NEWS
Newsdesk
■
■
New distros hit the mirrors ■ Competition for VMware ■ OSDL rejects collaboration with Microsoft
GPL 3 cracks down on infringements ■ Xandros founder dies ■ New distro for Asia ■ LPI updates
Linux licence gets tough
Worldwide consultation for GPL 3 launched with suggestion of patent retaliation clause
Developers and lawyers
drafting the new GNU
General Public License (GPL)
are considering adding a controversial
clause to deal with the patent menace
problems expected as Linux gains a
foothold in the mainstream. The so-
called patent retaliation clause would,
in certain circumstances, remove an
organisation’s right to use, alter or
redistribute software available under
“ONE OF THE MOST
IMPORTANT SOCIAL
DOCUMENTS OF OUR TIME.”
the licence.
The first draft of GPL 3 is due to be
published by the Free Software
Foundation (FSF) in January, and the
consultation period for it has already
begun. Joachim Jakobs, head of media
relations at FSF Europe, said it makes
sense to target developers or software
vendors who might use their patent
portfolios to stifle free and open
source (FOSS) projects, but that there
had to be a measure of justice in the
clause. “We don’t want to hinder
people from using free software if they
merely hold software patents,” he told
Silicon.com.
Georg Greve, the president of FSF
Europe, acknowledged that there has
been a lot of potentially damaging
speculation about the patent
retaliation issue, which could scare off
potential developers, vendors or users,
but defended the proposal. “What we
are concerned about is aggressive use
6 LXF73 DECEMBER 2005 www.linuxformat.co.uk
LXF73.news 6
of software patents against free
software authors,” he said
The proposed clause could bear
some resemblance to Section 8 of the
Mozilla Public License (MPL). This
removes the rights granted under the
licence to individuals or organisations
that make patent infringements
against developers and contributors.
“Whether such a clause would make it
into the GPL and how exactly it would
look is pure speculation at the
moment,” Greve said. “Only the first
draft will bring certainty. It is expected
for January 2006.”
Among the driving forces behind
the call for a patent retaliation clause
are the SCO case, in which SCO
launched attacks on the GPL while still
distributing software under the licence,
and fears of the offensive use of
patents by proprietary software
companies to kill competition from
Linux and other projects such as
OpenOffice.org and Openexchange.
Other changes to the GPL could
include elements to deal with DRM
technology and the use of web
applications (such as those used by
Google and Amazon), which are
transforming the way that end users
access software.
The FSF has initiated a funding
drive to raise $500,000 to bankroll
the consultation programme, whose
23/9/05 1:05:20 pm

Exciting times for Linux distros
Install freaks: brace yourselves for an exciting few weeks...
Inveterate upgraders are
bracing themselves for a new
round of distribution releases
featuring cutting-edge desktop
environments, the latest versions of
OpenOffice.org and Firefox and some
innovations designed to push Linux as
a genuine alternative to Microsoft’s
newly christened XP follow-up, Vista.
The arrival of Vista offers Linux an
“unrivalled window of opportunity”
according to Novell CEO Jack
Smart is like Synaptic, and improves package management in Mandriva.
scope will be unprecedented. The GPL
Version 3 Development and Publicity
Project will, according to FSF executive
director Richard Brown, reach out to
development communities across the
world. “We aim to increase the
international reach of the free
software movement. To develop this
new licence, we will be contacting
communities across the globe to
ensure their participation in the
update of one of the most important
social documents of our time.”
Global involvement
In an interview with Linux Format
published this issue, GPL lawyer Eben
Moglen said the FSF is expecting as
many as 150,000 people and 8,000
organisations to have an opinion about
how GPL 3 should look. “It’s going to
be a very large conversation,” he said.
The current version of the GPL was
drafted in 1991 (see GPL In A Nutshell
box, right) and, Brown says, has
LXF73.news 7
Messman. He told delegates at the
company’s European BrainShare
conference that the cost of migrating
from Windows XP to Vista was certain
to be significantly higher than that of
migrating to Linux, and so the launch
of Microsoft’s next generation desktop
would provide an opportunity for Linux
vendors to make a major push into
businesses that use high numbers of
computers, for whom migration costs
would be multiplied.
become central to the activities and
operations of companies and
governments all over the world. But
after 14 successful years, the licence is
in need of a review to reflect the
growing use of free software in both
institutional and commercial situations.
The Version 3 Development and
Publicity Project was kicked off with a
150,000 donation from Stichting
NLnet, a Dutch non-profit organisation
that supports research and
development in internet technology.
GPL software development is a core
part of the organisation’s work.
Teus Hagen, chairman of NLnet,
said: “Our support of the GPL and the
Free Software movement, and its
desire to see the successful adoption
of GPLv3 achieved in the spirit of
internationalisation, made the funding
of this project an important priority for
us. We hope to encourage other grant
making organisations to help fund this
unique project.”
www.linuxformat.co.uk
Novell’s future strategy will focus on
the fruits of its OpenSUSE project,
which will make its first ‘official’ release
in early October. SUSE 10.0 features
the latest Gnome 2.12 desktop as well
as KDE 3.4, and boasts excellent
Mono/Beagle integration (as you’d
expect from the owner of Ximian). In
tandem with the release of SUSE 10.0,
the first alpha version of 10.1 should
be available at the same time.
New Mandriva release
Mandriva’s first, and possibly only,
release candidate for its 2006 release
has also hit the mirrors. As expected,
this release is the first to feature large
elements of the Conectiva distribution,
including the Smart package manager.
By default, the distro still uses URPMI,
but Smart is available as an option.
Another interesting addition is the Kat
desktop search tool, which brings
Spotlight/Beagle-style search facilities
to the KDE desktop.
Barely a month after the release of
Gnome 2.12, the Ubuntu project and
its Kubuntu sister have released the
Breezy Badger distro which, like its
predecessors, swathes the user’s
desktop in tasteful, humanist browns.
Meanwhile, Foresight Linux is
claiming to have beaten Ubuntu to be
GPL IN A NUTSHELL
The GNU General Public License was
written by Richard Stallman in 1988.
The catalyst was his experiences
developing the Emacs text editor
and his legal tussles with a company
called UniPress. In May 1983 UniPress
acquired an application called Gosling
Emacs from James Gosling. Gosling
Emacs was based on Stallman’s original
Emacs application written in 1975, and
the source code was publicly available.
But when UniPress purchased Gosling
Emacs, Stallman was forced to remove
elements of the Gosling code from his
version of Emacs and replace them with
his own code.
From this argument, the Emacs
General Public License emerged in
February 1988. Other application-
specific versions of the licence were also
published, until 1989 when the licence
was changed to make it applicable to
any application. In 1991 the licence was
updated to version 2 to clarify issues
NEWS
Foresight Linux uses Anaconda as
its installer, but uses a distributed,
cross-repository system for
package management.
the first linux distribution based on
Gnome 2.12 Version 0.9 of Foresight
features the new desktop complete
with support for the Beagle desktop
search tool, Cowbell, F-Spot and a
Sudoku package. The latest version
also swaps out Rhythmbox to make
Banshee the default media player.
Foresight’s ethos is based on
choosing the right tool (and only one
tool) for a specific job rather than
populating menus with several options
that all cover the same few needs.
Foresight Linux is also interesting in
that it uses the Conary distributed
package management system rather
than relying on RPM or APT, and the
Howl networking system.
■ Look out for a new Distro Watch
section in LXF74!
surrounding its use in libraries,
permitting ‘linking the libraries into
proprietary executables under certain
conditions’. This version has been in
place for 14 years.
Broadly the GPL gives users the right
to acquire, alter and redistribute
software available under the licence,
with the caveat that any changes made
and (most importanly), released to the
public, are done so under the same
liberal licence. The source code to such
‘derivative works’ should also be
publicly available. Therefore, if a
company decides to implement a MySQL
system and makes changes or
improvements to the software for
internal use only, they are not obliged to
share them with the community.
However, should they later decide to
market their MySQL-based software,
they would need to comply with the
licence and publish the changes or risk
legal action.
LXF73 DECEMBER 2005 7
23/9/05 1:05:23 pm
 NEWS
MICHAEL
MEEKS
Michael works for Novell
full-time on developing
OpenOffice.org, and
dislikes herring.
COMMENT
A better
Gnome
I was particularly
pleased, while playing
with OpenSUSE’s Gnome 2.12, to
see some of the new features.
Firstly, the use of an
mmapped icon data cache that
saves memory and improves
performance, particularly
application startup time. Each
process uses far more virtual
memory with the cache, but that
increase is all shared, giving a
nice reduction in the real
memory used. Of course, the VM
can also be more aggressive
under pressure with nicely laid
out disk-backed memory.
The same technique is being
extended to fontconfig, which is
another memory hog used
across the Linux desktop. It’s
also great to see the results of
the Google-funded optimisation
bounties starting to appear,
which will give Gnome 2.14 yet
further improved performance
and memory consumption.
Secondly, it’s great to see
the system events layer and
volume monitor working well.
System events, such as
hotplugging USB devices, are
now handled in a consistent,
responsive and easily
configurable way that can be
locked down – and users will
feel the benefits right away.
This release added a large
amount of new functionality
under the hood, such as Cairo,
but we are still sticking to our
ABI guarantees. Gnome 2.12
forms the seventh release in
under four years of major
improvement that hasn’t
significantly altered the ABI for
ISVs. What that means is that my
C/Gnome binary I compiled on
Red Hat 8.0 in 2002 will still run
on my OpenSUSE 10.0
today – nice work.
8 LXF73 DECEMBER 2005
LXF73.news 8
Virtual space gets bigger
Xen-like calm disrupted by Virginian start-up
A new company has entered
the Linux virtualisation
market, launching a desktop
product that, it claims, is comparable
with VMWare and other systems but
retails at half the price. Parallels
Workstation 2.0, which is set for an
October release, and will be available
on both Windows and Linux.
The Virginia-based company behind
it, Parallels Inc, is unusual in the
virtualisation market for planning to
target enterprise as well as home
users. “As computers become more
and more powerful, virtualisation will
grow in importance,” explained
Parallels spokesman Benjamin Rudoph. two operating systems – Windows and
“Users at all levels of ability and
savvy will want to maximise their
machine’s potential so they can work
more efficiently and more powerfully,
while enterprises will want to curb
hardware costs, lower overall operating
expenses, and efficiently support their VMware strikes back
IT infrastructure. Virtualisation
addresses all of these concerns.”
Parallels even has hopes for use of to offer early access to its next
its software in education, to help
create student-friendly teaching
environments. “Parallels Workstation
Asianux unites three nations
Governments get involved in pan-Asian distro programme
In a startling example of
regional cooperation, the
governments of Japan, China
and South Korea have joined forces
to promote the benefits of Linux with
the release of Asianux 2.0, a server-
only operating system aimed at
www.linuxformat.co.uk
enables instructors to create self-
contained operating systems in which
users can work with multiple operating
systems and software sets
simultaneously,” said Rudolph. “Virtual
machines can be programmed to
automatically reset at shutdown,
ensuring that students will have a
clean, error-free environment to work
with each time they come to class.”
Parallels’s decision to offer Linux
support was linked, Rudolph said, to
the company’s wide target audience.
“Linux continues to grow in importance
and in use. It was important to us to
build a product that worked with the
Linux – that the majority of the world
uses personally and professionally.”
A beta release of the Workstation
application is available now, and a
server version is due early next year.
Not to be outgunned, VMware has
launched a beta programme of its own
generation application, which adds
support for 64-bit operating systems
and dual-processor systems.
enterprise users. The distro itself is
being rebranded and localised by
companies in the three countries
involved in its development.
The distribution has been developed
with government assistance by Red
Flag from China, Japan’s Miracle Linux
and Haansoft Inc from
Korea and will be
released as, respectively,
Red Flag DC Server 5.0,
Miracle Linux V4.0 and
Haansoft Linux 2006
Server & Server 64. It
follows version 1.0,
which was launched in
June 2004 and has
already become one of
Asianux has ambitions
to become the leading
distro in east Asia.
VMware finally has some real cross-
platform competition.
As well as opening up to new
architectures and platforms (including
BSD and Ubuntu), the new version
of VMware 5.5 can access and boot
up disc images saved in Symantic’s
Ghost format without changing the
underlying file structure.
Curiously, VMware is claiming that
users will be able to run a 64-bit
guest on a 32-bit host. For example,
users will be able to run 64-bit Linux
or BSD on a 32-Bit Windows XP host.
the top three Linux products in the
region, according to Chris Zhao, Red
Flag’s president. Zhao is confident that
by improving the technology, sales
presence and user support for version
2.0, Asianux can build on that user
base. “As a next step, we will recruit
more partners to join Asianux so that
we can expand the market for the
products based on it quickly,” he said.
The Asianux partners are also
hoping to make inroads into other
markets around the world traditionally
served by the likes of Red Hat, Novell
and Microsoft.
Project president Jong Jin Baek said
that a partnership was being forged
with an Indian company. “With India as
a new partner for the Asianux project,
we will have the most powerful Linux
platform in Asia,” he said.
23/9/05 1:05:26 pm

Keep music Live
Noodler-friendly distro Studio To Go gets a remix
Studio To Go, the Live CD
distro that provides a
complete audio studio on a
bootable CD, has had its second major
upgrade since its launch last year.
Version 1.50 of the audio suite has
been released in tandem with a
comprehensive guidebook covering
the use of the included Rosegarden
Midi and audio sequencer.
In addition to the sequencer
software, Studio To Go includes a
selection of VST plugins, a complete
JACK audio system and a range of
applications including Audacity and
Ardour, the audio editing environment.
The complete Live CD package is
available with the Rosegarden
companion book and a USB stick to
store compositions for £79.99 from
www.ferventsoftware.com.
OSDL says “no” to
joint MS study
We don’t want your steenking money!
Stuart Cohen, the director of
the Open Source
Development Labs (OSDL),
has ruled out the possibility of a joint
Total Cost of Ownership (TCO) study
with Microsoft. Microsoft executive
Martin Taylor approached Cohen after
the LinuxWorld Expo and Conference
in San Francisco last month with an
offer to jointly fund and publish an
independent study on the relative
merits of Linux and Windows.
But Cohen fears that bias would
appear, regardless of the results in an
objective report. Microsoft’s long
running ‘Get the Facts’ campaign has
been accused of skewing paid-for
research in its favour regardless of the
true outcome, which Microsoft has
denied. The OSDL (among others) has
LXF73.news 9
Studio To Go includes an enormous number of noisy applications.
faced calls to set up a Linux
equivalent to the campaign to
address the perceived bias in
the mainstream press.
The latest in a sea of Total
Cost of Ownership (TCO)
studies suggests that Linux is
40% cheaper to maintain than
Windows. The study, funded
by IBM and undertaken by
the Robert Frances Group,
looked at application servers
and found that businesses
running Linux found savings
of 40% over Windows and
54% over Solaris.
The OSDL was concerned
that Microsoft would insist upon a
short-term study emphasising
migration costs from Windows to Linux.
www.linuxformat.co.uk
Could a study involving the world’s
second-biggest brand ever really
be impartial?
NEWS
■ Last Man Standing is a new
Doom3 mod designed to breathe
life into an old friend. The game
comes into its own when used in
KillFest mode, when the intention
is to become the last man standing.
The multi-platform mod is available
from http://lms.d3files.com.
■ A battle has broken out
between the local authorities in
Massachusetts, who want to
mandate the use of “open standards
and formats” for its documents, and
Microsoft, which sees this move as a
violation of its right to compete in the
office suite sector. The Massachusetts
government will soon default to the
Open Document formats currently
supported by OpenOffice.org and
others but Microsoft says the state
should standardise on its “open”
format, which uses standard XML at
its core but has licence restrictions
that ensure competitors won’t be
able to use it.
■ SCO posted a loss of $2.4 million
for the second quarter of 2005. The
loss was blamed on a drop in profits
from the core UnixWare product;
its failing SCOSource programme,
launched with the intention of
convincing Linux users to insure
themselves against future litigation;
and legal costs of $3.1 million.
■ In some better news for the
company, SCO has signed a deal
with MySQL AB to jointly deliver
a certified edition of the open
source database on the company’s
OpenServer 6. MySQL’s Steve Curry
from MySQL AB said the deal
didn’t amount to an endorsement
of SCO’s business or legal activities:
“We appreciate the sensitivity of this
issue and did not make this business
decision lightly.”
■ The great Munich migration to
Linux has missed its deadline, with
administrators and planners wanting
more testing before they roll out the
new Linux and OpenOffice.org system
to users across the city.
■ Schoolchildren the Italian region of
Bolzano returned to their classroom
after the summer break to find their
computers upgraded and running a
combination of Debian and Gnome.
The FUSS Project saw the migration
of over 16,000 school computers
from Windows XP to Linux.
■ Sun’s on-off affair with Linux is
back on. The company has extended
support to Red Hat Enterprise
server on its newly-launched range
of x86-based Galaxy servers.
LXF73 DECEMBER 2005 9
23/9/05 1:05:31 pm
 NEWS

LPI wants your help Richard

Cobbett
Florentine telescope

to update courses
builder, physics pioneer
■ The Mozilla Foundation has and church worrier. Or
launched a new commercial was that Galileo?
arm to raise cash to assist in the
development of the Firefox browser.
COMMENT
■ Meanwhile, according to some

It’s show
estimates, Firefox is set to approach
Exam update will reflect real life
a market share of 10% by the end
of the year.
The Linux Professional The public is being asked to
■ Wikipedia has begun using
Mono to provide search services on
its main encyclopaedia.

■ SCO has rediscovered its roots as

a software company, and updated its
flagship UnixWare product.
■ Win4Lin has made a
breakthrough in its core technology
by getting Windows 2000, XP
and Server 2003 to run using the
Xen hypervisor.
■ After months of work, Linus
Torvalds has migrated the
entire Linux kernel to its new Git
management system. The ‘directory
content management’ system has a
new site at http://git.or.cz.
■ A new version of MEPISLite has
been released. The lightweight distro
has dropped OpenOffice.org in favour
of KOffice.
■ Recent open source convert
Computer Associates is following
in the footsteps of IBM, Sun and
Nokia in releasing a portion of its
patent portfolio to developers.
■ Researchers at Cambridge
University have discovered that
Gimp, the open source image
editing application, is quite adept
at revealing the contents of secure
letters that are sent out by banks
to accompany credit or debit cards.
Researcher Mike Bond said he first
became suspicious after being able
to read his own PIN through the
‘secure’ envelope thanks to a poor
printing process. Images from these
envelopes scanned into Gimp would
reveal their contents with a few
minor adjustments.
COMING ATTRACTIONS...
★ International PHP Conference, 6–9
November, Frankfurt am Main, Germany
Four days of database management,
security, application design, GTK
integration and more. PHP bliss!
The gathering will begin with a
programming fest and speakers will
include Georg Richter from MySQL AB
and Jeremy Johnstone, founder of
Hotscripts.com and current Yahoo
staffer.
www.phpconference.com
★ Forum PHP 2005, 9–11 November,
Paris, France
It’s a busy month for PHP developers:
Institute (LPI) is seeking
assistance in building a new
programme of courses and exams to
better reflect the reality of working
with Linux. The organisation says that
it relies upon community participation
to ensure its exams are at the cutting
edge, especially when it comes to
deciding on the technical objectives on
which the exams are based.
For the current round of
consultations, the LPI has started a
wiki dedicated to the process and is
inviting users, developers and
sysadmins to submit their comments
to the site.
The LPI’s exams are intended to reflect real-world applications of Linux.
first the shindig in Frankfurt, then this
two-day event in Paris targeting
technical and business issues. The
programme was not confirmed as we
went to press, but it will, the organisers
say, feature the crème of French PHP
talent.
www.afup.org/article.php3?id_
article=220
★ Open Source Developers’ Conference,
5–7 December, Melbourne, Australia
Kernel hackers get a chance to work on
their winter tans down under (that’s
Australia, not a euphemism) at the
annual OS Developers’ Conference. The
comment on the internal review
process, which was completed at the
end of June 2005. The LPI is
accepting comments at https://group.
(girls) time
lpi.org/cgi-bin/publicwiki/view/
Examdev/ObjectivesReview until
the end of February 2006. New
objectives will come into effect in April
and May of next year for LPIC-1 and
LPIC-2 respectively.
The LPI has already given an early
hint of the changes that might take
place: as of October, candidates sitting
the LPCI-1 must answer questions on
both RPM and DPKG package
management systems.
call for papers suggests a strong interest
in languages such as Perl, Python and
Ruby this year.
www.osdc.com.au
★ ApacheCon US 2005, 10 December,
San Diego, USA
The organisers of ApacheCon are sifting
through the 200 proposals submitted by
eager speakers. These are for the
conference, but the event will also
feature presentations from the likes of
IBM (which is running a Glucode
installfest), Sun Microsystems and the
Java Community Process.
www.apachecon.com
Red alert. Red alert.
Microsoft has a whole
redesign planned for its Office
suite, mixing a healthy dose of
Apple-style brushed metal with
a dollop of whimsy and/or crack
to produce an interface that you
may love, you may hate, but you
certainly won’t recognise. This is
it, folks. This is the Death Star’s
exhaust port, Achilles’ heel and
Paul Verhoeven’s Showgirls, all
rolled into one. Why? Change.
The fear of change.
Remember all those times
you’ve tried to get people to use
the latest open source tool, only
for its unfamiliarity to scare
them away? Just look at the
early Office12 videos. That
argument is gone. Indeed, for
the briefest sliver of time, all the
cards are in your hands. For once,
you’ll be pushing software that
looks similar enough to current
systems for your friends, family
and co-workers’ comfort, with
feature sets powerful enough to
tempt them away from 1999
forever. After all, if they’ve got to
learn a whole new way of
working anyway, it makes sense
to try that free open source
thingy before shelling out
hundreds of pounds on a whole
new OS and office suite.
The key is to make sure that
all the software is ready to go in
time – complete, usable, friendly,
and above all else, BS-free.
Forget betas, forget alphas, and
for the love of all things, forget
‘when it’s done’. The only way
that Linux can hope to have a
decent showing in the home is to
get in there at the changeover
from XP to Vista. Open source
software is often accused of
being behind the big names. It’s
time to turn that
accusation into a weapon.

