again. This dynamic can be seen, for example, in the current
broadcast flag debate in the U.S. This flag has already been defeated
once, but nobody doubts that it will return, perhaps repeatedly.
In Europe, the debate on software patents is likely to go the same way.
Those who have a substantial amount to gain if software patents are adopted
throughout the EU are unlikely to simply give up just because they lost the
battle last July. So software patents in Europe will almost certainly be
back. Now it is starting to look like the vehicle for the next attempt to
impose software patents might be a process called the "Community Lisbon
This program is part of an effort to improve the health of European
economies by making the EU as a whole more efficient and competitive. It
is a large undertaking touching on many areas, including regulation, internal
markets, environmental issues, global trade agreements and more. Deep
recently-released document [PDF] on the implementation of the program
is a section on intellectual property rights ("IPR"). It reads, in part:
Companies and their clients need IPR which stimulates innovation,
provides a stable context in which to make investment decisions,
and encourages the development of efficient new business
models. The debate engendered by the proposed directive on the
patentability of computer-implemented inventions has demonstrated
that framing IPR rules which balance the needs of all stakeholders
is by no means easy. The Commission will therefore launch a
dialogue with industry and other interested parties in 2006 to
determine what more might usefully be done to provide European
industry with a sound IPR framework.
It is not hard to imagine that the result of this process could be a
renewed directive establishing software patents in Europe. This time,
however, it could be buried within a much larger chunk of EU-level
industrial policy legislation, and, thus, harder to defeat.
Clearly, the free software community needs to be among the "other
interested parties" participating in this process. We have many thoughts
on what makes up a "sound IPR framework," and they should be heard early
on. In the later stages of this program, when it truly comes into public
view, it will be too late to effect changes on issues like patents.
Comments (12 posted)
Back in 1993, Bob Young created a company called "ACC Corporation," which,
among other things, dealt in early Linux distributions. In 1995, ACC
acquired Marc Ewing's Red Hat Linux distribution; the combined company was
then named Red Hat software. Over the coming years, Red Hat would
transform the Linux business environment, become the first Linux-related
company to obtain big-name venture capital, and the first to go public.
Regardless of how one feels about the company or its distribution, it is
hard to deny that Red Hat has had a big influence on the Linux community as
On October 18, Red Hat announced
that Bob Young had resigned from the company's board of directors, with the
intent of spending more time on his other endeavor: Lulu.com. Bob's role in the company had been
shrinking for years; he had not been involved in day-to-day management for
some time. Still, when one thinks of the names involved with the early Red
Hat (Marc Ewing, Donnie Barnes, Michael Johnson, Eric Troan, ...), it
becomes clear that they have all moved on. Bob was the last of the crowd
which helped to set new standards for Linux distributions and showed that
it was possible to build a business around Linux.
Bob's vision was not always perfect - remember that Red Hat went public
with a business plan
stating that its Internet portal was the key to its future
profitability. Still, he clearly got some things right. Seeking an
example of how he saw things in the early days, your editor spent some time
digging through his mailbox. What turned up was this message on how Red Hat chose Linux over
BSD, sent to the free software business mailing list back in 1998. It
makes an interesting read:
When we launched Red Hat Software, Inc, we planned to sell an
operating system. It doesn't take a rocket scientist to recognize
that being in the OS business meant that we were competing with
While our ambitions at the outset were quite limited, we can drink
as much beer as anyone, and on those occasions when our natural
intelligence was at its most limited, we'd speculate on what
Microsoft's reaction would be when we became a real threat.
They concluded that a GPL-licensed system would not be as vulnerable to the
famous "embrace and extend" strategy as a system covered by the BSD
license. Were it not for the licensing issue (and a couple of others,
mentioned in the message) and adequate supplies of beer, Bob and Marc might
just have gone into business with "Red Hat BSD."
Bob has been well rewarded for his role in the creation of Red Hat - he
still owns about 5% of the company, according to the proxy information sent
out for last August's board election. Still, it is worth a moment to say
"thanks, Bob." Linux would certainly have succeeded without Red Hat, but
it would have been a different, and possibly slower, path to success.
Comments (4 posted)
Page editor: Jonathan Corbet
Inside this week's LWN.net Weekly Edition
- Security: A survey of recent kernel vulnerabilities; Color printer watermarks; New vulnerabilities in curl, lynx, netpbm, PHP...
- Kernel: What's holding up 2.6.14; Nested classes and the future of the device model; The ktimers debate.
- Distributions: A Quick Look at SUSE Linux 10.0; new releases from Mandriva Linux, Ubuntu, SourceMage and OpenPKG
- Development: JPPF, the Java Parallel Processing Framework,
open-ezx launched, ModelSecurity for Ruby on Rails, Jini starter kit,
new versions of JPPF, ZODB, Zope, gnormalize, Dropline GNOME,
KDE, Logisim, Open Clip Art Library, Ember, MetalMech, MH-E, DSSI,
AbiWord, DrPython, TestOOB, DWARF.
- Press: Open formats make history, Open-Source and the FCC, the
Free Standards Group LSB Desktop Project, Linux Geek Cruise,
Race to Linux winners, Editing audio in Linux, Neuros' Linux-based
portable media player, National Center for Open Source Policy and Research.
- Announcements: Linspire school discount, SugarCRM Receives $18.77M,
Open Source EHR Katrina Relief Network, Vote Against Software Patents,
Award for Projects of Social Benefit, KDE at German Events,
LCA OO.o Miniconf CFP, PHP Quebec CFP.