www.linuxformat.co.uk LXF73 DECEMBER 2005 11

LXF73.news Sec2:11 23/9/05 1:05:34 pm

 NEWS
David
Cartwright
David Cartwright is an IT
consultant who specialises
in providing Linux Systems
and solutions.
COMMENT
Mono mania
As a fair number of our
readers no doubt already
know, the Mono project (www.
mono-project.com) has made
incredible progress with its
attempts to produce an open
source version of the .NET
development platform.
With the introduction of
Windows Vista, however,
Microsoft has raised the bar for
the open source community
simply by producing a new pile
of concepts (most notably the
new WinFX architecture) and
code that open source
developers really need to
emulate to keep up with MS.
The push within the Mono
development community is a
sensible one: as time and
resources are limited, they’re
spending their time
implementing the stuff that
people (ie application
developers) actually want. Even
with the existing .NET
specification it’s been decided to
leave out some of the more
esoteric things that are kind of
nice but, in reality, rarely used.
This approach will undoubtedly
be extended to WinFX as and
when Microsoft’s implementation
becomes less of a moving target
and the open source developers
can see what they need to do.
Mono has done an
admirable job so far, and I
sincerely hope that this will
continue. Credit must, of course,
go to commercial entities such as
Novell, which has put money
into the project and includes
Mono with its Unix packages, but
we should all give at least a
passing thought of gratitude to
the band of Mono developers
who are continuing to slog along
what must seem an
endless road.
12 LXF73 DECEMBER 2005
LXF73.news Sec2:12
Frederick H Berenstein
1946–2005
Xandros founder dies aged 59
Dr Frederick H (Rick)
Berenstein, the co-founder
of the Xandros distribution,
has died in his New York home from
cancer of the brain. He was 59.
Berenstein was one of the driving
forces in ensuring the acceptance of
the open source development model
among the reticent investors of Wall
Street. Before launching Xandros, he
made significant investments in start-
ups such as Ximian and CodeWeavers
and founded Linux Global Partners to
incubate new Linux projects and
companies. He also used his influence
and experience in the free software
community to encourage investment
among other venture capitalists.
“HIS BELIEF IN LINUX
WAS INFECTIOUS AND
HARD TO ARGUE WITH.”
In addition to being a canny
entrepreneur (who could have been a
concert pianist if his mother had
allowed it), Berernstein was an
accomplished software engineer who
EMBEDDED LINUX NEWS
● PalmSource has been acquired by the
Japanese company Access for $324.3
million. Access is known in Japan for its
NetFront embedded web browser and is
expected to build on Palm’s transition to
Linux. Access said it intends to port
NetFront to the Linux platform and expects
a first release of the new system in the
summer of 2006. The acquisition
means that most of the major players
in Japan’s enormous mobile phone
market – including leader NTT
DoCoMo – now have a significant
investment in the success of Linux in
the embedded space.
● Trolltech, the developer of the Qt
toolkit which underpins KDE, is
preparing to go public on the back
of its successes in the embedded
www.linuxformat.co.uk
started his career
programming in Fortran in
1965, and was responsible
for the design of some
landmark applications –
including Trans World
Airlines’ pioneering
automated reservation
system in 1985.
Xandros’s chief
executive, Andreas
Typaldos, said in a
statement that
Berenstein had a passion
for Linux and a strong
belief in the “inevitability”
of Linux. “Beyond Rick’s
vision, inspiration and
will, which enabled him to fight his
illness, that also inspired Rick to push
Linux forward, ahead of most other
people, through visionary investments
in companies such as Ximian,
CodeWeavers, and finally Xandros.
“His belief in the inevitability of Linux,
from the days when I first met Rick in
leadership, which our industry, our 1999 as one of his founding group of
company and all of us will miss, I will investors in Linux Global Partners, the
personally miss his example of quiet Linux incubator that he had co-
courage, and his indomitable will and founded with William Roseman, was
determination. “It is that courage and infectious and hard to argue with.” LXF
space. The company recently divorced itself company plans to sign partnership
from the Canopy Group – which also owns agreements with “four or five start-ups.”
SCO – and secured major funding from Panasonic has an interest in MontaVista – it
venture capitalists. Trolltech’s Qtopia software took a stake in the company in 2002 –
and application stack can be found on a and is expected to use MontaVista’s newly
range of PDAs and, increasingly, on Linux- released embedded OS as a standard for
based smartphones.
future development. Start-ups interested in
applying for support from the fund should
● Panasonic has established a new venture visit www.vc.panasonic.com.
and collaboration fund
targeting start-ups using ● Toshiba has joined forces with Sigma
Linux in the consumer designs to develop a Linux-based platform
electronics sector. The new for the next generation of smart DVD
Technology Collaboration players. The new reference design, built on
Centre will be opened in San a combination of Toshiba’s TX4938 Mips-
Jose, California and the based TX System Risc microprocessor, and
Sharp’s Zaurus the Sigma EM8620L digital media
has successfully processor, runs MontaVista’s OS and will be
used in DVD devices, media players, IP-
used Trolltech’s
Qtopia software. enabled set-top boxes and other devices.
23/9/05 1:05:36 pm
 NEWS EXTRA
Trademark trumps
Another Australian soap opera
Graham Morrison reports on a trademark action that has left a bad taste in the mouth of many free software users
At the end of July, the Perth-
based lawyer Jeremy
Malcolm, a committed
advocate of open source, contacted
around 90 Australian businesses that
use the word ‘Linux’, either within their
trading name or as part of the name
of any products they offer. Under the
title ‘Use of Linux® Trade Mark in
Australia’, Malcolm’s letter starts: “I act
for Linus Torvalds, and I am writing to
you about your use of the trademark
Linux. This is not a letter of demand,
but rather a request for your assistance,
and an attempt to inform or remind
you of the rights and obligations.”
The letter went on to explain that
the Linux Mark Institute (LMI) in the
United States had been appointed the
worldwide exclusive licensee of the
Linux trademark, a move instigated by
Linus himself to protect it from misuse.
The addressees were told it is their
legal responsibility to obtain a licence
from the LMI before using the word
Linux as part of a product, service or
brand. The purpose of the letter was
to pave the way for registration of the
trademark in Australia.
Shocked that the action was
“WE WANT TO ENSURE
THAT THE WORD LINUX
WILL NEVER BE TAKEN
AWAY FROM THE LINUX
COMMUNITY” PIA WAUGH
attributed to Linus Torvalds, many
people also felt that the legalese of
the letter was reminiscent of the
recent SCO debacle, as it seemed to
coerce people into handing over
money for the use of the Linux
trademark. This appears to fly in the
face of free software: surely if GPL
software can be left to fend for itself in
an uncompromising world, went the
objections, so too can the name under
14 LXF73 DECEMBER 2005
LXF73.xtra 14
which much of it is released. There Linux trademark for themselves,” president of Linux Australia. “We were
was an explosion of dissent on explained Towns. “Our aim has been directly threatened and we looked at
Slashdot and other internet sites, and to register the trademark and hand it how such actions might affect our
much confusion caused by ‘free’ not over to Linus so that we don’t end up community. Imagine if someone
seeming to be all that free after all. with different rules applying to the use trademarked the word ‘Linux’ and
of the word ‘Linux’ in Australia as then banned all user groups from
Under threat opposed to the rest of the world.” using it? Such issues are not
The organisation sponsoring Jeremy The LMI agreed that Linux Australia impossible, and thus we wanted to
Malcolm’s actions is Linux Australia, a should handle the issue in the country. extend protecting ‘Linux Australia’ to
group of volunteers formed to cater to “The motivation behind the action is ‘Linux’ and have done the process
the needs of the Australian Linux pretty simple,” said Pia Waugh, vice according to the Linux Mark Institute.”
community. Linux Australia has been
given the task of governing Malcolm has Licence confusion
won an award
the use of the ‘Linux’ for his work Before June, a perpetual licence was
trademark in Australia by for Linux in available from the LMI relatively
the LMI, which works Australia, but cheaply, and the LMI could not
directly under Linus his letter generate enough money to actively
annoyed many.
Torvalds’ jurisdiction. protect the trademark. Changes in the
Anthony Towns, secretary
for Linux Australia, said
Australia was behind other
countries such as the US,
where the trademark was
already registered. “My
understanding is that this has
been an ongoing issue since around
mid-2003, since the Linux Australia
organisation has begun to get itself in
order enough to do more than just act
as a legal body for Australia’s national
Linux conference [linux.conf.au],” he
told Linux Format.
Serve and protect
Linux Australia came into contact
with the LMI when its own name
was threatened. “[We received]
advice from IP Australia that a
company in Adelaide were calling
themselves ‘Linux Australia, Pty
Ltd’ and trying to register the
www.linuxformat.co.uk
22/9/05 9:48:24 am

RMS SPEAKS!
We asked Richard Stallman, founder of the GNU Project and free
software advocate, how the trademark wrangle was likely to
affect the perceived openness and freedom of GNU/Linux, and
how businesses may approach using the word ‘Linux’ in the
future. In particular, how did he feel about Jeremy Malcolm’s
distinction between subverting IP into creating ‘copyleft’ and the
lack of a similar space for the Linux trademark?
“That concept of free doesn’t apply
to the issue here, since it affects none
of those things, but only the question
of how you can publicise your
commercial activities. I am not a
lawyer, but as I understand it the
issue at stake here is not whether
you can distribute the sources of the
kernel, Linux, in a directory called
linux; and it is not whether these
source files can mention the name
‘Linux’ in their comments. It is purely
and simply about the name used for
a product or service. So if you want
to distribute a GNU/Linux system and
call it ‘Foobar GNU/Linux’ or ‘Foobar
Linux’, you have to pay – but if you
call it just ‘Foobar’ or ‘Foobar GNU’,
you don’t have to pay.
The freedom to share and change
software is important for specific
“SLASHDOT USERS MAKE
JERRY SPRINGER’S
AUDIENCE LOOK LIKE THE
VATICAN COUNCIL.”
licensing model will now generate
enough funds to globally police it. A
side-effect of this change is that it’s no
longer clear if you can use the word
Linux purely descriptively, as with
Debian GNU/Linux. Recipients of the
letters were also alarmed at the
prospect of having to pay to use the
word Linux, although the letter did
promise that the LMI would “waive”
any previous trademark “infringements”.
Pia Waugh denied that Linux
Australia is motivated by money. “In
fact, all trademark rights will be
transferred to the Linux Mark Institute
for protecting,” she said. “This is the
institute that Linus himself has
entrusted to protect the name.
“We want to ensure that in an age
where anyone can hold patents,
trademarks or whatever else over the
heads of people interested in freedom,
that at least the word Linux will never
be taken away from the Australian
Linux community or businesses.”
LXF73.xtra 15
reasons. The freedom to call your
product ‘Foobar Linux’ instead of
some other name is not very
important. Most likely Foobar is
actually a variant of the GNU
operating system, and calling it
‘Linux’ is a mistake.
The issue of free software relates
primarily to copyright law. This issue
is about trademark law. As Mr
Malcolm says, copyright law and
trademark law are quite different.
One reason that people tend to
confuse them is that the fashionable
term ‘IP’ encourages such confusion.
This is one of the reasons people
should avoid using the term. When
Mr Malcolm uses that term, he
perhaps unknowingly encourages the
same confusion that he is trying to
clear up.”
JEREMY MALCOLM
Linux Australia has been liaising
with Jon ‘Maddog’ Hall, executive
director of the advocacy group Linux
International. He told us he expects an
increase in the number of trademark
battles as free and open source
software becomes mainstream. “If
projects such as Linux, GNU,
OpenOffice and others do not protect
their names, they will find that people
are ready and willing to misuse them,”
he said. “It is a sad fact of life. I think
that when people get used to it they
will realise that the [Australian]
tempest in a teapot was just that.”
The “tempest in a teapot” phrase
was coined by a Groklaw reporter.
Malcolm in the middle
Linux Australia has also been criticised
for its use of Jeremy Malcolm, a lawyer
who was previously known for his use
of the DMCA against Google and for
his association with the Church of
Scientology. But the committee at
www.linuxformat.co.uk
Maddog Hall:
“People are ready
and willing to
misuse names.”
Linux Australia say they are
satisfied with his actions, even if
they’ve been surprised by the level of
interest his letters have attracted.
Those letters (which have reportedly
received a minuscule 10% response)
are themselves part of the trademark
registration process in Australia, as
feedback can be used to support the
case that Linux is still a well-defined
term and eligible to be protected with
a trademark.
Anthony Towns accepts that the
letters confused a number of people,
and that some of the criticism on how
they were phrased and delivered was
“well deserved”. Follow-up emails from
Linux Australia’s president and other
committee members have apparently
overcome much of the confusion.
“No other solution”
When asked if legal measures were
the only option for protecting the
Linux name, Malcolm said: “In an ideal
world, we would just rely on the
goodwill of the Linux and open source
community to ensure that the integrity
of the trademark is maintained. History
has demonstrated that this is not
sufficient, so we are forced back on to
legal solutions.”
He went on to say: “With copyright
law, we have been able to subvert the
IP regime to some extent by creating
‘copyleft’. There isn’t really an
analogous way of creating a more
++++ BREAKING NEWS ++++
The Australian Trademark authority has
made an initial determination on the
status of the Linux trademark, and has
specified that the current supporting
material provided by Jeremy Malcolm is
not enough to uphold the LMI’s claim
for protection.
NEWS EXTRA
friendly trademark regime for the
Linux and open source community.
So I don’t see any other solution
besides enforcing the use of the trade
mark and using licensing fees from
commercial users to support that.”
Malcolm also told us the reaction to
the letters was actually mixed, and
blames most of the hysteria on the
typical Slashdot user, who he
describes as creating so many
overblown rows and flame wars over
the years that “they make Jerry
Springer’s audience look like the
Vatican Council.” He insists that the
letters were not letters of demand,
and couldn’t have been interpreted as
such by any reasonable reader.
This action is the direct result of a
process that Torvalds himself initiated,
and he shares at least some of
Malcolm’s disquiet at the reactionary
nature of the Slashdot crowd. In a
recent post to the Linux Kernel Mailing
list (LKML), Torvalds said he was
concerned that the Linux community
would lose control of the name. “You
can call anything ‘MyLinux’, but the
downside is that you may have
somebody else who did protect
himself come along and send you a
cease-and-desist letter.”
It seems that as far as Linus is
concerned this is an open and shut
case, and as the owner of the
trademark, he is in the right. LXF
This is not unusual in such
applications, and does not mean that
the LMI will not ultimately have
trademark protection granted. However,
some existing trademarks incorporating
the word Linux already exist, which may
complicate the proceedings.
LXF73 DECEMBER 2005 15
22/9/05 9:48:27 am
 REVIEWS Mandriva Linux 2006

LINUX DISTRIBUTION

Mandriva Linux 2006

Graham Morrison hopes for fireworks from the Mandrake/Conectiva factory, but fears a damp squib.
KEY STUFF
The first proper release from the
new regime at Mandriva. See also:
SUSE 10.0 or Red Hat Fedora.
■ DEVELOPER Mandriva
■ WEB www.mandriva.com
■ PRICE: Free download edition

When Mandriva last released

an update of the popular
Mandrake distro, opinion in
the office was mixed (see LXF67 from
April). Mandriva Linux Limited Edition
2005 was a strong distribution; its
excellent configuration and package
management tools still made it a
contender for our top distro. But there
was something missing: Mandrake’s
trailblazing history of squeezing the
latest releases on to the disc was
replaced by sober consolidation with
Conectiva, and a new, mature outlook
both for the packages included and
for the company’s business strategy.
Mandriva promised a new version
for the final quarter of 2005, and sure
enough it has delivered an optimistic
and forward-thinking distribution in
Mandriva Linux 2006. There will be a
downloadable version, as usual, but
Mandriva really wants users to pay to
join its club, to be able to download
this version early and also take
advantage of an extra disc of packages.
(Almost) infallible
When it comes to ease of installation,
Mandriva can retain its seat close to
the top of the table. While not much
has changed, it still features one of
the most infallible installation routines
we’ve come across. It can deal with a
wide variety of hardware, and removes
much of the guessing that used to be
the mainstay of Linux installation.
Even more impressive is the greatly
improved speed of booting your
system: less than a minute on our
Pentium 4, 2.2GHz, 512MB RAM
machine, and 20 seconds of that was
for KDE. Mandriva has worked a little
magic behind the scenes, even if the
user interface doesn’t show it.
With such an impressive
improvement in the boot speed, you
might expect that changes have been
made at a low level, which should be
most evident in Mandriva’s
For the new release, Mandriva has added KDE 3.4, a new Socialist-Realist penguin… and very little else.
configuration panels. But the most
important addition we could find was
the provision for setting up a DVB
network, which is interesting if a little
niche. DVB networks use a data layer
embedded in the MPEG-2 transport
stream transmitted from a satellite.
This provides a high bandwidth
downstream connection, while a
terrestrial low bandwidth connection is
used for upstream data. It’s a
technology mainly of use to people
working in very isolated locations.
Into the details
First, the good news. The most serious
omission from the previous release
has been rectified with the inclusion of
KDE 3.4.2. Mandriva can now stand
alongside SUSE and Fedora; and
because it’s taken up KDE 3.4
relatively late it has also taken
advantage of all the speed and stability
fixes that were incorporated into the
3.4.2 release. Gnome too has been
upgraded to take in the goodness of
the 2.10 release. On the downside,
despite the recent kernel version the contents of those files, and makes
there’s still no sign of Gnome’s finding mislaid files much easier. If
wonderful search tool, Beagle. you’ve used Beagle, you’ll appreciate
Considering this, it’s surprising to that searching for the contents of a
find that KDE has been augmented file in this way rather than just its
with its own search tool, Kat, whose name becomes compulsive once you
feline form sits discreetly in your get used to it.
toolbar. Kat is obviously designed to Kat will eventually become an
compete with Gnome’s canine integral part of KDE’s file retrieval
scavenger, and works in much the once it is embedded in the file
same way. It’s a framework for requester, but in its current state you
indexing files described using certain can search the contents of compatible
MIME types. For example, descriptors files from the Kat application window.
for PDF or HTML allow Kat to index To use Kat, you first specify which
directories you wish to be indexed. Kat
will then wait for a period of CPU
MANDRIVA’S CORE inactivity before scanning through the
SOFTWARE directories to create an index. It’s quite
2006 LE 2005 processor-intensive, which is why Kat
Kernel 2.6.12 2.6.10 waits for a lull in your machine’s
KDE 3.4.2 3.3.2 activity. After the index has been built,
Gnome 2.10.2 2.8.1 it can be queried through either the
Apache 1.3.33 1.3.33
kioslave, by typing kat:/search in place
Firefox 1.0.6 1.0.1
of a URL in Konqueror, or using its
GCC 4.0.1 3.4.3
own user interface.
Gimp 2.2.8 2.2
OpenOffice.org 1.15 1.1.4 Another KDE feature added
automatically to the toolbar is Skim,

22 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.rev_mandriva 22 19/9/05 5:48:07 pm


HOW TO USE THE KAT SEARCH TOOL
1 2
You first need to specify just what you want
to be indexed, by adding a directory and the
kind of filetypes you want to be included.
which is KDE’s version of a Smart
Common Input Method platform;
functionally almost identical to the
GTK SCIM panel installed as default on
SUSE. Basically, it allows users of non-
Latin scripts to change their keyboard
configuration using just a couple of
quick mouse clicks. The KDE version
also lets you manage a couple of
functions of the KDE composite
manager, for adding transparency and
shadow effects to your desktop.
Applications like Beagle and Kat
modify changes to your system, so you
might think their presence in Mandriva
2006 indicates that the kernel has
been updated to include Inotify. Sure
enough, it’s the shiny new 2.6.12
kernel that’s at the heart of the new
Mandriva release. Not only that, but it’s
also been compiled to include support
for hyperthreading and Xen. Hardware
hotplugging has also been improved,
with the migration of certain modules
to Udev. We found USB devices in
particular worked better than in the
previous release, where repeated
removal of a USB key would eventually
mess up the automatic configuration.
New compiler
Sure to interest developers, Mandriva
has updated the Gnu Compiler
Collection to version 4 from version 3.
This is a big change, as the creators
had to compile and test every package
in the distro using the new version.
GCC 4 is new territory, as so many
things have altered for the new
release. We found 2006 to be very
stable, and compiling our own apps
using version 4 was unproblematic.
Mandriva stays at the head of the
distro pack by including version 5 of
LXF73.rev_mandriva 23
Clicking on the small Kat icon in the toolbar
opens another window, where you can start
each indexing process independently.
PHP, and also by using a cutting-edge
version of the X windows environment,
X.org (version 6.9). The desktop feels
smoother and slightly quicker, but
enabling the transparent windows and
shadows using the composite
extension still caused problems with
our NVIDIA hardware.
Other notable packages includes
with this release include Scribus and
Amarok, both at version 1.3. Mandriva
also bucks the recent trend of not
including patent-risking codecs by
enabling MP3 playback by default,
which is a good step.
The biggest disappointment is that
six months after SUSE 9.3 was
released with an almost fully
functional OpenOffice.org 2.0,
Mandriva is still using the previous
version as its office suite of choice.
This is understandable to a degree, as
not all pre-2.0 versions have been
very stable, but it would have been
nice for the download version to
provide a choice – especially when
SUSE has been able to make it work.
A ‘preview’ version of 2.0 is available
to Mandriva Club members on an
extra disc, but an office application
should be at the heart of a system,
and new users won’t be impressed if
they have to go out and get something
that they should have already.
For art’s sake
It seems traditional to pick on
Mandriva’s artwork, and we’re not
quite ready to break from convention
just yet. Thankfully, the gaudy starry-
eyed ‘crow’ from last time has been
replaced by a stylistic penguin staring
skyward. It’s an improvement, being
considerably more refined than its
www.linuxformat.co.uk
REVIEWS Mandriva Linux 2006
3
Once your sources are indexed, you can query
the database using the standalone Kat
application (above), or from Konqueror (below).
“This release has firmly
placed Mandriva in the #4
spot, after SUSE, Ubuntu
and Fedora. Choosing Kat
rather than Beagle? This
distro badly needs Mono.”
predecessor. But the slightly comical
essence of old has been replaced by a
degree of Kafka-inspired paranoia,
with the new penguin seeming to gaze
at the dawn of Linux on the horizon.
“MANDRIVA IS STILL GREAT
BUT IT’S DIFFICULT TO SEE
JUST WHERE IT’S GOING.”
Even Fedora has left the Galaxy
theme behind, and Mandriva really
should do the same. It’s easy to say
that you can change the whole style of
the desktop using a couple of clicks,
but for a distribution at the coalface of
converting people to the Linux fold,
the charcoal and blue colour scheme
is starting to give the feel of a disused
colliery under an icy sky. Mandriva
needs to spend time creating a lighter
default user interface, perhaps taking
some graphical cues from SUSE and
Red Hat Fedora.
You may have noticed that this
review has focused on several
relatively minor changes. After all,
KDE 3.4, Gnome 2.10 and GCC 4 are
essential, but other distributions have
already beaten Mandriva to the punch
by a considerable margin. Mandriva is
still a great distribution, and it will
always have a warm place next to the
hearth at LXF Towers, but it’s difficult
to see just where it’s going. The
bottom line is that there’s essentially
no difference between this release and
the previous one, other than the
essential upgrades necessary when
treading water.
There’s no evidence at all of the
Conectiva acquisition. This is still
Mandrake, even if it has tried to
eradicate every last instance of its old
name (having said that, Kat manages
to turn up a few). The configuration
utilities are still brilliant, and hardware
support is still unprecedented. We
finished our previous review asking if
Mandriva’s new strategy was the end
of the beginning, or the beginning of
the end, and we find ourselves six
months later putting the same
question back in cold storage. LXF
LINUX FORMAT VERDICT
FEATURES 6/10
PERFORMANCE 7/10
EASE OF USE 8/10
DOCUMENTATION 6/10
A solid distribution with much to offer,
but there are still questions about how
Mandriva will develop.
RATING 7/10
LXF73 DECEMBER 2005 23
19/9/05 5:48:09 pm
 REVIEWS SUSE Linux 10.0
LINUX DISTRIBUTION
SUSE
Linux 10.0
Stylish and feature-packed, SUSE
finally makes it to double figures.
Paul Hudson tries it out.
KEY STUFF
Home-user distro with features for
everyone’s tastes. Also consider:
Mandriva, Fedora Core or Ubuntu.
■ DEVELOPER Novell
■ WEB www.novell.com
■ PRICE Free to download under
GPL or £64.95 for boxed set with
install CDs, DVDs and manuals.
Over the years, we’ve come
to realise that SUSE is as
regular as the 7:36 from
Stuttgart to Munich. Six months ago
we reviewed SUSE 9.3, six months
before that we reviewed 9.2, and so it
has been pretty much since the
beginning of this magazine. SUSE has
traditionally lagged behind other
distros, with higher version numbers
eschewed in favour of fewer crashes.
April’s 9.3 release changed that by
eclipsing Mandriva for the first time in
our memory: it had more features and,
surprisingly, more stability than its long-
time competitor.
Despite the last release differing
so much from the one before that, it
warranted only a tiny 0.1 version bump.
What, then, has changed so much to
make this SUSE 10.0 rather than 9.4?
Well, the differences this time are
more philosophical than technical,
that’s for sure…
Tip of the hat
About two years ago we flew over to
Nuremburg – where SUSE’s
headquarters are – and took a tour
around the labs there. We made a
FEATURES
AT A GLANCE
Gnome 2.12
means spatial-
mode treeview
Nautilus. Another
good Mac idea
stolen borrowed!
24 LXF73 DECEMBER 2005
LXF73.rev_suse 24
point of asking when we would finally
see a truly open version of SUSE that
involved the community and was
available to download. The answer was
that the business model just didn’t
support opening the distro up. Clearly
the subsequent acquisition by Novell
has reformed SUSE completely, as
only a few months ago the new
openSUSE project was announced,
which aims to develop SUSE with the
community rather than hiding it away.
For end users this means weekly
builds of the distro, public roadmaps,
the chance to contribute and the
ability to get in touch with individual
developers at SUSE. In short, SUSE
has followed Red Hat, which forked off
its own desktop distro into the Fedora
community a while ago.
Before we get a barrage of emails
from novell.com addresses, we
should say that openSUSE is quite
different from Fedora – well, that’s
what the openSUSE people will tell
you anyway. The website contains such
choice quotes as: “The openSUSE
project embraces and develops several
additional important open standards
not included in Fedora, such as... Yast
(a standard, open source configuration
and management suite for Linux).”
The new,
more aggressive
SUSE password
checker.
www.linuxformat.co.uk
That’s somewhat akin to saying All new: cleaned-up Yast, a smart K
that Fedora embraces and develops menu and the Blam RSS reader.
several additional important open
standards not included in SUSE, such the best you’ll find – plus an inch-high
as Anaconda (the Red Hat installer). stack of CDs and DVDs ready to be
Here at LXF Towers we consider installed. Fedora, however, tends to
standards to be either de facto (made leave full stabilisation of a release for
standard by popular use) or de jure the Red Hat Enterprise Linux
(made standard through a ruling of a developers, which leaves the basic
standards body). Yast is neither, so Fedora Core installs unsupported and
keep your FUD alarm on while reading often unstable.
the openSUSE website. As part of making SUSE open, the
Perhaps the only real difference www.opensuse.org site hosted a
between openSUSE and Fedora is that new build of version 10.0 every week
the former is turned into a boxed before release, working through beta
product at the end of the releases 1 to 4, then release
development cycle. The SUSE boxes candidates. Downloaders could test
provide extensive documentation – out each new release and submit bugs,
and even host their own projects on
the site adding their own patches and
SUSE’S CORE customisations to the release for
SOFTWARE others to try. Although we’d rather see
10.0 9.3 full access to the build server (this is
Kernel 2.6.13 2.6.11 planned sometime next year) we have
KDE 3.4.2 3.4 to give kudos to SUSE for pulling off a
Gnome 2.12 2.10 great start to the project.
Apache 2.0.54 2.0.53
Firefox 1.0.6 1.0.4
GCC 4.0.2 3.3.5
The applications
Gimp 2.2.8 2.2.4
With all the glasnost swishing around
OpenOffice.org 2.0 2.0 SUSE, it’s perhaps not surprising that
SUSE 10.0 is less of a technological
The Blam
newsreader is
installed by
default – and
it rocks!
19/9/05 5:48:49 pm

revelation than 9.3 was. It’s partly a
victim of its own success – putting Xen,
Beagle OpenOffice.org 2.0 and Mono
all into one release didn’t leave much
behind to make headlines for 10.0.
But don’t despair – there are enough
new things here to keep SUSE fans
interested:
■ This is the first distro (apart from
Ubuntu) to release Gnome 2.12,
reviewed on page 30. It isn’t really a
user-centric release – the biggest new
feature is the new Cairo back-end for
GTK that makes the graphics much
smoother, but there aren’t many other
exciting new features around. Still,
you’ll be getting the most up to date
Gnome version available.
■ Evolution is now up to version 2.4.
The email client has been refined and
polished, with even the most obvious
new feature being fairly minor.
■ Xen 3 is included, bringing with it
support for x86-64 and SMP.
■ Kernel 2.6.13 is here, so you get
Inotify support out of the box.
The inclusion of Gnome 2.12 is
marred by the fact that the system
tools are available only on the DVD
release. This is also true for a few other
things, including PHP 5. SUSE really
needs to grow by an extra CD to avoid
annoying people without DVD drives.
There are some other, smaller
version bumps – the latest Mono and
Beagle are included, for example –
but most other components haven’t
needed to change much in the past
six months. KDE 3.4 and Gimp 2.2
remain, although perhaps with fewer
bugs this time around. One new
feature we did like in the K Menu is
the ability to type in a search term to
have it grey out menus that don’t
contain what you’re looking for.
Even the login screen has been smartened up, and various themes are
available for it. This is the default ‘SUSE 10.0’ theme.
LXF73.rev_suse 25
THE SUSE COMMUNITY
Distro community-making is almost
becoming a sport, and the latest event is
the different ways in which distro
makers try to form communities around
themselves. For Mandriva – arguably the
first distro to try this – the plan was to
sell membership and provide unique
features. Ubuntu seems to have done a
great job of making a community
through its plan of mailing out CDs free
of charge, but to be fair not many can
afford that! For SUSE the plan seems to
be to release beta versions of its distros
very regularly to keep people interested.
In fact, the only distro to have tried to
make a community and failed (so far) is
Fedora, where the plan was, “If we build
it, they will come”. As Graham Morrison’s
account of his Kalbum project showed in
LXF72, a community won’t build itself.
So, although this is nothing like the
version number jumps we saw in 9.3,
SUSE remains ahead of the pack on
every front.
Setting up
SUSE’s installation system was already
easy to use, and the team have made
a number of helpful little tweaks that
make things even easier. Some of the
more advanced options are
now hidden away under their
“At a time when boxed
distros seem to be falling
out of favour, SUSE has
upped the ante, once again
proving you can have your
proverbial new version
cake and enjoy eating it.”
www.linuxformat.co.uk
Can SUSE keep up the pace while
opening up even more of its systems?
Right now you can download CDs of
each test release as they come out, but
really you want to be able to download
nightly builds to see if the bug you
submitted yesterday really is fixed or
not. Similarly, can we expect SUSE to lay
down a strict trademark law for people
who want to build on top of SUSE?
Traditionally, Debian, Red Hat and
Mandrake have been used as parent
distros, but SUSE almost never – is that
going to change?
Let’s hope so. If SUSE really means to
be open, it should go all the way: let
people create SUSE-derivatives (and
mark them as such) to their hearts’
content, and that will be what truly
makes the SUSE community a success.
own tab, which means users who don’t
want all the dirty details just get
overview screens. To back that up,
there’s a clear ‘choose between
Gnome, KDE or other’ screen, which is
much easier for new users and also
places SUSE firmly in the middle of
both desktop camps.
The default installation is packed
with programs and so is quite sizeable,
but you do get just about every
program you need as a desktop user.
Mono is installed by default, along with
a number of Mono-based programs
such as Tomboy, Beagle and Blam. All
of the OpenOffice.org 2.0 office suite is
here, naturally, and it’s great to see
that SUSE has followed Red Hat by
using the Java-free version of OOo.
SUSE’s Yast installer and control
panel system has been tweaked in
various small but appreciated ways.
SUSE 9.3 (and prior releases) had a
large Yast logo at the top of the
window plus the SUSE toad, which
took up space needlessly and just left
people wondering what ‘YaST’ means.
Fortunately SUSE has redone the
interface. This change has been
mirrored in the Sax2 X configuration
tool, which used to have its own weird
interface stuck in the middle of Yast.
But now, praise Linus, this is fixed.
Yes, it took ten versions to stop Yast
sucking like an industrial vacuum
cleaner, but better late than never. Of
course, it now only sucks like a small
Dyson: much more work is needed,
particularly to clean up the mass of
jargon in the Network Services screen.
SUSE employees: please, if anyone
wanted to set their computer’s time
over the network, they would never,
ever look for ‘NTP client’!
REVIEWS SUSE Linux 10.0
In the freely downloadable release,
some commercial packages are
missing: Java is gone, as is
Macromedia’s Flash Player and Adobe
Acrobat. We’re happy to wave goodbye
to all three, but if you feel pulled by
the call of proprietary software and
don’t want to buy the commercial
SUSE you can still go ahead and
download it all – the free SUSE even
points you in the right direction.
More interesting is that the
openSUSE project has led to a freely
available SUSE release for PowerPC
chips (ie your shiny new Mac Mini).
SUSE has loved and left PPC before,
so this may only be a temporary thing…
For the coders
SUSE continues its dominance on the
programming front with the much-
needed inclusion of MonoDevelop.
Mono libraries are included for a wide
range of things, even down to
ipod-sharp so you can write C#
programs to play around with your
new iPod Mini. SUSE’s commitment to
detail here is almost fanatical:
GCC 4.0.2 is the absolute latest
version as we’re going to press – a
whole 0.0.1 version number higher
than in Mandriva. You also get
KDevelop and Glade for Qt- and GTK-
based development, although Qt
Designer is here in its ‘KDE Designer’
guise as well as plain old Qt Designer.
With Mandriva falling even further
behind, we’re starting to wonder
whether SUSE is beginning to focus
on Ubuntu as its closest competitor.
The Breezy Badger release (reviewed
next month) includes Mono and
Gnome 2.12, plus a slew of its own
new features.
Of course, no one really ‘wins’ in
the Linux market as we all consider
proprietary software our common
enemy, but nonetheless it is
interesting that SUSE has gone to so
much effort to be more open with this
release, and focused its efforts on
refining what’s already there rather
than introducing new things. LXF
LINUX FORMAT VERDICT
FEATURES 9/10
PERFORMANCE 9/10
EASE OF USE 8/10
DOCUMENTATION 9/10
An enjoyable release that builds upon all
the successes of the previous release.
Remains top of the game.
RATING 9/10
LXF73 DECEMBER 2005 25
19/9/05 5:48:52 pm
 REVIEWS Gnome 2.12

DESKTOP ENVIRONMENT

Gnome 2.12
Has Gnome managed to add updates without compromising its prized usability? Andy Hudson finds out.

KEY STUFF
Heavyweight window manager.
Also consider: KDE 3.4 or IceWM.
■ DEVELOPER Gnome Foundation
■ WEB www.gnome.org
■ PRICE Free under GPL

Nothing inspires a good

punch-up in the Linux
community as much as
choice of window manager. There are
three main camps: the lightweights,
the KDE supporters and the Gnome
advocates. In our review of Gnome
2.10 (see LXF66), we said that it had
reached a good degree of maturity,
both in features and stability. We’re
happy to report that Gnome 2.12 goes
even further, and reminds us once
again why Gnome is so many people’s
window manager of choice.
There are, as ever, two main ways
of getting Gnome 2.12. The first, and
most time-consuming, is to use the
source tarballs released by the Gnome
Foundation. The alternative, which the
majority of users choose, is to wait
until their distro is updated with the
latest release. This is the method we In typically usable fashion, Gnome 2.12 allows you to see exactly what you are dragging and dropping.
used. Testing versions of Gnome 2.12
were available on both OpenSUSE and Gnome community has put a lot of little addition is the new sidebar, where visibility of existing menu options, but
Ubuntu Breezy Badger in time for our effort in to this release. you can quickly access any it’s a step in the right direction (we’d
review. We went with OpenSUSE. One of the most significant bookmarked files, whether they’re stick with Smeg for now).
updates under the hood is the network drives or local directories.
Sleek inclusion of GTK 2.8, which uses Cairo Apart from Nautilus, improvements Simplicity
First impressions were as favourable as to enhance the graphics and make the have been made to the bundled Version 2.12 remains faithful to
ever, with 2.12 retaining the sleek look user experience more enjoyable. The software, including Evolution. This MS Gnome’s core beliefs – especially that
that Gnome’s followers know and love added bonus is that GTK 2.8 is totally Outlook replacement has had a of usability. Unlike KDE, with its horde
– but it now sports a new default compatible with earlier versions in the number of new plugins developed for of applications and seemingly endless
theme: Clearlooks. Users of Fedora 2.x family, so it’s easy for developers to it, and work has also started on ways of doing things, Gnome is still
Core 4 will already be familiar with this make the most of its functionality. making it compatible with CalDAV (the comparatively simple. But with two of
theme, as it was chosen to replace Another thing that GTK allows is the calendaring portion of WebDAV), the top three distros on DistroWatch
Bluecurve as that distro’s default ability to preview text when you drag which should make scheduling using Gnome by default, surely world
desktop appearance. It’s functional, and drop it across different Gnome- extremely easy with a messaging and domination can’t be too far away! LXF
clear and gives Gnome a fresh look. aware applications. As you drag the calendar server such as Hula.
Visually, not much else has changed. text around you can see exactly how Other changes include
LINUX FORMAT VERDICT
Don’t let this deter you though, as the much you are moving. enhancements to HAL that enable
Gnome to recognise even more FEATURES 9/10
Software plugged-in devices, and report more PERFORMANCE 9/10
Nautilus is still Gnome’s default file useful information. Weather Reporter EASE OF USE 9/10
manager, and love it or loathe it, it’s can now quickly find your location DOCUMENTATION 8/10
here to stay. There’s now a list view wherever you are in the world, and
very similar to that popularised on there’s a new CD and DVD-playing Another solid release, raising the bar for
features and ease of use.
Macs prior to OS X. Another handy applet in the toolbar. And to appease
those who have been campaigning for RATING 9/10
Your favourite tracks are just a one for years, a menu editor has been
click away with the new applet. added. It only allows you to toggle the

28 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.rev_gnome 28 21/9/05 1:49:01 pm

 REVIEWS Sugar Suite 3.5
CUSTOMER RELATIONSHIP MANAGEMENT SOFTWARE
Sugar Suite 3.5
CRM software is usually closed source. Will this open release please Graham Morrison?
KEY STUFF
Suite for a range of CRM tasks: use
email marketing templates, track
customer queries, create quotes,
record customer history and more.
See also: AbsolutelyBusy.
■ DEVELOPER SugarCRM Inc
■ WEB www.sugarcrm.com
■ PRICE $2,245 for five users
or free to download under
SugarCRM Public License
It’s surprisingly common for
businesses to keep their
customer details in one large
spreadsheet. This is especially true
when the details are only needed
once a year for the Christmas card list.
But if you operate like this, you’re
missing out on a wealth of information
about your customers with every sale
or enquiry. Even geeks can appreciate
the amount of data that’s lost when
you put the telephone down on a
customer without making any notes.
And so customer relationship
management was born, the tedious
name given to any smart customer
database and contact management
suite that helps businesses keep tabs
on their customers and measure the
success of marketing campaigns.
Sugar Suite is one such application.
It is a classic Lamp solution: combining
Linux, Apache, MySQL and PHP. You
can use it to enter customer details,
making notes and to-do lists as well as
planning meetings or collaborating on
sales opportunities. These are all
avenues of potential for a business,
paths that present themselves and
need to be followed up. In layman’s
terms, they’re goods and services.
From a system administrator’s
perspective, installation is relatively
easy. As long as your server is running
the required versions of the software,
FEATURES
AT A GLANCE
Add and manage
your sales
contacts from a
button bar.
30 LXF73 DECEMBER 2005
LXF73.rev_sugar 30
You can manage your entire customer database from the home page,
but the price is a rather cluttered design.
installation is as simple as extracting
the entire Sugar Suite application,
copying it to your web directory and
running the included install.php page
from a web browser.
You then need to answer a few
questions about database
configuration and you’re done.
How it works
Each user within the company has an
account. There you create customer
accounts for your contacts and report
on any progress using the contact
manager. You can raise sales
opportunities as they occur and mark
any preliminary contacts as ‘leads’.
Your company’s product details are
also part of the system, and any
problems with your product can be
registered as cases then, if it’s not your
job to fix them, assigned to colleagues
for resolution.
The breadth of this software is
sometimes staggering. You’ll find
modules for managing customer
quotes, and a bug tracker in case
Create custom
reports with
complex
queries.
www.linuxformat.co.uk
■ Free version
you’re selling software. Sugar Suite has
an RSS module for news streams,
integrated email, and an extension to
use with Microsoft’s Outlook. And as
you might expect with a system that
generates more data that the Hubble
space telescope, there’s also a
comprehensive reporting module
where you can add fields from just
about any of the other modules.
Sensitive, uncompetitive types
might like to look away while we
discuss the dashboard. This is where
you can see your sales scores pitted
against those of your fellow team
members. Data mined from the
database is used to determine your
yield and efficiency from the various
customer-facing modules. Scary.
After logging into your account, the
first page you see is the SugarCRM
‘home’. This is an overview of your
contacts and appointments, and shows
you notices from other members of
your team. The wealth of information
you’re presented with here is a little
overwhelming, and not made any
Identify the most
profitable sales
channels with
the dashboard.
easier by the presence of more than
20 tabbed buttons for getting to the
other pages in the system. The
problem is that Sugar Suite tries to fit
too much information on one screen,
producing an interface that’s busy and
hard to get used to. We’ve seen worse,
though, and the sympathetic colour
scheme and framing soften the blow.
Try it for free
The Enterprise version of Sugar Suite
supports Oracle databases as well as
the default MySQL and an advanced
reporting module that features
complex queries and a comprehensive
support package. There are other
versions, including a hosted ‘Sugar On-
Demand’ service, and a cheaper
Professional version.
If enterprise-level pricing leaves a
bitter aftertaste, consider the free
edition. It uses yet another licence
(the SugarCRM Public License, natch),
but it is free, and can be a massive
boon if your business needs to really
get under the bonnet and customise
its CRM application.
Sugar Suite is so big and so
complex that you really need to give it
a long-term trial. Regardless, this is
impressive software, perfectly capable
of becoming an integral part of your
business strategy. It’s feature-rich,
comprehensive, and best of all, you
can see the source code. LXF
LINUX FORMAT VERDICT
FEATURES 8/10
PERFORMANCE 6/10
EASE OF USE 6/10
VALUE FOR MONEY 7/10
This offers the best of both worlds: a
professional and well-supported solution
that comes with source code exposed for
all to see.
RATING 7/10
21/9/05 3:13:47 pm
 REVIEWS CrossOver Office 5.0
WINDOWS EMULATOR
CrossOver Office 5.0
‘Naughty but nice’ redefined for geeks: eating
cream cakes while using MS Office on Linux. Would
Graham Morrison pay $40 for such an illicit thrill?
KEY STUFF
Get Windows applications working in
Linux. See also: Wine.
■ DEVELOPER CodeWeavers
■ WEB www.codeweavers.com
■ PRICE $39.95
CrossOver Office last graced
our pages back in LXF63, in
February of this year. And
since then? Well, most people use the
summer months for playing volleyball
on the beach, or Unreal Tournament
in the office. But those crazy folks at
CodeWeavers have worked all summer
long to bring us a shiny new version,
and a winter break isn’t even on the
horizon: CodeWeavers’ product
roadmap includes another major
release of CrossOver Office before the
end of the year. That will be version
6.0, and targeted specifically at
running Windows games on your Linux
desktop. Development of this version
is most likely in response to a recent
announcement from erstwhile games
porting specialist TransGaming that it
intends to broaden its compatibility to
include Windows applications, clearly
stepping on CrossOver Office toes.
Back to the present, and version 5
of CrossOver Office. For the uninitiated,
CrossOver is an application designed
to run Windows applications from
Linux. It uses Wine to provide all the
A mouth-watering array of
Windows applications are
supported.
32 LXF73 DECEMBER 2005
LXF73.rev_coo 32
■ Wine, the base
of CrossOver Office
Windows system components, and
CodeWeavers contributes to this
hugely popular open source project in
turn by ploughing its own changes
back into it. Each release provides
compatibility with a desired Windows
program (you can register your interest
for those you need on the
CodeWeavers website). Last time it
was Apple’s iTunes software, this time
it’s Microsoft Office 2003.
Cross redressing
Considering CrossOver’s complexity, it’s
surprisingly easy to install and manage. Above: CrossOver Office lives up to its
You just need to execute a single name with Microsoft Office 2003.
installation script, after which you find Right: Each application can share a
‘bottle’– its own Windows version.
a setup utility added to either your
Gnome or KDE menu. It’s from the
setup utility that you add and remove There’s another advantage to this
programs and repair any installation approach; you can create bottles
that fails to work. tailored for a specific Windows is usable, but only if you absolutely
For the most part, you install operating system, such as Windows 98 need to buy music from Apple on your
applications that have already been or Windows 2000, and even run Linux machine – you’d be mad to use
proven to work, and you select which applications from different bottles at it over a native Linux media player.
you need from a list. If this is a freelythe same time. Other applications don’t seem to
downloadable application like iTunes or CrossOver Office suggests which suffer the same problems at all, and
Windows Media Player, the setup bottle is most appropriate for which even Windows Media Player works well.
software will automatically download application at installation time. ITunes, Using CrossOver’s filetype editor, you
the files before attempting to install for example, uses a Windows 2000- can manually specify which files you
them. There’s almost no user tailored bottle, while other applications, would like Media Player to open,
interaction, and the whole process is Windows Media Player included, seem making it the perfect solution for
very smooth. If the software you’re to prefer Windows 98. some of more prickly proprietary AVI
installing comes on a CD-ROM, it’s The other big new feature in this formats you often find on the internet.
just a matter of pointing CrossOver release is, as we’ve hinted, CrossOver Office seems to be going
Office to the location of the drive on compatibility with Microsoft’s Office from strength to strength, and we
your system. 2003. It worked exceptionally well, can’t wait for the game-centric version
even going as far (too far?) as adding 6.0 at the end of the year. If you need
Bottles of Wine the Microsoft Customer Improvement Windows compatibility, and CrossOver
The big addition for this release is Program applet to the tool bar. It’s Office supports the applications you
what CodeWeavers calls a ‘bottle’. A tempting to say that Office 2003 need, it’s highly recommended. LXF
bottle holds a single Windows performs better than OpenOffice.org –
installation, and you can install as it certainly felt like it. And it loads LINUX FORMAT VERDICT
many or as few applications into any faster too, which is really rubbing salt FEATURES 8/10
one bottle as you like. In the previous into the wound. Overall, MS Office ran
PERFORMANCE 7/10
version of CrossOver, you had basically wonderfully (with a couple of crashes),
a single bottle, and a single Windows which makes this upgrade essential if EASE OF USE 7/10
installation for each program. you need to use Office 2003. VALUE FOR MONEY 8/10
One bottle includes everything a Performance has improved
Get seamless integration of Windows
Windows applications needs to run considerably over the previous version applications on your desktop without
properly, such as the program files and – with one notable exception. ITunes is crossing over to the dark side.
the registry settings, and it means that still an ugly resource hog, making the
an ill-behaved application can be user interface very difficult to use. RATING 8/10
quarantined and left to run riot in its This is no doubt down to Apple’s
own virtual Windows installation. brushed-aluminium window design. It
www.linuxformat.co.uk
21/9/05 3:52:45 pm

BACKUP SOFTWARE
BrightStor ARCserve Backup
It’s scalable, but is it usable? Graham Morrison looks at a backup solution for the big boys.
KEY STUFF
Client–server backup system for
business. Alternatives include Bacula
and Storix.
■ DEVELOPER Computer Associates
■ WEB www.ca.com
■ PRICE $775
BrightStor ARCserve Backup
isn’t for the faint of heart. If
you’re looking for a backup
tool that’s quick and easy to use, you’ll
need to look elsewhere. To be fair, this
software hasn’t been designed for
backing up your MP3 collection. It’s
intended to be a scalable solution that
can cope with the dozen or so
machines in a typical small business or
the incredibly complex requirements
of a large data centre. That’s an
ambitious remit for a product that has
the potential to cut short an IT
manager’s career if it goes wrong.
The implementation of ARCserve
Backup is equally ambitious. A server
is used for managing your backup
data, regardless of whether this exists
on external tape machines or internal
SCSI disks. A client makes a
connection to the server for
configuring and managing the various
backup jobs and processes. The client
can be just as easily on the same
machine as it can across the other
side of a network.
Arcane installation
Installing ARCserve Backup is not
anywhere as easy as it should be.
You’re first drawn into the realm of
false security by the presence of an
“DOES THE JOB IT’S
DESIGNED FOR SO THAT
YOU CAN KEEP YOURS.”
installation script. Executing this script
goes well, and the reassuring output
from each stage of the process gives
you the impression that your
sysadmin hacker skills are
unquestionable. All the tools are
installed along with the Ingres
database, and the script even goes to
LXF73.rev_arc 33
If it’s not pre-installed, you’ll have to unravel the thorny installation
process to arrive at this neat client interface.
the trouble of creating the correct
username and permissions for you.
It’s only afterwards, when you’re
left with a blank command prompt,
that you’re left wondering just where
to turn. Fortunately, there’s a ‘Getting
Started’ guide in the form of a PDF.
The trouble is that it doesn’t really
help you get started.
This is compounded by the cryptic
error messages you’re sent as you
wade through the command-driven
setup utilities. Here’s a classic example:
‘ERROR: There was an error starting
the httpd daemon. Please examine
the error message and check the
httpd directory.’ Repeatedly looking at
the error message doesn’t seem to
make the error any clearer, no matter
how hard you scratch your head.
Once it’s set up, ARCserve Backup
operates on several levels. The lowest
is occupied by a series of
management processes or daemons,
designed to access and control the
back-end database, backup devices,
your backup jobs and user alerts.
www.linuxformat.co.uk
REVIEWS BrightStor ARCserve Backup
On top of that is a management
level for controlling these daemons.
They can augment the system by
providing user control, managing
media or copying tapes. You can query
their status with the cstatus command.
Understanding what each core
management service does is an
important part of getting your head
around the ARCserve way of doing
things, especially when it comes to a
couple of the more important ones.
The Discovery service finds all the
available backup servers on your
network, and fields connections
between the back-end and the clients.
The job of the Authentication service
is to ensure the integrity of the
network resources, as well as to
manage user permissions for each
resource. Another vital service is the
Media Manager. This handles the
actual transfer of your vital data
between the server and your chosen
back-up device.
An applet a day…
This sounds complicated, and it is! But
things get slightly easier when you
finally get around to using your system
via the Java-encumbered web
interface – the top level of the system.
Launching a browser and connecting
to the backup server embeds 14
separate Java applets, shown in the
screen shot at left. It’s from here that
you carry out the real tasks of an
enterprise backup: creating backup
jobs, managing the media that your
data is stored on, and recovering data
in the event of a failure.
With the obvious caveat that Java
applications can be somewhat slow
and unwieldy, each of the separate
components uses its own window,
which makes the division of labour
easy to understand.
You’ll probably spend the most
time in the first window, the Backup
Manager. This is where you create your
backup jobs, in simple
and straightforward terms. That
means specifying a source directory
and how you would like the job to be
handled by the backup server. You can
elect for the compressed archive of
your files to be backed up to a local
or remote disk, a server, tapes or
another destination.
Other applets are overwhelmingly
comprehensive. Tasks including user
access, virus checking and remote
tape management can be handled
from here.
BrightStor ARCserve isn’t meant for
mortals. It’s meant for super system
administrators who feel completely
comfortable dealing with gigabytes of
data on a daily basis. Computer
Associates should really provide some
on-site training with every installation,
because the unfathomable installation
procedure is far trickier than it needs
to be. This is one of the most
comprehensive yet convoluted
applications of its kind that we’ve seen.
But when disaster strikes, it does do
the job it’s designed for, so hopefully
you can keep yours. LXF
LINUX FORMAT VERDICT
FEATURES 8/10
PERFORMANCE 7/10
EASE OF USE 5/10
VALUE FOR MONEY 6/10
The cost of scalability is complexity, but this
is a versatile and capable solution.
RATING 7/10
LXF73 DECEMBER 2005 33
16/9/05 5:08:34 pm
 REVIEWS Books
Learning Perl, 4th Edition
The fourth edition of an old book, but Paul Hudson thinks it suffers from the same problems as before…
KEY STUFF
■ AUTHORS Randal L Schwartz
et al
■ PUBLISHER O’Reilly
■ ISBN 0-596-10105-8
■ PRICE £28.50
■ PAGES 283
There has been a recent
resurgence in Perl books, despite Perl
6 apparently being as far away as it
has ever been. This is Learning Perl,
reviewed below is Advanced Perl
Programming, and we’re also in the
middle of Perl Best Practices: the slew
of treeware is pitched at a range of
ability levels. As its name suggests, this
book is a junior effort – but being an
O’Reilly Learning title, is geared
towards people who have experience
in a related area.
At the advanced end of the
O’Reilly spectrum lie the Programming
books (Programming Perl,
Programming Python etc), and at the
other end there’s the Nutshell books
Advanced Perl Programming
Going from Learning to Advanced Perl in one page takes a master, but we only have Paul Hudson…
KEY STUFF
■ AUTHORS Simon Cozens
■ PUBLISHER O’Reilly
■ ISBN 0-596-00456-7
■ PRICE £28.50
■ PAGES 281
Being able to program in Perl does
not make you a Perl programmer.
While you might finally understand
Damian Conway’s Lingua::Romana::
Perligata and have all but memorised
Programming Perl, you’ll probably still
be a bit confused if you go along to a
Perl Foundation fundraiser and find
people offering money for Michael
Schwern to keep his shirt on. Suffice
to say, the Perl community is closely
knit, with people in the inner circles
writing code that pluralises English
words, obfuscates the string ‘just
another Perl hacker’ and reads as
haikus – as long as everyone agrees
that $foo is pronounced “dollar foo”.
This book is aimed at people who
have finished Programming Perl and
want to go still higher. However, we
34 LXF73 DECEMBER 2005
LXF73.rev_books Sec1:34
(Perl in a Nutshell, Python in a Nutshell,
etc). These two series are both highly
prized and work well in their area, but
there’s quite a gap between them –
you can’t expect to finish Perl in a
Nutshell then go on to read
Programming Perl, because the latter
discusses things you still won’t know.
This is where the Learning books are
intended to sit: they should give you
all the low-level building blocks
were pleasantly surprised to find that
it’s not full of back-references to the
other book, which means you can
jump straight in here if you’re an
experienced Perlmonger. The contents
start off pretty crazy, and only get
wackier the more you read – there’s
no real organisation, so you just keep
turning the pages wondering what the
heck will come next. And turn we did:
the code is well thought-out and
interesting, and the topics cover
www.linuxformat.co.uk
required to take you up to
Programming level.
However – and this is key – this
book isn’t suitable for people coming
to Perl as their first language.
Admittedly, part of the problem is that
learning Perl will always be difficult if
you’re not already familiar with Unix
tools such as sed and awk, but the
bigger worry is that the book reads
like another Nutshell: though it’s and
enough bases that everyone will find
something of interest. The chapters we
particularly enjoyed were Parsing
Techniques (how to parse your own
language in Perl), Natural Language
Tools (to get on the English word
pluralisation bandwagon) and Fun With
Perl, which explains how to play Perl
Golf and more.
Our experience with Perl is pretty
sizeable, but we learned a lot from
reading this book. For example, the
is written in a tutorial style, Learning
Perl is a reference guide at heart,
which means it goes through Perl
programming topic by topic instead of
slowly building your knowledge up.
You won’t get to learn about
subroutines, for example, until you’ve
learned everything about operators.
Some parts of the book are a bit
random, and this is most apparent in
the use of footnotes: almost every
page has one or more footnotes, with
some having up to five.
Our last niggle is with the index. It’s
nowhere near big enough (only nine
pages), and it’s so bad that we had to
read every entry to find what we were
looking for.
LINUX FORMAT VERDICT
An adequate introduction if you already
program competently, but you’ll need the
Nutshell book alongside.
RATING 6/10
chapter Inline Extensions covers how
to embed C, Ruby and Python inside
your Perl scripts. For a long time this
was one of those things we dismissed
as cute but useless, but Crozens
provides enough examples not only to
get you started but to spark a real
interest in making you want to find
uses for inlining other languages – it’s
crazy, but this book makes it fun.
Is there anything to drag Advanced
Perl Programming down? Again, we
find a small index, but so much of this
book is already code and diagrams
that a short index is pretty inevitable.
Kudos to Cozens: at last this is a book
we’re proud to place next to
Programming Perl.
LINUX FORMAT VERDICT
Packed with tools and ideas to try if you’re
looking to push your Perl envelope. Highly
recommended.
RATING 9/10
21/9/05 3:52:19 pm
 REVIEWS Books

Astronomy Hacks
Observe Nick Veitch mixing with the stars.

KEY STUFF and why to star-collimate your

■ AUTHORS Robert Bruce
Thompson and Barbara Fritchman
Thompson
■ PUBLISHER O’Reilly
■ ISBN 0596-10060-4
■ PRICE £17.50
■ PAGES 390
You don’t even need so much
as a pair of binoculars to get into
astronomy – a single eye will do.
Still, before you set the controls for
the heart of the sun you should do
some swotting, and Astronomy Hacks
is a good place to start. It is not
actually aimed at beginners, but
newcomers to astronomy will find
plenty of useful information here,
particularly on the differences
between types of telescopes and how
to choose one that suits both you and
what you want to observe.
Though it starts off with some basic
astronomy, at least half of the book is
devoted to more advanced observation
with telescopes. For such an
accessible book there is a
phenomenal breadth and depth to the
information presented, and even the
most knowledgeable astronomer is
sure to find something worth reading.
Much of the material only adds
weight to the notion that astronomers,
amateur or otherwise, are anal
retentives who must be avoided at
parties. Excellent tips on observing
shallow space objects are mixed with
advice on how to take proper
observation records: “The log sheets
you fill out during observing sessions
are journals because they record
transactions chronologically.
Periodically, you should transfer those
records to a consolidated ledger”.
Zzzzzz.... However, gems such as how
telescope are not only thought-
provoking, practical and well explained,
but also interesting. All of the topics
discussed are done so with a depth
and clarity that many other astronomy
books can only aspire to.
As ever with the Hacks series this
is a mixed bag, but if you have more
than a passing interest in astronomy
there is a lot of practical, real-world
advice here, blended with useful
background information and
presented in an easy to understand
way. This is an excellent companion
to a night under the stars (and let’s
face it, you aren’t going to be invited
to many parties). LXF
LINUX FORMAT VERDICT
This book will save you time and
help you see more.
RATING 8/10

www.whosbehindyourwebsite.co.uk

If you’re designing & hosting your clients’ websites, or building &

testing your own applications - what you really want is the flexibility and
performance of your own server. But sometimes the cost of managing
one is just too high, both in terms of time and money.
So wouldn’t it be great if you could share the cost of a server’s hardware
but not share its resources? In other words, the full power of your own
server with multiple domain support - in an easy-to-manage, shared
environment.
Hostway Virtual Private Servers - Now you can really get on with
doing what you do best...
To find out why visit:
onth
www.whosbehindyourwebsite.co.uk 9.95 / m
From £3 set-up fees.
and no

call us now on 0808 180 1880 • sales@hostway.co.uk • www.hostway.co.uk

Domain Registration • Email Hosting • Web Hosting • Ecommerce • Resellers • VPS • Dedicated Servers

LXF73.rev_books Sec1:35 21/9/05 3:52:24 pm

 Roundup
Every month we compare tons of software,
so you don’t have to!
PHOTO MANAGERS
If you need to get your digital photo collection in order,
Andy Channelle has just the collection of apps for you.
Digital technology has
transformed photography. In
the past, cameras came out
for special occasions and users had to
shoot sparingly because it was so
expensive to get prints developed. In
contrast, the advent of decent camera
phones means people often have a
camera to hand all day, every day.
Moreover, you no longer having to
process 23 blurred images of your
thumb just to get a decent picture of
your wife/kids/dog. Both of these
developments have led to a huge
increase in the number of pictures
people take and, subsequently, a
growing need for computer
applications to handle the results.
36 LXF73 DECEMBER 2005
LXF73.round 36
There are loads of Linux
applications capable of dealing with
your digital image collection, but which
of these is best for you?
For this Roundup we looked for
software that makes managing, editing
and distributing snaps easy and fun.
Of particular importance is the ability
to add notes or tags to images and
then search the collection based on
this information – this is a prized
feature because most digital images
start out called something unhelpful
like dcf00056.jpg. And while we
don’t expect any of these apps to
challenge Gimp in the editing stakes,
having to fire up a fully-fledged
graphics application to change the
www.linuxformat.co.uk
orientation of a portrait shot or
remove a bit of red eye is a pain, so
some editing capability is also helpful.
As these are desktop-focused
applications, we put a premium on
ease of installation and user-
friendliness. Finally, we like applications
that make it easy to show pictures OUR SELECTION
either locally in the form of slideshows, AT A GLANCE
or on the internet with HTML galleries
or, even better, through services such DigiKam ...................................38
as Flickr and Fotopic.net. F-Spot .......................................37
We tested each piece of software
GThumb ...................................39
on an AMD Athlon XP 2500+ with
512MB of RAM running SUSE Kalbum.................................... 40
Professional 9.3. This was loaded with KimDaBa .................................39
2,730 digital photos of various sizes KPhotoBook ....................... 40
taken over a period of six years.
19/9/05 5:56:48 pm

F-Spot
Mono-based app with some nice ideas.
■ VERSION 0.0.13 ■ WEB www.gnome.org/projects/f-spot
■ PRICE Free under GPL
F-Spot has been acclaimed in this
magazine as one of the most
promising Linux programs to emerge
this year. It was built using the new
Mono platform, so you’ll need the
Mono runtime installed first. This is a
recipe for dependency hell, but
fortunately most distros should ship
with both Mono and F-Spot available
through the package manager.
Once it’s set up, you can add
photos by selecting File > Import and
choosing a directory. It took us just
over two minutes to index and display
2,500-plus images in the browser
window. F-Spot is EXIF-aware, so
pictures will be arranged by the date
of the photo; the rest of the EXIF data
can be viewed by selecting an image
and hitting Ctrl+I. F-Spot can also grab
pictures directly from a camera.
“THIS IS A VERY SMART
APPLICATION THAT MAKES
TAGGING IMAGES EASY.”
F-Spot’s tagging system can add
tags to hundreds of images at a time if
necessary. The tags are all user-
definable so you can build up a
comprehensive list of people, places,
events and concepts on your PC and
then assign different tags to images.
Sorting images is then simply a case
of selecting tags in the sidebar;
everything tagged with that word will
be displayed.
Ranged along the top of the
browser window is a calendar widget,
which uses a bar chart metaphor to
display how many pictures you’ve
taken in each month. Clicking on any
of these ‘bars’ jumps the browser to
the first image of that month. In
tandem with the tags, this is a great
way to find the right picture every time.
Double-clicking on an image will
open it up into the main window, and
from here you can make minor colour
adjustments, crop and resize the
picture to a number of preset sizes,
and remove red eye. Changes are
made to a copy of the image (the
original remains untouched) so it’s
LXF73.round 37
possible to revert even after the
application has been shut down and
restarted. This could lead to wasted
disk space if you do lots of edits, so
you should remove some of the
different ‘versions’ as you work toward
perfection. To make this process easier,
versions can be renamed, saving the
trauma of having to remember
whether it was modified1 or
modified3 that had the red-eye
removal done.
For local display there is a quick
‘full screen’ option and a good but
basic slideshow option that displays all
the images in the browser window.
Online sophistication
Another great feather in the F-Spot
cap is Flickr integration (see What Is
Flickr? box, below). Images can be
uploaded to your account from within
the browser and, joy of joys, it turns
the F-Spot tags into Flickr ones,
appending existing ones as needed.
WHAT IS FLICKR?
Flickr is a photo sharing service that has
managed to attach itself rather
effectively to the blogging phenomenon.
Users can join the service for free and
post their pictures to Flickr’s servers
(paying customers get extra features,
such as more storage space).
Each image (or folder of images) can
be designated public or private and can
also be tagged with a number of
keywords. You might, for example, take
a picture of the moon each evening from
the same spot in your garden; these
could then be tagged with the words
‘moon’, ‘garden’ and ‘August’ and the
date, and posted to your Flickr space.
Users searching for pictures of the moon
would probably then be presented with
your images, especially if their search
string also involved the words ‘garden’
or ‘August’. You can get around this by
restricting access to your pictures.
www.linuxformat.co.uk
ROUNDUP PHOTO MANAGERS
F-Spot’s tagging is second to none.
F-Spot is a very smart application. simple to use, it could really do with
It makes it easy to systematically tag some proper documentation.
your pictures, which is essential if The future’s bright for F-Spot.
you’re going to do any cataloguing. Further integration with Gnome and
You can remove images from the other Mono projects such as Beagle
library or delete them entirely, and the search and perhaps even the Tomboy
whole thing runs along at an note application are very exciting
acceptable speed. prospects, but even at this early stage,
It would benefit from a good undo
system – this is in the pipeline – but
this is a welcome addition to the
desktop ‘lifestyle’ sector.
as it stands, the system of creating
LINUX FORMAT VERDICT
new images for each change is at
least workable, but does rely on the
A product well on its way to becoming the
user to organise things. It would also iPhoto of the Linux world.
be nice to have some sort of ‘virtual’
file system to build collections for RATING 7/10
specific purposes without affecting the
library as a whole. And despite being
As well as being an
excellent complement
to the likes of F-Spot,
the service has an
excellent tool called
Organizr, which runs
entirely within a web
browser using Flash.
This makes it easy to
access and manage
your photos regardless
of your location. Flickr
is also brilliant if you
plan to use photos in
tandem with a weblog,
as images can be
hosted by Flickr but Flickr can make all your photos
accessed from, say, a Blogger account. available from anywhere.
The free Flickr account has a 20MB
per month upload limit and will display upgrading to the Pro account (at $24
only your 200 most recent images, so if per year) is recommended.
you plan on making the most of it, www.flickr.com
LXF73 DECEMBER 2005 37
19/9/05 5:56:54 pm
 ROUNDUP PHOTO MANAGERS
DigiKam
Comprehensive project.
■ VERSION 0.7.3 ■ WEB www.digikam.org
■ PRICE Free under GPL
DigiKam has been around for a
while, but doesn’t have the glamour of
some other photo management apps.
This is a shame, because it does the
job efficiently and with style. It also has
a vast suite of plugins available for
adjusting, styling and distributing
images and collections. In fact,
DigiKam is the archetypal all-rounder,
and because it works with an already
established file system – appended
with its own small database file – it’s
worth having on every system.
The user interface is based on
familiar KDE conventions: the left of
the window displays the file tree while
the main section shows either
thumbnails or large versions of each
picture. A pop-up appears when the
mouse is hovered over an image, and
this displays the EXIF data and any
notes associated with it. It’s possible to
display information just below the
thumbnail images, but this can slow
the application down quite markedly.
We also noticed that when the app
displays a date, it shows the date
when the photo was uploaded rather
“WHERE DIGIKAM REALLY
SHINES IS IN OUTPUT
AND DISTRIBUTION.”
than the date that the photo was
taken, which would have been far
more useful.
File management is very easy, but
be aware that any changes made
within the album tree will be
performed in your actual filesystem.
So creating a new album will make a
new folder on your system, rather than
just build a virtual folder and populate
it with links.
Handy deletion
The tagging system is similar to
F-Spot’s, but without the advantage of
having tags available in the side panel.
It’s still possible to assign or remove
tags to multiple selections, but this is
done through the right-click service
menu. Other excellent management
tools include a duplicate image finder,
38 LXF73 DECEMBER 2005
LXF73.round 38
which creates
a digital
thumbprint of
every image and
offers up similar
ones for manual
deletion. Beware!
The images it
finds may not be
exact duplicates,
so delete with care.
Building and
comparing the
thumbprints on our
whole collection
took just under
three minutes, and once deleted the
pictures are moved to the desktop’s
wastebasket. This is very efficient, and
we can recommend DigiKam on the
strength of this alone, especially if your
image collection has been mangled by
some other management program.
DigiKam handles EXIF data well,
and can be used to embed short
notes into the data space, so those
notes will be accessible by other
compatible applications.
Thanks to the plugin system this
application can access quite a
selection of external image
manipulation tools, such as an ‘oil
painting’ filter and various colour filters
(sepia, black and white, and so on).
There is also a handy red-eye removal
tool and options for cropping, resizing
and blurring.
Where DigiKam really shines,
though, is in output and distribution.
You can opt for exporting in HTML
gallery format, as a new album or as a
slideshow. Slideshows to be displayed
locally can have a number of
transitions (including OpenGL ones
with appropriate hardware), and the
whole thing, including audio, can be
output to an MPEG file on formats
including VCD and DVD. You can
also email images, a feature that, while
www.linuxformat.co.uk
DigiKam makes everything available from a smart, comfortable interface.
not as elegant as F-Spot’s Flickr
integration, does at least mean you
can send images to the service from
your desktop.
This is an excellent all-round
application that does most tasks with
aplomb. The fact that it doesn’t make
an enormous impact on the filesystem
means that any (small) shortcomings
in the app can be remedied by using,
DIGITAL CAMERAS AND LINUX
You can almost guarantee that any
digital camera you buy will be usable
with your Linux PC. It wasn’t always the
case, but most devices purchased these
days will be accessible as a standard
USB mass storage device – that is,
once it’s plugged in to a PC, the
operating system will recognise it as an
external hard disk and mount it to the
desktop as a normal hard disk. You will
then be able to move photos from the
camera to your disk using a simple drag
and drop procedure.
If your camera fails to mount properly,
it may mean that the OS isn’t set to
auto-mount new devices, and you’ll
have to do the job manually. To do this,
first plug the camera in – making sure it
is switched off – then open a console
and type
dmesg
This will output a ton of system
messages to the console when you
switch on your camera. If this output
includes something along the lines of
localhost kernel: Initializing USB mass
storage driver...
or
localhost kernel: scsi1 : SCSI emulation
for USB mass storage device
for example, F-Spot for Flickr
management or Kalbum for HTML
gallery building.
LINUX FORMAT VERDICT
If you have a digital camera, install this.
RATING 8/10
then you know the computer has
recognised that the device is attached. It
also says it is using SCSI emulation to
access it, which means it will be given
the device name ‘sda1’. It should now be
possible to mount the device using the
following command:
mount /dev/sda1 /mnt
The device will then be accessible
from within the /mnt directory using
your normal file manager. If ‘sda1’
doesn’t work, you may have other SCSI
devices attached to your PC. Try ‘sdb1’,
‘sdc1’ and so on until you hit the mark.
For cameras that don’t use the USB
mass storage protocols, you will need to
install GPhoto2, which is available on
virtually every Linux distribution and
provides automatic access to over 600
different digital cameras.
Applications such as DigiKam, F-Spot
and KimDaBa can automate the process
of importing photos into your central
library further and also allow the
creation of a tag that can be applied to
every image in the whole set, and that
can be used as a ‘roll name’ to make
searching more efficient in the future.
■ For more on auto-mounting devices,
read Neil Bothwick’s tutorial on page 82.
19/9/05 5:56:57 pm

GThumb
The new kid on the block.
■ VERSION 2.6.6 ■ WEB http://gthumb.sourceforge.net
■ PRICE Free under GPL
GThumb is usually installed as part
of a default Gnome system, and its
primary purpose is as a multi-format
image viewer. It does, however, have
some features that help it to rise
above that humble station.
As with DigiKam, GThumb works
with an already established filesystem
rather than creating its own hierarchy.
Its weak point has to do with access:
where other application on test here
have visual representations of the file
tree available, GThumb suffers from
having the old Gnome system
whereby only the folder you are in and
its subfolders are displayed.
To go to a higher folder, you
double-click on the two periods (..)
widget until you’re in the right place. It
is possible to manually enter a full
path (something newer Gnome
applications have done away with), but
these are still inefficient ways to work.
KimDaBa
Nice app, shame about the name!
■ VERSION 2.1 ■ WEB http://ktown.kde.org/kimdaba
■ PRICE Free under GPL
The really good thing about
KimDaBa is its ambition. The authors
have definite ideas about the quickest
and most efficient way to navigate
through hundreds and thousands of
Unusually, the slideshow runs in its own window.
LXF73.round 39
Rather than using tags, GThumb
uses a system of ‘comments’ and
‘categories’ to define locations,
people or events. The problem with
this system is that data is not
explicitly saved with the images,
meaning that if an image is moved
from one folder to another outside
of GThumb, the associated
comments and categories are lost.
However, one thing GThumb does
do is enable the creation of catalogues,
which the developers handily compare
with the playlists you might create in
an MP3 application. It’s possible to
create a catalogue called, for example,
‘Holiday’, and populate this with links to
images without moving the images
themselves. Adding pics simply
involves right-clicking and selecting
Add To Catalog… from the menu.
These can then be exported to CD or
to a web gallery, or used as the basis
images. The only problem is that their
ideas will take some getting used to.
The first few times you use it, you may
find that you’re wandering back and
forth through different lists, clicking
www.linuxformat.co.uk
ROUNDUP PHOTO MANAGERS
GThumb is a good companion for Nautilus but lacks solid features.
of a slideshow. The slideshows are of
the fixed variety with few options to
beautify the experience.
Overall, GThumb feels a little in
need of attention. The minimal tools
for fixing and displaying images work
within their limitations, but stacked up
against even the newest applications
here it doesn’t really cut it. For Gnome
users it’s a handy utility to have
around – working in conjunction with
icons wildly and wondering where on
earth the photos are. There is a
sensible feature in there somewhere, it
just needs a little more thought in the
presentation department. The problem
lies in the positioning of the icon that
will actually take you to the images; it
appears at the bottom of a list of icons
that doesn’t really change, so it’s easy
to miss.
KimDaBa stands for KDE Image
Database and, as this suggests, the
application is built on the idea of an
internal database that handles the
tags, changes and movement of
images. The benefit of this system is
that it’s really fast even when
navigating enormous collections.
When we changed the thumbnail
preview size with over 100 images on
screen it was almost instantaneous –
even when the computer was busy
doing other processor-intensive stuff in
the background. The disadvantage is
that it’s difficult to export comments or
tags to different systems. It also means
that you must remember to save your
changes – though there is an autosave
function buried in the Configure
KimDaBa dialog.
KimDaBa comes with a range of
plugins, which deal with importing,
Nautilus – but for anything
approaching serious or sustained use,
GThumb is out of its depth.
LINUX FORMAT VERDICT
A good, stable image browser that lacks
the features to make it essential.
RATING 5/10
exporting and adjusting images. Some
of these are shared with those
available in DigiKam, so the facilities in
that application are largely duplicated
here. Perhaps a decision between
KimDaBa and DigiKam will come
down to which interface you find most
useable. KimDaBa’s database is
powerful, but the interface obscures
too much during early use. The UI
becomes easier to use with time and
practice, but as a piece of consumer
software it should probably work within
the conventions that most people
expect. Perhaps splitting the interface
in two might help, keeping the
navigation tools on screen even when
browsing the thumbnails.
There is also a serious bug in the
version we tested that prevents the
thumbnail gallery from wrapping round
the window. This made for extremely
long windows running off both edges
of the screen.
LINUX FORMAT VERDICT
A little confusing on first use, but gets
better with age. And it’s really fast.
RATING 6/10
LXF73 DECEMBER 2005 39
19/9/05 5:57:00 pm
 ROUNDUP PHOTO MANAGERS
KPhotoBook
Immature but extremely promising.
■ VERSION 0.0.6 ■ WEB http://kphotobook.berlios.de
■ PRICE Free under GPL
KPhotoBook (KPB) is spanking new,
so it’s the only application reviewed
here that we had to build from source.
This was simple enough as KPB uses
mainly chunks of KDE, as you’ll see
when you start it up.
We included it here because it
combines the best bits of DigiKam with
the superior tagging skills of F-Spot.
There are plans to implement support
for the same Kipi plugins system that
DigiKam and KimDaBa use. KPB goes
further, though, with the ability to
combine tags using Boolean operators
so you can search for images of Rita
and Sue, excluding images with only
Sue or only Rita. Navigating the photo
database and tagging images is not yet
as simple as it is in F-Spot, but it’s
getting there. KPB can also lock tags
to prevent mistaken deletions.
The program uses Konqueror’s
display/rendering system, so it
Kalbum
A focused application in need of some TLC.
■ VERSION 0.8.0 ■ WEB www.paldandy.com/kalbum
■ PRICE Free under GPL
This application, written by LXF
scribe Graham Morrison, is aimed at
the export and distribution end of the
Despite its age, Kalbum has a really useable interface that gives the user
lots of information.
40 LXF73 DECEMBER 2005
LXF73.round 40
integrates well with the rest of KDE.
The familiar interface has options to
use a standard tabbed interface to
access the file tree, tagging tools and
EXIF data, or the IDEAl/sidebar system
that a number of KDE apps including
Konqueror and Amarok have recently
adopted. The sidebar system makes
better use of available space, but it
feels a bit clumsy moving from the file
browsing to the tagging interface.
Some sort of compromise – perhaps
involving the as-yet unused right-click
service menu – is really needed here
to streamline the all-important process
of tagging images.
At present, and in contrast to
many of the apps here that centre on
the viewer, KPB is strictly a
management/database application;
there are no tools for slideshows,
image editing or even for displaying
larger versions of thumbnails.
spectrum. That is, rather than being a
tool for managing an entire photo
collection, it works in tandem with the
www.linuxformat.co.uk
It’s still early days yet, but a quick
look at the project’s roadmap suggests
that the developers have ambitious
plans for turning KPB into the standard
image management app for the KDE
desktop. There are some important
bits missing at present – in fact, many
of the tools that make this kind of
software so desirable – but what is
available is smartly done and easy to
use. It will be well worth keeping an
eye on this project in the future.
file manager (or a more
comprehensive management app) to
create good-looking HTML galleries for
uploading to a web server. However, it
can be pressed into more extensive
use with a little lateral thinking. For
instance, you can make individual
albums for each month and then open
them as and when needed. Pictures
can be renamed and have notes
associated with them, and can be
moved around within the structure of
the album.
Kalbum’s forte is in preparing and
outputting complete albums for
viewing on the internet, and to this
end it integrates a decent FTP client to
make uploading as painless as
possible. As such, Kalbum may be
worth exploring if this forms a large
part of your photography habit, though
to get the most of it you should use it
as a complement to an application
such as F-Spot.
There is an unfinished air about
Kalbum, especially as the only
available template for HTML output is
looking a little dated, though it uses
tabs quite well. In truth, the whole
application is in need of a little love
and attention. The final results look OK,
but there’s not much beyond changing
It’s nowhere near finished, but
exciting nonetheless. KPhotoBook
is one to watch.
LINUX FORMAT VERDICT
An application with the potential to
become the standard KDE photo app.
RATING 5/10
the colour of the background that can
be achieved for the actual output.
That said, for static pages of images it
does the job; but stacked up against
the abilities for management access,
design and tagging that is displayed by
Flickr, Kalbum falls well short.
One thing Kalbum does well is
displaying information. The user
interface is divided up into fairly
sensible sections and so makes
accessing notes, EXIF data and other
information associated with an image
an easy job.
While not up to the job of general
image browser, Kalbum could have a
future as a smart Flickr ‘uploadr’; with
the FTP end already sorted, the
author would just need to integrate
some of the tagging features of Flickr’s
API and this would be a good, simple
and focused application.
■ Read Graham Morrison’s diary of
developing Kalbum in LXF72.
LINUX FORMAT VERDICT
Limited but focused, KAlbum could do with
some attention from its developer...
RATING 4/10
19/9/05 5:57:03 pm
 ROUNDUP PHOTO MANAGERS

PHOTO MANAGERS
THE VERDICT
As OS X and Windows have
excellent digital photo
managers in Apple’s iPhoto
and Google’s Picasa respectively, Linux
needs an application approaching the
form and function of these. Thanks to
some smart work in the development
DIGIKAM
community, it does. While DigiKam is
the clearly the best application here,
F-Spot and KPhotoBook both ave real
potential to steal the top spot.
DigiKam combines an easy to use
interface with a stupidly
comprehensive
OVER TO YOU
Did we over-de
velop our
opinions of DigiK
feature set. It doesn’t yet have the am at the
expense of your
Flickr integration that F-Spot offers, but photo app? So favourite
und off on the
there is someone somewhere LXF forums at
www.
linuxformat.co
currently beavering away on a plugin .uk.
as you read this. The plugin is in the
early stages at present (we couldn’t
get it to install, but it’s nice to know
that the issue is being tackled). makes finding the right picture easy,
While DigiKam’s developers have and this is vital for an application like
paid a lot of attention to the interface this. Efficient tagging means effective
and have added genuinely useful searching, especially when a collection
features, the adoption of Kipi plugins is counted in thousands of pics rather
broadens the application’s scope than hundreds. Throw in the class-
enormously. The facility to output high- leading Flickr support and you have an

8/10 quality multimedia slideshows to DVD

would be especially useful to a
dedicated photo-sharer, as not
everyone will have the hardware and
software available to play back a Linux-
centric one.
F-Spot deserves an honourable
mention, especially as it is such a
young project. So simple even a child
could use it, F-Spot also has the best
tagging system that we have
experienced on any platform, easily
beating iPhoto’s keywords. Combined
DigiKam: the best combination of ease of use and advanced features. with the calendar widget, F-Spot
app that well deserves a place on your
system. If the plethora of plugins that
DigiKam offers seems like overkill, go
for this one.
Not quite ready for the big time,
but also worth checking out, is
KPhotoBook. We were impressed with
the way the developers appear to be
thinking about navigation and tagging,
and look forward to some really strong
releases from them in the future.
All in all a pretty strong Roundup
this month, offering a, er, Canon of
apps that Linux can be proud of. LXF

TABLE OF FEATURES
Name Import from camera Export to HTML Export to Flickr Tags Slideshow Red-eye removal Colour correction
DigiKam ✔ ✔ ✔ ✔ ✔ ✔
F-Spot ✔ ✔ ✔ ✔ ✔ ✔ ✔
GThumb ✔ ✔ ✔ ✔
Kalbum ✔
KimDaBa ✔ ✔ ✔ ✔ ✔ ✔
KPhotoBook ✔

www.linuxformat.co.uk LXF73 DECEMBER 2005 41

LXF73.round 41 19/9/05 5:57:07 pm

 HOTPICKS

HotPicks The best new open source software on the planet!

WEB-BASED DESKTOP icons and an adequate supply of

Mike Saunders

Mike has installed and
tested more apps than
he’s had hot dinners.
As you can see.
This is the place where we get
to profile some of the hottest
software around.
Each month we trawl
through the hundreds of open
source projects that are
released or updated, and
select the newest, most
inventive and best for your
perusal. Most of the HotPicks
are available for you to try out
on our coverdiscs, but we've
provided web links if you want
to make sure you have the
very latest version.
If you know some open
source software that you’d like
us to cover, email me at mike.
saunders@futurenet.co.uk.
HOTPICKS
AT A GLANCE
Amaya ......................................47
CSSED ..................................... 44
EyeOS...................................... 42
FreeSpeak ........................... 46
Gillo........................................... 45
KScope ....................................47
Luma ........................................ 46
Mailtc ....................................... 43
Snowball ............................... 45
TruBlog .................................. 43
LOOK OUT FOR THE
HOTPICKS AWARD
Everything covered in our HotPicks
section is unmissable, but each
month we single out one project for
outstanding
brilliance. Only
the very best
is chosen!
EyeOS
■ VERSION 0.7.04 ■ WEB http://eyeos.sourceforge.net
Two EyeOS apps: web browser (an embedded Gecko here) and calculator.
Linux Live CDs and cheap USB
memory keys have led to a mini
revolution in computing: the portable
desktop. This is not in the laptop sense,
but in the way you can carry around a
full OS and your own files in your
hand – booting the OS from a Live
CD and loading your personal data
from the USB key.
Still, even this isn’t ideal: locked-
down machines (such as public library
computers) won’t let you use anything
but their default settings. EyeOS is an
astounding project to get around these
issues for the geek on the move,
running off a server and working
entirely in a web browser window.
At its heart, EyeOS is a CMS
(content management system)
designed to look and operate like a
desktop OS. Via nifty use of JavaScript
and PHP, it creates a familiar working
environment complete with file
manager, text editor and more.
It’s also easy to install: just extract
the main tarball somewhere in your
web server’s document root, change a
few permissions as explained by the
docs, and go (you’ll need PHP 4.2 or
newer). By default, there’s one
administration account already, and
“THIS IS AN ASTOUNDING
PROJECT FOR THE
GEEK ON THE MOVE.”
you can add general user accounts
with fewer privileges once you’ve got
things set up.
After the login screen, EyeOS
presents a cosmetically smooth
desktop of subtle green with grey text.
In fact, this grey on white and green
can be horribly eye-straining at times –
the default theme would benefit
greatly from stronger contrast.
That aside, the desktop has a
modern gloss, combining
round-cornered shapes with effective
tooltips. The panel along the top acts
as a launcher for EyeOS’s bundle of
programs, while a clock sits along the
bottom along with a trash can icon.
Extras, extras
So EyeOS is immediately recognisable
as a desktop OS of sorts, and the
mini-programs also work in a familiar
fashion. A calculator, text editor and
calendar make up the essentials, and
do a decent job. The more advanced
utilities include an unusual but
workable file manager, a messaging
system and a web browser. Yes, a web
browser in a web browser – but it just
uses the browser’s rendering engine in
a smaller box, so it works great.
EyeOS’s developers could have
made thing easy for themselves by
simply making these mini-apps open
in separate browser windows. Instead,
they’ve ambitiously coded in a
primitive window manager, so you can
move the programs about, resize their
windows, and so forth. Multi-tasking
leaves a lot to be desired, as does the
performance while dragging windows;
although in fairness the latter is an
issue with the host browser rather
than with EyeOS itself.
EyeOS is a truly amazing
accomplishment, pushing back the
boundaries of web-driven software
and opening up a whole new world of
portable computing. It can’t be a
replacement for a desktop OS – you’d
never be able to run a complex
graphics or sound-editing app through
it, for example. But with some work on
stability and a few feature additions to
flesh it out, EyeOS could lead to some
incredible developments. Just imagine
if Google were to build on this...

42 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.hot 42 20/9/05 11:33:17 am


WEBLOG SERVER
TruBlog
■ VERSION 0.2 ■ WEB www.trublog.org
Current projections are that by
2010 there will be more blog
postings than emails sent in any
given day. By 2015, blogs will be the
largest bandwidth consumer after
spam, accounting for 5% of all
internet traffic; and by 2020 vocal
communication will cease to exist as
everyone exchanges thoughts by
blogs. Instead of asking your spouse
“Can you pass the salt please?”, you’ll
write a blog entry about it, and
receive six comments with people
offering salt if you refer them for a
free iPod Maxi. In short, there is
plenty of demand for a new entrant
in the blog server battle, TruBlog.
“Designed by bloggers for
bloggers”, TruBlog puts great
emphasis on creating a system
that’s easy to use. It’s driven by PHP
with MySQL as the database back-
As you’d expect, the developers use TruBlog themselves to keep
people informed of its progress.
LXF73.hot 43
end, ideally running on top of
Apache. Installation is remarkably
simple: just extract the tarball
somewhere among your web
server’s HTML files, point your
browser at the resulting directory,
and the setup procedure will begin.
This nine-step, mostly automatic
process involves configuring the blog
title and owner, selecting a theme
and entering the database
connection info. Three themes are
supplied with the tarball; they’re not
exactly gorgeous but they give you a
clean and respectable page.
TruBlog supports multiple
authors and topics, exporting of RSS
feeds, commenting, trackbacks and
other features obligatory for a lively
blog. The developers have put a
great deal of effort into accessibility
and standards compliance – it
should work on nigh-on any browser,
and supports assistive technologies
for those with disabilities. This is
refreshing to see, when so many
projects simply rush for the latest
buzzwords and features without
taking into account the wider effect.
The mighty WordPress would
beat TruBlog in a feature-for-feature
match, but that’s only to be
expected for a 0.2 release. Similarly,
there are always security concerns
when using a relatively untested
system on a popular blog, but we
saw no major glitches or worries
with TruBlog in its current state. It’s
fast, clean, charmingly easy to set
up and has bags of potential.
www.linuxformat.co.uk
EMAIL CHECKER
Mailtc
■ VERSION 0.6.3 ■ WEB http://mailtc.sourceforge.net
Mail checkers are one of those
simple everyday tools that none of us
pays much attention to, but they can
be tremendously useful. Instead of
having to fire up your email client to
see if you have new messages (a time-
eating and memory-hogging event
with large GUI apps like Evolution), you
can have a small background applet
that periodically checks for new mails
and indicates this accordingly. Mailtc,
written by LXF reader Dale Whittaker,
is a mail checker that pops up an icon
on your desktop panel when new mail
is waiting to be read.
You won’t need any obscure
dependencies to compile from source
– just GTK 2.0. Optionally, you can
install the development packages for
OpenSSL and GNU SASL for extra
security and authentication features.
Mailtc works perfectly well with
window managers and desktop
environments such as KDE, Gnome
and Xfce. If you use a different WM,
you can try running Fbpanel, which
should display the Mailtc icon properly.
When first started, Mailtc needs
some initial configuration. Enter the
relevant settings for your POP3 or
IMAP email server in the clean,
sensible dialog boxes, give it an
account name (multiple accounts are
“BEING ABLE TO COLOUR
CODE MAIL ACCOUNTS IS
AN EXCELLENT TOUCH.”
HOTPICKS
By default, Mailtc is set to check
for new mail every minute, but you
can alter that if you’re conserving
every byte of bandwidth.
supported) and select an icon colour
for the panel. It may seem like an
unnecessary frill, but it’s possible to
choose a colour for the new mail icon.
This is an excellent touch, as it means
you can give priorities to different
accounts. At work, for instance, your
business-related email account can
pop up as bright red when a mail
arrives, but personal messages can be
denoted with a less urgent blue.
And that’s Mailtc in its entirety. It’s
worth looking at because it sets out to
do a job and does it well, with no
pointless animations, sound effects or
other fluff. Similarly, the configuration
dialogs are comfortable, and it’s not
tied down to any particular desktop or
WM. Give it a go if starting up
Thunderbird every few minutes is
making your swap space cry.
LXF73 DECEMBER 2005 43
20/9/05 11:33:21 am
 HOTPICKS
CASCADING STYLE SHEETS EDITOR
CSSED
■ VERSION 0.3.0 ■ WEB http://cssed.sourceforge.net
Cascading Style Sheets (CSS) has
been a hugely welcome technology for
improving the presentation of websites.
For developers, it provides a means to
change the overall appearance of a
site with a single edit; and for users, it
gives much more flexibility (the option
to select a plainer style sheet for some
sites, say, improving accessibility).
Books, tutorials, guides and tips
abound on using CSS – yet there’s
little in the way of CSS editing software.
Many of the web development tools
doing the rounds, such as Quanta and
Bluefish, focus on the HTML side.
There are JavaScript editors and XML
editors, but CSS rarely gets a look in.
CSSED aims to fill this gap,
providing a comfortable graphical
editor geared specifically towards CSS
(although it will handle other web
documents too, such as HTML and
XML, and even caters for Python, Perl
and C source files). It doesn’t require
any estoteric libraries when building
from source: if you have GTK 2.0 with
its header files you shouldn’t have any
difficulties. The standard ./configure,
make and make install (as root)
process will do the trick. Binary
“THE KIND OF APP WE
ENJOY COMING ACROSS –
FAST AND RELIABLE.”
packages are available for
Fedora Core 1 and 2, which can be
coerced into working with other distros
or converted in to Debs with Alien.
The interface
The CSSED main window has a bundle
of widget furniture in a busy (but still
navigable) layout. The ever-useful
tabbed editing panels are present to
eliminate taskbar clutter, and the icon
set picks up on your current GTK
theme. Two non-detachable toolbars
sit at the top of the screen, giving
quick access to the program’s most
commonly used features. A
particularly helpful touch is the scratch
pad, which can be toggled on at the
bottom. This provides two textboxes to
jot down miscellaneous notes during
editing, one of which retains its
44 LXF73 DECEMBER 2005
LXF73.hot 44
contents even after you’ve closed and
restarted the app. It beats the need to
have additional text editor windows
floating around.
Elements of style
Instead of writing a text-editing
component from scratch, CSSED’s
author has used the mature and
popular Scintilla editor. This provides all
the features you’d expect (copy and
paste, undo and redo, drag and drop) The CSSED interface can be cluttered on lower resolution screens, but
along with various extras. Syntax thankfully the panes are resizeable.
highlighting colours keywords in CSS,
HTML, several programming languages present (instead, it’s an HTML file without any stability scares along the
and even Apache .htaccess files, while available via the website) but when it’s way. It’s the kind of app we enjoy
code-folding is useful for collapsing fully included, it’ll truly round the coming across – fast, reliable, not
blocks of code into a single line, program off. There’s not a great deal in dependent on a truckload of weird
making it easier to concentrate on the the way of configuration options; just libraries, and supplied with solid
matter in hand. A simple bookmark the colours for syntax highlighting, the documentation to boot. The author
system has also been rolled in, which font settings and a handful of choices has a few more features in the
lets you define markers in a for the editing component. It covers all pipeline for future releases, most
document’s margin and step back and the basics, then, but we’d like to see notably a browser preview option and
forth between them via the toolbar. some options for redefining the ability to create style sheets from
CSSED can validate documents and keybindings. HTML. Similarly, as the program grows
notify the developer of syntactical Despite the early version number, in popularity, we may see it expand in
errors via the Digest panel on the CSSED 0.3.0 already looks and many more directions as users and
right-hand side. Font and colour behaves like a maturing editor, developers write their own plugins. A
dialogs assist in setting properties via delivering a healthy range of features must-try app for all web developers.
checkboxes, sliders and other friendly
means, while changes made in the
property pane are reflected
immediately in the code. Intriguingly,
CSSED sports a plugin system to
expand the program via add-ons.
There are only a few of these at
present, but they’re worth checking out.
The virtual terminal plugin, for instance,
provides a tabbed shell window next
to the scratch pad, further saving
taskbar/window list space and making
it seem a little like an integrated
development environment.
A special mention should go to the
quality of CSSED’s documentation.
Most projects in mid-development
tend to have lacking or incomplete
docs, but CSSED has a top-notch
guide peppered with screenshots and
explanations. Unfortunately, this isn’t
available through the program itself at
20/9/05 11:33:29 am

PUZZLE GAME
Snowball
■ VERSION 0.1 ■ WEB www.snowball.retrovertigo.de
We’re always chuffed to see
games involving penguins here on
HotPicks. We think that penguins
deserve a more prominent role in the
world, even taking into consideration
the fame that comes with being the
Linux mascot. Snowball is a game in
which you control a penguin (good).
But the sprite is small (hrm), and
doesn’t exhibit any real-life penguin
abilities (bah). Instead, this game’s
black-and-white bird is simply
capable of jumping its body height
without moving its legs, and
constructing ice blocks in
nanoseconds. Stick to March Of The
Penguins, realism fans.
Snowball is written in Python, and
you’ll need the PyGame libraries to
run it. Once it’s extracted, entering
./snowball.py should start the gam
up and drop you into the main
screen, from which you can configure
3D ACTION GAME
Gillo
■ VERSION 1.0-beta1 ■ WEB http://sourceforge.net/projects/gillo
Last month, as we were hunting
down the best new software to put
on our coverdisc, we came across
Gillo – a wacky 3D game involving
sinister smiley faces, magnetic balls
and other bizarreness. The more we
Three-wheeled vehicle capers as the Smiley Face of Doom stares on...
LXF73.hot 45
the keybindings or dive straight into
the gameplay. Essentially, Snowball is
a blend of platform and puzzle
elements – it’s reminiscent of
Solomon’s Key from the 8-bit days.
We must be blunt: the graphics
and sound are both pretty ropey,
with bland sound effects and
mediocre sprites scattered around
the playing area. True, this doesn’t
detract from the puzzle elements of
the game; nonetheless, Snowball
would improve with some proper
artwork, so it’s a relief that the
programmer Willi Kappler has taken
a new artist on board for the next
release. With a fittingly mellow
soundtrack, the game’s presentation
would be solid.
Controlling the penguin with the
cursor keys and Enter, your job is to
leap around a 2D platform
environment, dropping ice blocks to
played it, the more we realised that it
deserved an appearance in HotPicks.
Gillo is one of those cheerfully insane
nonsense-fests that somehow comes
together as a playable game, and
while it’s no king of longevity, it keeps
www.linuxformat.co.uk
Thanks to the built-in level editor, you can create your own perplexing
puzzles. Our heroic penguin can just be seen in the bottom left.
cordon off enemies and provide a path
for an ice ball. This ball rolls around,
aiming to reach an exit somewhere on
the screen, but only by putting ice
blocks in the correct place can you
guide it to its goal. You’re limited to
having 12 ice blocks on screen at any
one point – this is where the strategy
comes into play.
you coming back for one
more go.
Usefully, Gillo’s
developers have supplied
highly compatible binaries, so
you almost certainly won’t have
to install from source. It doesn’t
even need installation; you can just
extract the gillo-1.0beta1-bin-x86.
tar.bz2 file with tar xfvj and run
Gillo/bin/gillo3 to fire it up. If you
take the source route, you’ll need
development packages for OpenGL
and SDL installed beforehand.
Your objective in Gillo is to fire a
ball from your vehicle into a goal.
Simple as that. Oh, but the ball is
magnetic, and you have to attract it to
your vehicle’s cannon first. And the
arena in which you play has banked
edges and can flip upside-down. And
there’s a smiley face that descends
into the goal, hampering your efforts
to acquire the ball. And controlling
your vehicle is like trying to
manoeuvre a rocket-powered fork-lift
HOTPICKS
HOT GA
ME
S
Forward planning is essential, and
the later levels (there are ten in all)
are fiendishly puzzlesome. Despite
the shoddy cosmetics, Snowball has
a promising underlying gameplay
engine, and already it’s supplied with
enough levels to make it well worth
playing. Oh, and there’s a penguin in
it too. Quadruple hurrah!
truck on an
iceberg.
By sliding
around the arena,
picking up the ball
and firing it into the
goal area before your
opponent does, you score points,
thus providing you with ample
opportunity for sneering remarks.
There’s no CPU-controlled opposition
(though there is a training mode), but
with two players battling it out the
game is charmingly weird and frantic.
Visually and sonically it’s stock fare –
but the lively background music
makes the gameplay seem that little
bit more insane.
There’s one sizable flaw with Gillo,
though: by compressing the full
arena view into a small window, it’s
often hard to keep your eyes on the
ball. Ideally, some form of network
play would be introduced in future
releases, giving each player a full
screen of magnetic buffoonery.
LXF73 DECEMBER 2005 45
20/9/05 11:33:38 am
 HOTPICKS
TRANSLATION FRONT-END
FreeSpeak
■ VERSION 0.11
. ■ WEB http://home.gna.org/freespeak
The internet has changed the
way most people look at translating
short sections of text. Rather than
thumbing through dictionaries or
faffing around with limited electronic
translators, there are now websites
that translate words, sentences and
even full texts to and from a
multitude of languages. Alta Vista’s
Babel Fish is one of the most
popular translation engines, but
there are several more, and
FreeSpeak gives them a smart
Gnome front-end on your desktop.
FreeSpeak is written in Python,
and requires the PyGTK bindings for
its GUI. You may also need the
Gnome-python-extras bundle along
with Mechanoid if you want to
exploit its full featureset. As Python
and PyGTK are provided with most
popular distros, you can typically just
install by entering python setup.py
install (as root), and then
freespeak.py from the shell prompt.
When started, FreeSpeak pops
up a simple dual-paned box
LDAP MANAGER
Luma
■ VERSION 2.2 ■ WEB http://luma.sourceforge.net
The Lightweight Directory Access
Protocol (LDAP) system for managing
users, resources and other information
is best known on Linux in its
OpenLDAP form (of which Slapd is the
server). Luma is designed to make
browsing and managing data stored on
LDAP servers a fast and friendly
experience, with a Qt interface and
widgets galore. It boasts a plugin
system to provide the core functionality
and has been translated into a wide
array of languages.
To get Luma up and running, you
may need to hop around your distro’s
repositories for a few dependencies.
It’s written in Python with Qt as the
front-end, so you’ll need those
installed along with the PyQt bindings
to tie it all together. Additionally,
Python-ldap is required to interface
Python with the LDAP server. Once
46 LXF73 DECEMBER 2005
LXF73.hot 46
containing a widget
in which you enter
the text to be
translated, and a
section at the
bottom which shows
the resulting
translated text. That’s
it – no gimickry. A
useful touch is the
tabbed display, so
you can keep
multiple translations
open at the same
time without turning
your taskbar or
window list into an overgrown mess.
By default, FreeSpeak uses the
Free Translation and Alta Vista
engines, but Google is also available
depending on settings. Due to the
immense complexities of languages,
translations are typically rough and
need rewording, but they’re
acceptable for reading instructions
or translating short passages. With
Alta Vista, you can translate between
the archive has been extracted, you
can either install or run it in place by
entering luma-2.2/bin/luma.
www.linuxformat.co.uk
Translations may need rewording,
but it’s an impressive service in all.
English, French, German, Spanish,
Chinese, Japanese and others. The
speed of translation will vary
depending on your internet
connection and how loaded the
translation server is, but in most
cases it’s nigh-on instantaneous.
FreeSpeak is impressively clean,
straightforward and fast, with no
After a slow startup, Luma drops
you straight into the fray – there are
no wizards, no welcoming dialogs.
Some more polish in the presentation
would be good, then, but as it is Luma
will suit many administrators who
would rather get straight to work than
battle with superficial glitter. At the
top-left of the window you’ll find a
drop-down plugin list, which lets you
pointless bells and whistles. It’s only
as good as the translation engines
that back it up, but these are
effective enough and make
FreeSpeak a handy desktop extra if
you frequently encounter chunks of
foreign text that you need to
decipher in a hurry. Es esto
correcto, Colin?.
switch between various aspects of
LDAP server administration:
managing the address book, browsing
through objects, searching, managing
users and more, with straightforward
editing panes.
Multiple servers with their own
security settings can be defined, and a
sprinkling of admin helper utilities have
been included too – password
generation, Unix date format
conversion and mass user creation.
The browser and user management
plugins are accompanied by icons and
neatly organised boxes to make editing
easier, so while there’s little in the way
of documentation, most of the options
are self-explanatory.
Luma works well, and while it’s not
the most advanced admin tool of all
time, it’s ideal for basic management
and editing chores. Some additional
refinement would be welcome in
places, as long as it weren’t to the
detriment of its overall cleanliness and
no-nonsense design.
Handily, Luma indicates the
strength of a password as you
enter it.
20/9/05 11:33:41 am

HotPicks REVISITED
SOURCE CODE NAVIGATOR & EDITOR
KScope
■ VERSION 1.3.0 ■ WEB http://kscope.sourceforge.netx
In LXF45, we took a look at
KScope, a graphical front-end to the
source code browsing app CScope. A
command-line-based search tool,
CScope that aims to make large source
code projects, such as the Linux kernel,
easier to navigate. You can pinpoint
places where a function is called, for
instance, or search for all references to
a certain symbol. KScope adds a slick
KDE layer to this raw functionality.
Masses of development activity has
taken place since our coverage of
KScope 0.2. The most notable highlights
are: querying via a context menu; a new
manual; the ability to run in read-only
mode; an option for using external
editors; position history for going back
after search results; code-completion;
drag-and-drop support; and a spiffy
new call graph based on Graphviz.
We’ve included some binaries on
our coverdisc, but if you have to
compile from source, make sure you
have the KDE and Qt development
packages. Upon starting KScope is, you
WEB BROWSER & EDITOR
Amaya
■ VERSION 9.2.1 ■ WEB www.w3.org/Amaya
We haven’t looked at Amaya in
HotPicks since right back in LXF10,
when LXF Towers was based in
Pangaea. Amaya is the web browser
and editor for the World Wide Web
consortium, which started
development in 1996 and is still going
strong to this day. Many long-time
Linux users will wince at just hearing
its name; despite the good technology
underneath, Amaya was notoriously
ugly and stood out like an especially
sore thumb.
Not any longer, though. The old
interface has been ditched in favour of
a WxWidgets GUI, resulting in a
smooth GTK front-end when the
browser is run on Linux. It looks orders
of magnitude better. Amaya’s
rendering engine isn’t as advanced as
Gecko or KHTML (used in Firefox and
Konqueror respectively), but for
relatively undemanding sites it’s
LXF73.hot 47
KScope’s new call graph feature in the bottom right, showing a visual
representation of the code.
start a new project by pointing it to a
source code tree. The program then
presents a multi-paned view in which
you can switch between files, execute
queries and edit the source by hand.
This editing component sports code
folding and syntax highlighting,
making it effective for basic editing
jobs – but it’s not intended to be a
fully-fledged IDE.
acceptable. But the program’s real
strengths lie in its editing facilities.
Amaya functions as a versatile
WYSIWYG web page editor:
meaning, you edit the visual layout of
a page rather than the raw HTML
behind it. It supports CSS, MathML,
SVG and more. A sidebar down the
left contains collapsible sections for
changing formatting, selecting
colours, inserting math symbols and
so on. It’s clearly geared towards
experienced website designers rather
than newcomers (who’d feel more at
home with Nvu and the like),
www.linuxformat.co.uk
KScope has matured from an
intriguing bolted-on GUI into an
excellent application in its own right.
Few coders will find it of great use
for their own personal projects, with
a relatively small number of source
code files, but if you’re grappling
with a gigantic codebase such as
OpenOffice.org, it could be
supremely useful.
Editing the BBC’s snooker
page – if only our addition
(highlighted) were true...
although it doesn’t compare on a
feature scale with bigger editors
such as Dreamweaver.
Still, it’s free (as in freedom and
beer), it’s extremely stable and it
doesn’t hinder you with pampering
and pointless fluff. Amaya is a good
choice if you’re looking for
something more advanced than the
simplistic My Page Editor-type tools,
but don’t want to get bogged down
in hacking raw HTML. LXF
HOTPICKS
ALSO RELEASED
New and updated software
that also deserves a look...
■ Guten 0.3 Project Gutenberg book
reader http://guten.sf.net
■ UseBB 0.6 PHP/MySQL bulletin board
system www.usebb.net
■ Neko1.0 Compiler and virtual machine
http://nekovm.org
■ Linphone1.1.0 Gnome internet
telephony www.linphone.org
■ StormWar 0.14.1 3D strategy game
http://stormwar.free.fr/intro.html
■ Galleria 0.1 Web image gallery
generator http://qwoot.net/galleria
■ Xplorer 0.6.0 Fast file manager
http://cxplorer.sourceforge.net
■ Kalva 0.7.95 KDE video recorder
www.andreas-silberstorff.de/ktvapp
■ FONTpage 0.1 A PyGTK font viewer
www.pcbypaul.com/linux/
FONTpage.html
■ Geda suite Electronic circuit design
toolkit www.geda.seul.org
■ Ptyaim 0.5.6 Text-based AIM and
IRC client http://ptyaim.sf.net
■ Echelon 0.1 Gnome game server
monitor http://echelon-applet.
sourceforge.net
■ xlayout 0.7 Control X apps via CLI
http://freshmeat.net/projects/xlayout
■ ReactOS 0.2.7 Open source
Windows clone www.reactos.org
■ Ogrian 0.9.1 A Magic Carpet-esque
game http://ogrian.sourceforge.net
■ KGuitune 0.5 Guitar tuner
http://home.planet.nl/~lamer024/
k3guitune.html
■ GNUnet 0.7.0 P2P network (see
HotPicks in LXF68) www.gnunet.org
LXF73 DECEMBER 2005 47
20/9/05 11:33:45 am
 WHAT ON EARTH Parrot
What on Earth is...
PARROT?
Originally planned to help run new Perl code, this bird could go far.
Amias Channer examines Larry and the Perlmongers’ latest creation.
>> Go on then: what is Parrot?
Parrot is new breed of virtual machine (VM)
are compiled in real time before execution. C is a
compiled language: to make a program you write the
tailored to the needs of interpreted languages. source code, then compile and link it to get a binary,
which you then run. With an interpreted language
>> Interpreted you just run an interpreter and tell it which source
languages being
Ruby, Perl and so on?
code to run. Perl test.pl, for example.
Yes, they are languages that
>> Why use a virtual machine?
Virtual machine architecture runs on top of
your real architecture. Instead of compiling to
machine code you compile to bytecode, which is run
by the virtual machine. This separation allows greater
control over security and better sharing of resources
in exchange for a little bit of performance.
>> What’s the difference between
bytecode and machine code?
The key difference is that machine code runs
directly on a real processor while bytecode
runs on a made-up machine which is
emulated by the real machine. This means
that bytecode can be run anywhere with an
interpreter, whereas machine code only works
on one type of computer. They are both equally
headache-inducing and shouldn’t be written directly
unless you are some kind of pain-loving cyborg or
absolutely have to.
Parrot allows the compilation of source code for
a variety of languages into bytecode, which can be
run on a variety of different computers as long as
they have a bytecode-targeting compiler.
>> I don’t care what my code
runs on – what can a VM do
for me?
You can build components of a program
in several languages, choosing whichever
is the most appropriate to the task in
hand. You can then compile your
www.linuxformat.co.uk
LXF73.woe 48
components to Parrot bytecode and use them from
any of the other languages that can be compiled to
Parrot bytecode. And of course, your program can be
run or ‘hosted’ on any computer with a Parrot VM.
>> Can’t I use any language with
existing system libraries?
Well, you could, but currently only through a
separate external interface such as a database or
serialisation API. Interpreted languages have always
had problems when it comes to sharing data
structures because of their lack of a common binary
interface. Perl, Python and Ruby all use completely
different binary structures internally (for reasons of
platform neutrality) and so cannot share data
structures directly in memory without the aid of
external tools. It’s also very common for some
external interfaces to work on a subset of the
available platforms, which makes porting a bit of a
lottery. That’s why there’s so much hope for Parrot.
>> Will I be able to distribute binaries
of my interpreted programs?
A bytecode file isn’t strictly a binary, but if you mean
not distributing the source code, then yes. This can
also be a good reason to push for the development
of a Parrot targeting version of your favoured
language. If you are using GPLed modules in your
code you should check you aren’t infringing their
licences if you plan to distribute them in bytecode.
>> Wouldn’t my code be faster
without the VM underneath it?
Native code, that is, code written to run on a specific
machine, will always be faster than virtualised code.
But if you are using a high-level language, especially
an interpreted one, you have already indicated that
your priority lies in functionality, because there will
be some sort of VM involved anyway.
20/9/05 10:18:54 am
 WHAT ON EARTH Parrot

>> Why’s that?

Well, take the interpreted language PHP. This
typing is where variables have to be declared;
dynamic typing just creates the variables when you
bytecode translator is already available, so consider
the doors open.
is both a language and a virtual machine platform to assign to them.
host it. Static typing is generally more efficient of the two
in terms of speed and memory usage, because >> You mentioned that Perl was the
inspiration for Parrot.
>> And you don’t think the
presence of a VM slows
the compiler will know more about
what the code is doing and can
Yes. The idea was to create a self-hosting
environment for the new Perl 6. Self-hosting means
things down? optimise it further. This is great for written in its own language, like Java and C.
Given the falling cost of operating systems where you need Most of the people who work on Parrot are Perl
hardware, performance to know exactly what is hackers, and quite a lot of the ideas for
speedups can be more happening, but it’s not so Parrot are logical extensions of
effectively achieved using faster chips good for high-level Perl paradigms. This self-hosting
than by streamlining code. And to business logic environment would also allow
compensate for any performance hits, processing where Perl users to hack the insides of
when you use code that runs in a good Perl in Perl where previously this
VM, upgrades and platform changes
should be seamless.

>> There must have been

other open source VMs
before Parrot came along. Are they
any good?
There is no other open source VM with the same
features, and that’s one of the reasons why the
Parrot project was started in the first place. Though it
is cross-platform and multilingual, it was conceived
by Perl programmers primarily to host the new Perl
6 they were working on. If there was an alternative
out there, they would have used it.

>> Java and .NET are VMs. What

about them?
For a start, unlike them, Parrot is truly open source.
This allows it to be cross-platform, because there is
no underlying platform allegiance. Even though both
Java and .NET are cross-platform, they don’t
necessarily treat each platform evenly, so moving to
another platform may not work. Parrot is expected
to outperform both Java and .NET, and current tests
show it leading the pack in VM performance
although there is a lot more code to be added,
which may well change this.
Also, Java and .NET have their own native
languages –

>> Don’t tell me: Java and .NET?

Indeed. You’re a genius. Parrot, on the other you want rapid coding and clear business logic, was only
hand, is more language-agnostic. Parrot is designed regardless of how it actually works in memory. possible if
for interpreted, dynamically-typed languages in you knew
general, unlike most other VMs, which are designed
for specific statically-typed languages. >> Will Parrot talk to Java or .NET?
It will, but not for a while. The view from the
some C.
Most of the Parrot
Parrot’s nest seems to be that this can be added at development goes on in the

>> I type with one finger. Is that

static or dynamic typing?
a later stage, when bridging between a non-
dynamically-typed VM and one with type checking
Perl 6 internal mailing list.

Er… no, it’s nothing to do with keyboards. Static becomes simpler. A proof-of-concept Java to Parrot
>> How did we
get to Parrot
from Perl?
“PARROT SHOULD OUTPERFORM The moniker derives from an
April Fool’s joke, which alleged

JAVA AND .NET, AND IS MORE that Perl originator Larry Wall
and Guido van Rossum (Python
CROSS-PLATFORM THAN BOTH.” leader) were to merge Perl and Python into
a new language called Parrot. Simon Cozens,
>>

www.linuxformat.co.uk LXF73 DECEMBER 2005 49

LXF73.woe 49 20/9/05 10:18:58 am

 WHAT ON EARTH Parrot
>> a senior Perl hacker and poster of the April Fool,
then took over as Parrot’s lead developer, and the
name has stuck – because in some ways it does
merge Perl and Python.
>> I thought Perl died out with the
failed dotcoms?
Nope, Perl is still going strong. Quite a few of the
core developers have moved on to Parrot
development and very little new code is added
to the Perl 5 code base, although CPAN
modules for the Perl repository continue to
roll in.
>> If people hate Perl, will
they hate Parrot?
Parrot is very Perl-centric, and it will be hard
to get away from the Perlisms. You can use
Parrot without writing a single line of Perl,
but you will have to have a working Perl
installation to use it. Most Linux distros
include Perl, so this shouldn’t be a problem
It is possible to use Parrot with other
languages than Perl – in fact, the idea with Parrot
is that you can stick to your chosen language and
get the benefits of others. Your language can
become interoperable with other languages and
platforms that previously it might not have been able
to use. But you will need to have a Parrot bytecode
compiler for it to achieve this.
>> OK, so what other languages
compile to Parrot bytecode?
Loads, although there is nothing especially stable
or complete yet and there won’t be until the core
matures. At the moment you can try prototypes of
Perl 6 of course, Parrot Assembler (PASM), Java,
Jako, Integer BASIC, Scheme, Forth, TCL, Ruby,
Python, Ook, Lisp and Befunge bytecode compilers,
but don’t expect them to work very well yet. If your
favourite language is not on this list, it could be if
you start nagging its developers now.
>> Urgh, Assembler! I thought Parrot
was about high-level languages?
Calm down dear, it’s only a virtual machine. Anyway,
you don’t have to use PASM: there will be more
comfortable ways to generate Parrot bytecode in the
near future. PASM is the most complete option at
the moment and is being used to write unit tests for
the Parrot VM, which is the only serious work done
with Parrot at the moment. When Parrot matures,
PASM will disappear into the sidelines and nobody
will use it unless attempting some dark voodoo
hackery. You’ll probably never actually use PASM
directly but it’s handy to have the option to use it. As
Larry Wall says: “There is more than one way to do it.”
>> Do you think this grand idea of
Parrot will take off beyond the
Perl community?
Parrot is already gaining popularity and is starting to
connect with other open source projects as both a
50 LXF73 DECEMBER 2005
LXF73.woe 50
VM and an embedded language engine, so there
should be no need to even think about nailing it to
its perch. Leo Tötsch, the current pumpking, pushed
out the current release, Serenity (or 0.23), on 5
August 2005 and work continues apace.
>> Pumpking?
Sorry, it’s a Perl thing. From the pool of
developers, a ‘pumpking’ is chosen.
This pumpking oversees the addition
of new code to the main release
versions. Linus Torvalds could
be considered the
Linux 2.6 pumpking. Apparently the phrase was
coined when a pumpkin-shaped keyring that was
once passed between mythical programmers back in
the digital dreamtime. The keyring contained a USB
key that allowed updates to the back-up media that
“PARROT IS STARTING TO CONNECT
WITH OTHER OPEN SOURCE
PROJECTS AS A VM AND LANGUAGE.”
contained the code they worked on. The pumpkin
king was the one who had the key.
>> What’s with all the weird names
and acronyms?
The linguistic playfulness of the Perl community is a
legacy from Mr Wall. He’s a deeply odd but
fantastically gifted individual, who among other
things is a linguistics expert, prodigal quipper and
wearer of frighteningly loud shirts. His playfulness
and linguistic skill have imbued Perl with an
extremely flexible – some might say lethal – syntax
that appeals to the more linguistically-minded
developer. The result is that Perl is the only language,
programming or otherwise, used regularly in poetry
www.linuxformat.co.uk
as well as shell scripts, resembling line noise and
beautifully structured web automation projects.
>> Perl people seem pretty quirky.
Perl as a programming language celebrates
the values of clever quirkery, flexibility and
pragmatism. This focus has attracted the weirder end
of the hacker spectrum, people who love to revel in
their oddness at any opportunity. These people
helped to create the guts of the language, so Perl
took on their preferences and eccentricities. Perl
developers often do not follow the traditional
computer science degree route, and tend to
see coding problems in a less mathematical,
more linguistic, higher-level way.
>> Will this lunacy infect
Parrot as well?
Darn tootin! It’s what’s made Perl so
successful! There are Perlmonger groups
all over the world developing their own
recursive acronyms, writing Perl modules that
turn your code into pictures of camels and finding
time to drink copious amounts of beer. Miraculously,
this digital bedlam has bred strong, vibrant
communities that churn out top-notch code.
>> But how does that make for
good code?
I suppose if you have a group of happy people who
enjoy each others’ company and are constantly
trying to out do each other technically you get good
code. The tomfoolery creates a pleasurable
environment and, most importantly, a light-hearted
approach makes discussing possible conflicts much
easier. Perl excelled during the dotcom years
because you could rapidly prototype and adjust your
code to fit with the constantly changing business
practices. A more rigorously defined language would
not have provided the flexibility to change so quickly.
>> What are the new features of
Perl 6 ?
One of its main features is better object orientation
(OO). Perl 5 got a lot of flak for its hackish but
rather clever approach to OO, but Perl 6 will have
proper keywords for declaring classes so you don’t
have to keep fixing typos in basic constructors. This
should keep your code smaller and better organised.
A major criticism of Perl 5 was that it encouraged
bad coding practice by having too many quirks.
Indeed, it’s common to use a special ‘strict’ grammar
to turn off some of its more dangerous features
when using it in production environments. To avoid
this, Perl 6 will be strict by default, but you can turn
off the ‘strictness’ easily enough.
20/9/05 10:19:00 am

>> Wasn’t there some kind of
problem with Perl 6?
Yes, the initial plan to extend Perl 5 all the way to
Perl 6 failed. Because of its high-level nature, Perl is
not a great language for writing compilers, so a new
development language was chosen: Haskell.
Although similarly high level, it has much better
formal debugging tools, which are essential when
writing something as sophisticated as a compiler.
This interim compiler is called Pugs, and is intended
as a bridge that will allow the development of a
working Perl 6 grammar. Most of the Perl 6 designs
have been implemented in Pugs, and it is finding its
way into production systems. Pugs has a limited
lifespan because it will culminate in a port to Perl 6
on Parrot. See www.pugscode.com for more info.
>> Will my Perl 5 program run in
Perl 6?
Pretty much, yes, unless you’re doing something
really weird in which case you will probably know
exactly why. Perl 6 will assume Perl 5 behaviour
unless it detects Perl 6 constructs (class and module
keywords) in your code. You might want to port to
Perl 6 at a later date, especially if you are writing
object-oriented code.
>> Can’t I just have Perl without all
this VM stuff?
No, the new features of Perl 6 are what necessitated
Parrot’s development. The insides of Perl 5 are so
twisted around themselves that a complete rewrite is
needed. There will still be a Perl 5 for quite a while,
but you should expect the main body of Perl
development to move to Perl 6. Most of CPAN will
move over, and so newer concepts may only get
coded in Perl 6, although Perl 5 has such a large
userbase that this will not happen for a long time.
>> Was Parrot coded in Parrot or
Perl 5?
Ah. Actually, it was written in C.
>> Why is Parrot coded in C if it’s
supposed to be self-hosting?
Self-hosting creates a chicken and egg situation. To
get this environment to work it needs to be coded in
C and compiled to machine code first of all. From
there it can be used to rebuild itself in its own
language. It is not unusual for compiler writers to
switch languages during the development phase.
This is possible because many compiler writers use
special toolkits that generate the same functionality
in several languages allowing them to choose the
one that makes the fastest or safest machine code.
>> Isn’t C a bit obsolete? I thought
Parrot was all new and shiny?
C is getting on a bit but it’s still the best language
we’ve got for this sort of job. It runs on the widest
range of platforms, has the biggest pool of
development talent and is mature enough to safely
generate small and fast machine code. No other
LXF73.woe 51
WHAT ON EARTH Parrot
language has all three of these
characteristics. Remember also that
the code that really counts here is the
machine code, and C compilers and linkers
create the best machine code.
>> I use Mod-Perl on
Apache. Is Parrot of
any use to me?
Oh yes. The programmer Jeff Horwitz
is working on Mod_parrot, but it’s
only at version 0.3 so is nowhere
near production grade yet. Still, the
possibilities with Mod_parrot are great.
It will allow multiple languages to
manipulate the same data structures
from inside the same web server. You can
use whichever language is most appropriate to
the task in hand instead of the one that seems
to cover the most situations.
>> What else uses Parrot?
At the moment only seriously early
adopters should use Parrot for real work.
It is rumoured that some of the core Parrot
developers are using it in production environments,
but of course they practically live
in the Parrot development lists and are thus fully
aware of what works and what doesn’t. If you are
thinking of using Parrot for something serious you
should read all the PDDs and subscribe to the
Perl 6 internals list first.
>> What are PDDs?
Parrot design documents. These are
a formal guide to what’s happening and an
historical record of the Parrot design and
implementation process. This means
that anyone who missed out on the
original development can know
why a certain decision was made.
The format is basically POD,
but they can be easily converted to
other formats. To read them visit
http://dev.perl.org/
perl6/pdd.
>> So what can I
do with Parrot
right now?
In short, play, learn and bide your time.
Parrot is still relatively new, so it’s
easier to pick up its nuances
now before the complexity
gets layered on top of
it.Also, if you learn how
it works you might be
able to help develop
a Parrot targeting
version of your
favourite language or
help write documentation. For more information
visit www.parrotcode.org. LXF
www.linuxformat.co.uk
20/9/05 10:19:01 am
 CRACKING THE KERNEL

COVER
FEATURE

52 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.kernel 52 23/9/05 3:54:47 pm

 CRACKING THE KERNEL

Kernel hacker Greg Kroah-Hartman looks inside the

latest kernel to show you what makes your box tick.

“From a technical standpoint, I believe the kernel

will be ‘more of the same’, and that all the really
interesting stuff will be going on in user space.”
Linus Torvalds, 2001

THE AUTHOR espite what out With the release of the 2.6 kernel

Greg Kroah-Hartman works full time

for Novell on the Linux kernel. He’s
been writing kernel drivers since
1999, maintains udev and hotplug,
and also looks after various kernel
D benevolent bazaar
keeper thought, the
Linux kernel is proving
to be a very interesting and
important part of all Linux
series at the end of 2002, the kernel
engineers unleashed upon the world
an extremely stable, multiprocessor
operating system that runs faster
than any other operating system, on
subsystems. Greg has also been systems. At the lowest level, the a wider range of processor types.
known to be
kernel is responsible for making What other operating system can
an avid
the hardware (the stuff you’ve just scale from a tiny MP3 player or
Gentoo user,
but how that spent so much money on) work embedded robot controller to the
pans out now properly. It enables you to run largest supercomputers – some with
he works for your favourite web browser, email over 512 processors? All of this is
Novell is client, IRC program, 3D shooter only possible thanks to the kernel
anyone’s
guess...
and more, all at the same time.
Without it there’d be no Linux.
itself: everything on running on top
of it is just fluff.
>>

www.linuxformat.co.uk LXF73 DECEMBER 2005 53

LXF73.kernel 53 23/9/05 3:55:08 pm

 CRACKING THE KERNEL
Cool kernel features
>>
We should all take some time to get better acquainted with the kernel. The
latest version has some great additions that will excite all kinds of users.
The 2.6.13 kernel was released on
28 August, 2005. It added some
200,000 new lines of code, and had
the same number of lines modified.
Most of the changes were bugfixes
and stability improvements; but 2.6.13
also has some new and interesting
features that can make using a Linux
machine better – for everyone.
FOR DEVELOPERS
Inotify
For such a simple concept (letting the
user know about all changes that
happen to a file), Inotify has had a lot
of press. This is all down to one tool:
Beagle. Beagle is a Gnome application
that enables you to instantly search
your machine for anything – whatever
the type of file it is, wherever it is on
your machine. With Beagle, you can
easily find documents, emails, web
history, IRC conversations and more
(for more information on
Beagle, see http://
beaglewiki.org).
In order for tools like
Beagle to run efficiently
and know what has
recently changed on your
machine, they need
some method of
knowing what has
changed on the system,
rather than having to
The Linux kernel can as easily
run a supercomputer as a
small, embedded device.
54 LXF73 DECEMBER 2005 www.linuxformat.co.uk
LXF73.kernel 54
trawl the whole hard disk every time
you run a search. Inotify is the kernel
part that provides this functionality to
user space. Older versions of the
kernel had a mechanism called
Dnotify, but it was cumbersome to use
and did not work very well. Inotify fixes
all the problems in Dnotify, and allows
individual files as well as entire
directory trees to be watched.
Kexec and Kdump
The Kexec and Kdump features have
been around for a long time in
distribution kernels, but it’s only now
that they’ve found a place in the main
kernel tree. Kexec allows a user to
load and run another Linux kernel
without requiring the currently running
kernel to shut down to the BIOS –
saving a lot of time at reboot. Kdump
then uses this feature to grab kernel
dumps of system in the event of a
kernel crash, enabling developers to
analyse the system at a later date.
Users who want to keep kernel
upgrade disruptions to a minimum (no
one wants to wait and watch the BIOS
re-probe the SCSI bus) can use Kexec
to keep their downtime to a minimum.
FOR USERS
Bind and unbind
For a device to work in Linux, it needs
a device driver to control it. The kernel
chooses which device is bound to
which driver through a set of tables
that the individual drivers register with
the kernel. Before now, users had
never been able to control this
matching up between device and
driver, except by unloading a driver,
which causes all devices controlled by
that driver to be released.
The 2.6.13 release includes a way
for users to bind and unbind individual
devices to specific drivers. This can
allow administrators of large systems
to disable a single device, load a
newer driver and then bind the newer
driver to that device, without
disconnecting any other devices that
were controlled by that original driver.
Just think of the hassle involved in
upgrading a large disk system and
you’ll realise this is a Very Good Thing.
This module also enables you to
choose which device is controlled by
which driver. Some devices can be
controlled by multiple drivers, and the
kernel will just use the first one it sees.
Manual binding lets you use your own,
specific driver, which might work a lot
better for your device than the default
one chosen by the kernel.
Devfs is dead
The devfs filesystem has been disabled
in 2.6.13. This has been done to flush
out any remaining users of this in-
kernel device filesystem before it is
finally removed entirely. People who
are still using devfs should move to
udev (see LXF66). It may seem rather
odd that we’ve highlighted a feature
being disabled as a step forward, but
the truth is that devfs sucks. It’s a
huge amount of bloat, makes it harder
to read through the kernel code and
the sooner we’re rid of it, the better.
Better I/O queuing
The ‘Completely Fair Queueing’ I/O
scheduler (arguably the best overall
disk scheduler so far) has been
enhanced to allow better handing of
timeslices. Also in this area, two new
kernel system calls have been added
23/9/05 3:55:49 pm
 CRACKING THE KERNEL

to allow user programs to modify the
I/O priorities of different processes in
the system. This lets administrators
tweak system performance of different
programs much more easily.
FOR EMBEDDED
Xtensa
A whole new CPU architecture has
been added to the 2.6.13 kernel:
Xtensa. Xtensa is a 32-bit embedded
processor that can be placed within
system-on-chip solutions with a wide
range of embedded uses – phones,
PDAs, set-top boxes, MP3 players and
all sorts of other common gadgets.
With this new architecture, Linux
now supports 24 major processor
types, and many more sub-processor
types. Good news for embedded
devices. For those who like to brag,
Linux supports more different
processors than any other operating
system ever has. Combine that fact
with the realisation that Linux supports
more devices out of the box than any
other operating system, and it’s clear
why the Linux kernel is recognised
as the most flexible and powerful
kernel around.
Execute in place
The ext2 filesystem now enables
programs to be executed in place
instead of having to be loaded into
memory first. This is very useful on
embedded systems that have a limited
amount of RAM but a large amount of
Flash or ROM (like a mobile phone).
Adjustable frequency
We set the 2.6 kernel’s interrupt
frequency to 1,000Hz, which caused a
few complaints from a range of users.
Now this frequency can be configured
at compile time to three different
values, depending on what the
machine is used for.
■ 100Hz This is best for low-power
devices.
■ 200Hz Ideal for servers.
■ 1,000Hz Optimum for desktop
systems, or other systems that require
fast, interactive responses to events.
Voluntary pre-emption
Those users who are not comfortable
with a fully pre-emptable kernel now
have another option: voluntary
pre-emption. This allows the kernel to
pre-empt processing at a number of
known ‘explicit pre-emption points’
that have been selected to reduce the
maximum latency of rescheduling
different programs. Applications have a
faster response time, albeit at the >>
expense of a slightly lower throughput.
Interrupt
frequency
can now be
tweaked.

WHAT’S COMING SOON?

With luck, the 2.6.14 kernel will have been released by
the time you read this (it’s expected sometime around
the beginning of October 2005), and will contain these
nice new features:
■ relayfs
Sometimes the kernel developers need to get a lot of
data out of the kernel very quickly and with minimal
side-effects. For example, when you’re trying to debug
the kernel, you don’t want the debugging code to
modify the timing of the code you are trying to debug.
Relayfs was written to solve this problem, and a
number of external kernel projects have been using it
for quite some time in order to help them with their
data transfer issues. The Linux Trace Toolkit is one, and
now that this feature has been added to the main
kernel tree, LTT has come a little closer to acceptance.
■ ipw2100 and ipw2200 drivers
It seems that almost every other new laptop being
produced these days uses the Centrino
chipset for wireless capability. The main
kernel tree can now support these
devices without forcing the user to dig
through a SourceForge site, download
some source files and figure out how to
build and load a kernel driver just so they
can get their new laptop to work properly.
With the acceptance of these two drivers and
a larger wireless network infrastructure, it is
now much easier to add new wireless drivers to
the main kernel tree.
■ DCCP network protocol
As if Linux didn’t support enough network protocols
already, a new one has been added. This protocol is
so new that the 2.6.14 kernel is the first one to
actually be used in the real world to test it. DCCP
stands for Datagram Congestion Control Protocol and is
much like the existing UDP protocol – with some
important differences. For more details on this protocol,
and what it’s going to be used for, see www.icir.org/
kohler/dcp/draft-ietf-dccp-spec-11.txt.
■ Removal of the PCI IDs database
Prior to this kernel release, every kernel contained a list
of strings describing all possible PCI devices that existed
in the world. It did this just to make a few /proc files
look pretty, and it wasted a lot of space. This file has
been removed, so the wasted kernel memory can now
be used for something more useful. For those users who
really liked the pretty names, they can still be accessed
through lspci.
■ ThinkPad hard drive protection system
Most new IBM/Lenovo ThinkPad laptops contain a small
motion detector that can determine if the laptop has
slipped off of a table
and is about to come
crashing down on
the ground. If the
device detects the
laptop is in
freefall, it
quickly shuts
down the disk
drive to help
prevent any
nasty data
loss issues.
Linux now supports this very nice feature with this
new driver. But kernel programmers being what they
are, they’ve added a few new features that are not
found in any other operating system. It turns out that
the device that does this detection can be used to
detect what angle the laptop is being held at. So they
have hooked the device up to the mouse pointer, so that
you can hold the laptop in your hands and tilt it around
to push the mouse in different directions.
Once the initial fun of this feature wore off, they
realised that they should just export the tilt information
to user space so that any program can access it. This
has spawned a whole range of tiny programs that show
the laptop rotating in a window in real time according
to the way it is being held. Hopefully something useful
will come of this feature some day.
■ Fuse
Finally, it is now possible to write a filesystem in Courtesy of International Business Machines Corporation. Unauthorized use not permitted.
userspace, using whatever language you want (Python
anyone?). Fuse is the magic that allows this to happen,
exporting the internal kernel virtual filesystem to user
space for all to play with. A number of user space
filesystems are already written using Fuse, including an
SSH filesystem and a tar filesystem. See the main Fuse
development page (http://fuse.sourceforge.net).
As with all new features provided with kernel
releases, you must enable the feature and build it into
your kernel at compile time in order for it to work
properly. For instructions on how to build, install and
boot a new kernel, please look at the excellent kernel
build HOWTO at www.digitalhermit.com/linux/Kernel-
Build-HOWTO.html.
The IBM ThinkPad parks
the hard drive head within
500 milliseconds when it
detects sudden motion.

www.linuxformat.co.uk LXF73 DECEMBER 2005 55

LXF73.kernel 55 23/9/05 3:55:53 pm

 CRACKING THE KERNEL
>>
“HALFWAY COMPETENT IN C? YOU CAN
GET TO GRIPS WITH INOTIFY STRAIGHT
AWAY TO ADD WATCHES TO FILES.”
MANUAL DRIVER BINDING – USING THE BIND AND UNBIND FEATURE
Here’s a simple example of how to use
the new bind and unbind feature in the
2.6.13 kernel. You’ll need a USB Flash
device, or any other device that is
controlled by the usb-storage kernel
driver. First off, build a kernel (see URL at
the bottom of page 55) and select the
CONFIG_BLK_DEV_UB and CONFIG_
USB_STORAGE drivers (you can build
them both as modules, or into the main
kernel image).
When you use this new kernel, plug in
your USB Flash disc. Since the ub driver
is found first on the kernel load path, it
is bound to your device. This can be seen
by looking at the /sys/bus/usb/drivers/
ub/ directory:
$ tree /sys/bus/usb/drivers/ub/
/sys/bus/usb/drivers/ub/
|-- 1-1:1.0 -> ../../../../devices/pci0000:00/
0000:00:1d.7/usb1/1-1/1-1:1.0
|-- bind
56 LXF73 DECEMBER 2005
LXF73.kernel 56
Get started here
Are your programming fingers itching to go? Delve into the new kernel and
explore two of its finest features.
OK, so we’ve been covering lots of
interesting theory so far – and you’re
probably waiting to get to the
hands-on ways you can start using
new kernel features now.
Well, here you go: if you’re half-
way competent in C, you can get to
grips with Inotify straight away, learning
the new kernel functions to add and
remove watches on files.
If you’re a bit less intrepid, see the
Manual Driver Binding box (below) to
learn how you can customise the
driver that gets loaded when you
insert a USB Flash drive.
Into Inotify
The Inotify feature added three new
kernel syscalls:
int inotify_init(void);
int inotify_add_watch(int fd, const
char *path, __u32 mask);
|-- module -> ../../../../module/ub
`-- unbind
Notice the bind and unbind files in this
directory. In order to unbind a device
from a driver, simply write the bus ID of
the device you wish to change to the
unbind file.
echo -n “1-1:1.0” > /sys/bus/usb/drivers/
ub/unbind
The device will no longer be bound to
the driver.
$ tree /sys/bus/usb/drivers/ub/
/sys/bus/usb/drivers/ub/
|-- bind
|-- module -> ../../../../module/ub
`-- unbind
To bind this device to a different
driver, you first have to check that the
device is not controlled by another
driver. To check this, look for the driver
symlink in the device directory:
$ tree /sys/bus/usb/devices/1-1:1.0
www.linuxformat.co.uk
int inotify_rm_watch(int fd, __u32
mask);
Intended as a replacement for
Dnotify, Inotify works with something
called ‘watches’. A watch is a pair of an
object and a mask that describe an
event that the user wants to receive
notification events of. The object is
either a file or a directory (as
represented by an open file descriptor)
and the mask is a bitmask of events.
The different types of events that can
be monitored are:
■ IN_ACCESS The file was accessed.
■ IN_MODIFY The file was modified.
■ IN_ATTRIB The metadata was
changed.
■ IN_CLOSE_WRITE The writeable
file was closed.
■ IN_CLOSE_NOWRITE The
unwriteable file was closed.
■ IN_OPEN The file was opened.
/sys/bus/usb/devices/1-1:1.0
|-- bAlternateSetting
|-- bInterfaceClass
|-- bInterfaceNumber
|-- bInterfaceProtocol
|-- bInterfaceSubClass
|-- bNumEndpoints
|-- bus -> ../../../../../../bus/usb
|-- modalias
`-- power
`-- state
Then simply write the bus ID of the
device you wish to bind into the bind file
for that driver:
echo -n “1-1:1.0” > /sys/bus/usb/drivers/
usb-storage/bind
And check the binding was successful:
$ tree /sys/bus/usb/devices/1-1:1.0
/sys/bus/usb/devices/1-1:1.0
|-- bAlternateSetting
|-- bInterfaceClass
|-- bInterfaceNumber
■ IN_MOVED_FROM The file was
moved from location x.
■ IN_MOVED_TO The file was moved
to location y.
■ IN_CREATE A subfile was created.
■ IN_DELETE A subfile was deleted.
■ IN_DELETE_SELF Self was deleted.
If the user wants to monitor all
events, a handy IN_ALL_EVENTS
macro is defined, which includes all
the above event flags combined.
To create a watch and register it
with the kernel, a file descriptor must
first be created with a call to the
inotify_init() function, like this:
int fd = inotify_init();
Then, using this new file descriptor, you
can add a watch to it by calling the
inotify_add_watch() function:
int wd = inotify_add_watch(fd, path,
mask);
where fd is the handle returned from
the call to inotify_init(); path is the
path to the file or directory that you
wish to watch; and mask is the type of
event you want to monitor.
The inotify_add_watch() function
returns a handle to a watch descriptor,
which must be held on to in order to
clean up properly.
|-- bInterfaceProtocol
|-- bInterfaceSubClass
|-- bNumEndpoints
|-- bus -> ../../../../../../bus/usb
|-- driver -> ../../../../../../bus/usb/drivers/
usb-storage
|-- host2
| `-- power
| `-- state
|-- modalias
`-- power
`-- state
In order for a device to bind
successfully with a driver, the driver
must already support that device. This is
why you can’t just arbitrarily bind any
device to any driver. To help with the
issue of adding new device support to
drivers after they are built, the PCI
system offers a dynamic_id file in sysfs,
so that userspace can write in new
device IDs that the driver should bind to.
23/9/05 3:55:56 pm
 CRACKING THE KERNEL

FOUR WAYS TO CONFIGURE YOUR KERNEL

You should be root when compiling the kernel in /usr/src/linux (where it’s
normally installed) so you have the access rights to change files. Editing the
.config file in /usr/src/linux allows you to enable or disable individual settings by
name. This is how we do things in this feature, because it means you don’t need
to hunt through user interfaces to find options!
Running make config prompts you to choose the setting for each configuration
option. This is great if you want the entire kernel just as you want it, but slow if
you only want to change a few things.

Running the make menuconfig program loads a simple text-based GUI where GUI addicts can run the make xconfig program, which loads a Qt-based GUI
you can select the options using your cursor keys. This is usually the best choice, where the options are neatly organised under a tree structure and you can use
particularly for the times you’re working on an old server. your mouse to enable what you want.

To remove a watch that is already
in place, use the inotify_rm_watch()
function, thus:
int retval = inotify_rm_watch(fd,
wd);
where fd is the original file handle
returned from the call to inotify_init()
and wd is the watch descriptor
returned from the call to
inotify_add_watch().
Once a watch has been registered
with the kernel, a simple read() call to
the file descriptor is used to retrieve
events based on that watch (and all
other watches that have been
registered for this process).
The structure of the data read
from the kernel is described in the
following C struct:
struct inotify_event {
__s32 wd; /* watch descriptor */
__u32 mask; /* watch mask */
__u32 cookie; /* cookie to
synchronize two events */
__u32 len; /* length (including
nulls) of name */
char name[0]; /* stub for possible
name */
};
All of the fields are pretty self-
explanatory, with the exception of
cookie. If this field is not set to 0, it
will tell user space that multiple events
happened at the same time on the
same object. An example of this is
renaming a file. If we were watching a
directory where a file was renamed
mv foo baz, two events would be
generated: IN_MOVE_FROM and IN_
MOVE_TO. They would both have the
same cookie value, which allows user
BEHIND THE SCENES
space to co-ordinate the events.
The file descriptor returned by For more details on the kernel design
inotify_init() allows select() and decisions that the Inotify developers
poll() to be called on it, so a blocking went through in creating this system,
read() is not necessary, which would check out the Inotify documentation in
the file Documentation/filesystems/
tie up a program’s thread.
inotify.txt.
When the program that had the
original file descriptor returned by
inotify_init() exits, or if that file
descriptor is closed, all watches that
were registered with the kernel are
destroyed and cleaned up properly.
For a very simple example
program that shows how to register for
events and read events as they
happen, see the inotify-utils package
that can be found at www.kernel.
org/pub/linux/kernel/people/rml/ >>
inotify/utils.

www.linuxformat.co.uk LXF73 DECEMBER 2005 57

LXF73.kernel 57 23/9/05 3:55:57 pm

 CRACKING THE KERNEL

The quest for stability

>>

“The problem we’re having at
present is that I don’t think the
kernel quality is as high as it
should be. We’re getting quite a
few regressions sneaking into
each release.”
Andrew Morton, OSDL, Aug 2005
Frankenstein patches, bugfixes, two trees becoming one... How the kernel
development process has changed for the better in five short years.
Back before the 2.6 kernel was
released, it was easy to determine
which kernel development branch was
stable and which one was the
development tree: the second digit of
the version number told all. Even
numbers were stable and odd
numbers were development, so the
2.0, 2.2, and 2.4 trees were stable and
the 2.1, 2.3, and 2.5 trees were
development ones. Each branch had
individual releases on it, so 2.4.12 and
2.4.13 would be different releases of
the 2.4 branch.
That has all changed: now there is
only one tree, the 2.6 tree.
Backport bungling
Back during the 2.5 development
process (from late 2001 to 2003),
several major changes were added to
the development kernel. These
changes were so good that users and
distributions alike really wanted to use
them, even though the kernel was still
at an unstable, developmental stage
(ie the odd-numbered 2.5).
The big enterprise distribution
vendors duly got their engineers to
backport these features to their
custom 2.4-based kernel versions.
Application vendors needed to be able
to determine which kernel features
were enabled in these custom kernels.
The problem came when users of
these distros were forced to rely on
their distributions to support these
Frankenstein-style patched kernels, as
the kernel community wanted nothing
to do with them. Over time this
became very hard to maintain, and
turned into an engineering nightmare
(just compare the Red Hat Enterprise
Linux 3 and SUSE Linux Enterprise
Server 8 kernel source trees with the
kernel.org trees on which they’re
based for an example of this).
After almost a year of development
of the 2.6 kernel series, the Linux
kernel developers met to talk about
what they wanted to do for the next
development series (2.7).
It turns out that all of the
developers really liked things just as
they were, and didn’t want to have the
same problem that the 2.4 kernel had
with the backport mess – which, by
the way, helped stretch the 2.6
development cycle out to over
two years.
When they huddled round and
looked at how the 2.6 kernel was
working, they decided to leave it alone
and state that this was both the stable

NINE RULES FOR STABLE KERNEL DEVELOPMENT

The rules on what is accepted into the -stable releases are The change must be accepted by the relevant
quite simple:
8 subsystem maintainer.

The change must follow the normal rules of how a

1 The change must be obviously correct and tested. 9 change is submitted to the kernel (as documented
in the file Documentation/SubmittingPatches in the
The change must not be bigger than 100 lines,
2 including context.
kernel source tree).

3 The change must fix only one thing.
The change must fix a real bug that bothers people (not
4 a potential bug that hasn’t bothered anybody yet).
The change must fix a problem that causes a build error,
5 an oops, a hang, data corruption, a real security issue,
or some “oh, that’s not good” issue. In short, it must be
something critical.
No ‘theoretical race condition’ issues may be fixed,
6 unless an explanation of how the race can be exploited.
The change must not contain any trivial fixes in it
7 (spelling changes, whitespace cleanups etc).
The -stable kernel tree is also run with a public
review cycle, which is started when the
maintainers decide they have enough changes
for a new release. All the potential changes
are posted to the Linux kernel mailing list,
and sent to the individual kernel
maintainers responsible for the area that
the change modifies. At least 48 hours
are allowed to pass, for any
developer to object to any
individual change. Once the time
has expired, and no one objects, a
new -stable kernel is released.

58 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.kernel 58 23/9/05 3:56:00 pm


and the development kernel at the
same time.
Of course, there was a downside:
not having a development kernel (or
not having a stable kernel, depending
on how you looked at it) meant that it
“THE BACKPORT MESS
STRETCHED THE 2.5 CYCLE
OUT TO OVER TWO YEARS.”
was up to the Linux distributors to
better test and stabilise individual
kernel releases for their customers.
Out of date...
So the kernel developers started to
release new kernels every two or
three months, with a relatively large
number of changes in each one. This
worked out for a few months, until
people realised that if a security
problem or a simple bugfix was
needed for a specific kernel release, it
could be a number of months before
an updated kernel release happened
with these fixes in it. And when that
kernel release was done, it also had a
number of new changes in it, some of
which also needed new bugfixes...
LXF73.kernel 59
To get round this, a ‘-stable’ kernel
release tree was created. This release
is run by two kernel developers who
package up a small kernel release
every other week or so, and adds
nothing but obvious fixes and security
updates to the latest kernel release.
These releases are marked by a fourth
figure in the release number; version
numbers now follow the format
‘2.6.x.y’, where y is the -stable kernel
release number. For example, the
2.6.13.1 release is the first -stable
release of the 2.6.13 kernel branch.
Release tweaks
That was great, but the gap between
main kernel releases seemed to be
growing. It took one month to go from
2.6.1 to 2.6.2, but we’re now up to
almost four months between releases.
Also, it didn’t seem that anyone was
actually testing the intermediate
‘release candidate’ kernel versions that
were being released between the
major releases.
Something had to change. What
happens now is that in the first week
after a kernel release, all the different
subsystem maintainers submit their
major changes to be merged. These
might be new features, bugfixes or
rewrites of sections. At this point a real
‘release candidate’ kernel is released,
and all the developers settle down and
work on making this release stable.
Bugs are tracked using the site
http://bugzilla.kernel.org, and
developers are temporarily barred
from adding new features.
Within a few more weeks the
kernel is stable enough to be
released, and the whole
cycle starts over again.
This should provide a
short release cycle
and provide a better
shakedown of
>>
any bugs that
show up.
www.linuxformat.co.uk
CRACKING THE KERNEL
BYE-BYE BITKEEPER, GOOD MORNING GIT
During the 2.5 development process, the
kernel developers adopted a source code
control tool called BitKeeper. This
program provided a distributed way of
handing changes and kernel trees from
traditional source code control systems
(like CVS and Subversion) and was
designed specifically for the needs of the
kernel developers.
Unfortunately it was a closed source
program, but the kernel developers were
able to use it for free. Many electrons
were wasted in arguing if the Linux kernel
should be developed using a closed
source tool, but in the end it enabled a
Linus Torvalds only used to
lot of good things to happen.
publish kernel updates every
Previously, the kernel was developed
few weeks, a frustrating system.
by sending individual changes, via email,
to Linus Torvalds, the kernel maintainer. He would apply some of them to his copy
of the kernel, and every couple of weeks or so would issue a new development
release. This meant that the individual developers had to wait a relatively long time
to see whether their changes had been accepted what other changes had
happened during this time.
This worked pretty well for the first ten years of kernel development, but was
really starting to show problems as the kernel grew and grew. When BitKeeper
began to be used by a number of core kernel developers, two important things
happened: there were nightly snapshots of the state of Torvalds’ kernel tree; and
each individual change was sent to a mailing list, including a description of why that
change was made.
The change was nothing new for most other open source projects that used CVS
(“Welcome to the nineties, kernel team,” was how one group put it), but for the
kernel, it was a radical change. Torvalds no longer reviewed each individual change,
but started to rely more heavily on the subsystem maintainers of the different
portions of the kernel to enforce the proper rules on what should be accepted.
Pulling the rug on BitKeeper
Eventually, all of the people who complained about BitKeeper being a closed source
program were vindicated in a way when BitMover, the company behind BitKeeper,
decided to stop the developers using it for free. The kernel developers had to stop
everything and look around for a tool that would work properly for them. They
didn’t see anything they liked, so Torvalds decided to write his own tool, which he
called Git. Git is a distributed source code control system that works much like
BitKeeper did, but is tuned exactly for the Linux kernel development process. It has
the following attributes:
■ It allows for completely distributed development (no central server like CVS
or Subversion).
■ It can apply a change quickly.
■ It can handle the large size of the kernel source tree, and the number of changes
that happen over time (the kernel development process is constantly increasing the
rate of change as the code base gets larger, which is unlike any large open source
project in the past).
Because it has a
relatively cryptic
command line interface,
a number of wrapper
programs have been
created to help make
using Git easier (such as
Cogito). Since the
2.6.12-rc2 kernel
release, the kernel
developers have been
using Git with no
complaints. For more
information on
getting and using Git The removal of BitKeeper and emergence of
yourself, go to Git threatened the unity of Linux kernel
www.kernel.org/git. development. See LXF68 for the full story.
LXF73 DECEMBER 2005 59
23/9/05 3:56:06 pm
 CRACKING THE KERNEL

Into the future

>>

Kernel developers are taking a stronger line on closed source – but they’re
just as keen that the new development model doesn’t shut out innovation.
In the past, the Linux kernel is proving to be very effective, as no which enables more developers to see
developers have looked the other way Linux company wants to be taken to the code and work on it, getting the
when kernel modules were created court over this issue. benefit of the open source
and the source code was not made Now to the third deterrent. Most development process.
available for them. But that has all complaints about the kernel For a more detailed description of
changed in recent years. There are development process come from why Linux does not have a stable
now three ways that the kernel
developers handle such instances.
First is a flag that is that is enabled “THE CHANGE IS PUSHING
whenever a closed source kernel
module is loaded into it. This lets the DEVELOPERS TO BECOME
kernel community easily identify and
refuse to support all users of such
BETTER ENGINEERS.”
modules, as it is impossible to
determine exactly where the problem people working on drivers outside the in-kernel interface, see the file
might be (in the closed source section, main kernel source tree – closed Documentation/stable_api_
or in the rest of the kernel). source developers, for example, such nonsense.txt at http://lxr.linux.no/
Almost all new kernel functions are as NVIDIA. source/Documentation/stable_api_
explicitly marked by the compiler as These complaints are usually about nonsense.txt.
‘only to be used by GPL code’. This how the internal kernel interfaces are
prevents any closed-source module always changing, unlike other Will 2.7 ever happen?
from using these interfaces, and slowly, operating systems that have a well Back when the kernel developers
over time, will choke off all defined and stable in-kernel interface decided to stick with one kernel tree,
functionality from them. A good for them to use. a number of big features were
example of this is the driver core and The Linux kernel programming mentioned that might lead to a 2.7
sysfs interfaces, which are not interface changes frequently for a tree to fork from the main kernel. Two
available to closed source modules. reason: the kernel developers routinely of those features were four-level page
change how the kernel works tables, and a rewrite of the kernel’s
Letter of the law internally, fixing bugs, security time subsystem. However, the 2.6
As if that weren’t enough to dissuade problems, or extending it in new ways. kernel tree now has four-level page
errant coders, legal action has been To do this in an efficient manner tables (allowing virtual memory usages
initiated by the kernel they do not keep old interfaces, but in the terabytes), and there’s a rewrite
developers where closed remove them or change them, then fix on the cards for the kernel’s time
source modules have up all uses of these interfaces at the subsystem, with code waiting to go
been shipped by Linux same time. This lets the Linux kernel into a future kernel release.
distributions (as it is not stay as small as possible without It turns out that by not having a
legal to redistribute a closed breaking any drivers that are development kernel tree, where
source Linux kernel module, contained within the kernel source developers are free to tinker with
as per the rules of the GPL tree. It also forces developers to move whatever they want, the developers
licence of the kernel.) This their code into the main kernel tree, are forced to be very careful with what
they do. Although there have been
JOIN IN! some major changes to the way that
the kernel works, any improvements
Do you want to help out with kernel development, but don’t http://selenic.com/mailman/listinfo/kernel-mentors. This will
have only come about by steady,
know where to begin? If you have a solid understanding of help developers learn the proper rules of the road when it
incremental change.
the C language and know how to build and install a custom comes to getting their changes accepted into the main kernel
kernel, try looking at the Kernel Janitors project at source tree. This has enabled these very big
http://janitor.kernelnewbies.org. It contains a list of easy, And if you’re keen to do it all on your own, you should start changes to happen within the middle
‘janitor’-type projects that are very helpful to developers who out by reading the Linux kernel coding guidelines at http://lxr. of a ‘stable’ kernel without users
are just starting out. The mailing list for the project will help linux.no/source/Documentation/CodingStyle. Then check out noticing any bad side-effects. The new
you find out proper procedure and format for generating a the directions on how to submit individual changes (http://lxr. model is pushing the kernel
kernel change and sending it off for inclusion. linux.no/source/Documentation/SubmittingPatches) and whole
developers to become better
For developers who just need a gentle nudge in the right new drivers (http://lxr.linux.no/source/Documentation/
engineers, which can only lead to a
direction, the Kernel Mentors project is available at SubmittingDrivers).
faster, more stable operating system. LXF

60 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.kernel 60 23/9/05 3:56:11 pm

 INTERVIEW RICHARD HIPP

THE LXF
INTERVIEW

perfectionist Richard Hipp

LXFtrumps
If you’ve heard of the database engine SQLite and
would like to contribute… good luck. Its author
Richard Hipp is very, very picky. He tells us why.

L
Created the SQLite earning SQL and database about sponsorship from AOL,
ase
embeddable datab ones, normalisation are difficult coding standards and life in the
tem po pu lar in ph
sys
gly
PDAs, and increasin Won topics for computer public domain...
in desktop software. students, but how much
the Best Integrator Source award harder is it to learn how to LXF: Let’s start at the beginning:
UK LTD

en
Google/O’Reilly Op his work.
at OSCon 2005 for develop your own database Dr Hipp before SQLite. When did
44
MOVES

management system then release you start getting interested in

AGE eri ca n
Am it to the close inspection of the databases?
NATIONALITY
URTESY: WINNING

UX 12
YEARS USING LIN 7
wider world? Dr Richard Hipp did RH: When I came out of graduate
OGRA MM IN G LANGUAGES 1 just that when he created SQLite school with my fresh PhD in 1992 –
PR l work
S On ly 5 tha t stil
NUMBER OF PC 1 cup
(pronounced Ess-cue-ell-ite), which
TAKE
DAILY COFFEE IN a time
is one of the most successful public LXF: That was in computer science,
at
TOP TRUMPS CO

D 1 pa ir
SANDALS OWNE We’ve
domain software projects around. wasn’t it?
rket’s saturated.
HE SAYS “The ma ady”. We caught up with Richard at RH: Yes, I took a Master’s degree in
source products alre
got excellent open
OSCon and had a poolside chat electrical engineering and computer

62 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.hipp 62 16/9/05 4:59:55 pm


ON STANDARDS
“People send me patches and things, and I
appreciate the idea, I really do, but I think,
‘You haven’t really though this through’.”
science. So I came out of graduate
school and at that time there was this
enormous glut of people applying for
jobs as faculty, university. And so
there would be an opening for one
position and there would literally be
500 applications. And of those about
475 were more qualified than me, so
I quickly recognised where this was
going and I just decided to go in
business myself doing consulting
work. I developed a good practice
where I would work with a client for
years on end... they would have a
hard problem and I would develop
solutions to problems that really
hadn’t been solved before, not your
IT kind of jobs, but engineering types
of computation.
I was working on one of these
[problems], and you know, businesses
that buy services are never a
monolithic entity, there are always
factions within the customer
organisation, and one of them just
insisted that this thing be driven by a
database. And because of legacy
constraints it was an Informix
database, running on legacy
hardware and an HP-UX system.
Informix ran fine on HP-US, and did
everything it needed to do, but
sometimes they would power the
system down and bring it back up,
and the Informix database would not
come up correctly. Then the program
that I was writing, together with
another group, would get error
messages and couldn’t connect to
the database. Because I was doing
the user interface I got the blame
even though it was the Informix
server that didn’t come up! We talked
about it and said, “What we really
need is an embedded SQL database
engine that just reads data out of the
file and still handles the SQL.”
That project ended and we didn’t
take action on it in the end, but
during a period when I didn’t have
any work later on I said, “I’m going to
get busy and write that. I’m curious
how an SQL database engine works
LXF73.hipp 63
so I think I’ll just write up my own
database engine cause that’s always
a great way to learn how things
work.” And that’s how SQLite started.
LXF: So that’s how you got into it...
RH: Yes, I did not study database
theory, I had zero courses in
database. When I was at university I
was browsing the library once and
came across a copy of... the SQL-89
Standard I guess it was, and I read
over it and I thought, “Well, that’s
nice and all, but I don’t see any
application to what I’m doing,” but
then I came back to it ten years later
and what I was doing had shifted and
it made some sense.
LXF: Was it always planned to
throw the first release away? Was
it pretty dire, or something, is that
what you’re saying?
RH: No, it wasn’t, I mean, version 1.0
used GDB for the storage so it had
multiple files, you had one file for
each table and index. And it didn’t
have transactions because GDB
doesn’t support that. Also there were
a lot of things I couldn’t do because
GDB is a hash-based thing, it doesn’t
keep things in sorted order, so I
couldn’t take advantage of indices
using any quality constraints, there
were a lot of permutations there. I
really needed to go to a btree-based
system. I looked at BerkelyDB, and I
spent a day or two days just reading
the documentation, which went on
and on. After a couple of days of
reading that stuff I thought, “This is
really complicated”– you know, the
API – “I think I’ll just write my own.”
So I pulled Cummings off the shelf
and read what he had to say about it,
sat down and designed my own btree
layer. And that led to version 2.0.
You know, I’m still learning while
I’m doing all this, so version 2.0 was...
well, I did a lot of things wrong with it
initially, but it eventually got to be
pretty good – by about version 2.7,
2.8 it was pretty nice. Everything was
www.linuxformat.co.uk
INTERVIEW RICHARD HIPP
still a string. But it was good enough at
that point for people to notice it. PHP
picked up on it –
LXF: 2.8 it was then, wasn’t it?
RH: Yeah, they got 2.8. Apple first
noticed when it when it was at version
2.7 or 2.8. A lot of people started
using it.
But then AOL came along and
they wanted to use it for some of their
products, and they said, “Hey Richard,
we need to support international
languages and we need some other
features.” I’d been planning to do
version 3.0 and fix all the problems
that I had, so I told AOL, “I’m gonna
do that, it’s just a matter of time – it’s
gonna be about a year, I’ll have
something by June of this year”– this
was 2004. And they said, “Can we pay
you to expedite your schedule?” So I
said, “Sure,” and gave them a number.
LXF: And did you hit the deadline?
RH: Well, Dan Kennedy worked with
me closely on it. Dan used to work for
a company in Brisbane that used
SQLite, and he had sent me some
patches, which were really, really good
– he does really good work – so I got
him to help me and together in a
couple of months we went from
nothing to version 3.0. We wrote the
btree back-end and rewrote all kinds
of things and worked around the clock
for two months pretty much but we
made the schedule. And that was
version 3.0.
LXF: SQLite 3.0 was very, very
different. The file format changed,
which is probably a result of you
writing it from scratch, and AOL
obviously provided some funding –
RH: AOL permitted me to devote full
time to it, because I didn’t have to do
other things to make a living, and it
also permitted me to pay Dan to work
with me on it.
LXF: Is that help still ongoing
from AOL?
RH: No, that was a one-time thing. In
fact AOL doesn’t even pay for support,
though because they did give us that
funding, whenever they call I jump
through hoops to help them. But
technically that contract has long since
expired. I’m led to understand that
they are very happy with the product
and continue to use it, though I do not
know exactly what they’re using it in.
LXF: So in terms of the file format
changing it’s obviously quite a bit
different from version 2.0. Was it
really necessary to rewrite the
btree layer?
RH: Well, no. I just incorporated all the
lessons learned, is what it came down
to, and I wanted to relax the
constraints. In the old version I had a
hard limit of 16MB, and now there is a
practical limit of 264 [MB] or
something like that. There were a lot of
lessons learned and I had been
accumulating these over the previous
couple of years: “When I redo the
btree layer I’m gonna do this
differently.” But I couldn’t just do it
incrementally because each one of
those changes would have been
incompatible and I wanted to preserve
compatibility; you know, I thought,
“When I go to 3.0 and make that
clean break I might as well get this all
in here at once.” And that was the plan.
LXF: It almost seems a shame that
Apple and PHP picked it up before
version 3.0.
RH: Apple started working with it at
2.8 but it’s three-point-something
that’s in Tiger. They made the change
because they could do it in time. But it
didn’t come out in time for PHP.
But version 3.0 is in the PDO [see
the PHP tutorial this issue on page >>
90] stuff of PHP.
>>
LXF73 DECEMBER 2005 63
16/9/05 5:00:57 pm
 INTERVIEW RICHARD HIPP
>> LXF: Tell us about releasing SQLite
into the public domain.
RH: Well, version 1.0 was GPL. And it
had to be because it depended on
GDB, which was GPL. So we were
coming up on version 2.0, and I
thought, “What licence should I use? I
could do LGPL, but it’s kind of
restrictive, or I could do a BSD-style
licence.” But I looked at that and
thought, “What’s the point of the BSD
licence? What does it really give you?
Why not just put it in the public
ON RELEASING INTO THE PUBLIC DOMAIN
“Putting the code into the public domain,
that was a big step. It was sort of like
marrying off your firstborn.”
domain and be done with it?” Because
with the BSD, anybody can take it and
use it for whatever they want, so just
give it to them, make it explicit...
renounce this whole copyright thing
altogether. So that’s what I decided to
do. Now, since that time a lot of
lawyers have come to me and said,
“Well, there’s liability, someone could
come and sue you,” but I’m sceptical of
these arguments.
LXF: Well, it’s in the public domain!
RH: Yes, that’s not going to play well
to a jury in my mind. Someone pulls in
some public domain source code and
incorporates it in a larger application,
and that larger application fails and
somebody gets hurt and they’re gonna
blame me for that? That’s a real
stretch. Even in an American court
that’s a real stretch. So I’m not real
worried about it.
LXF: So how did you find going
from GPL to public domain? Or
were you the only person really
contributing to the code?
RH: I was the only contributor, so I
could do what I liked. At that point
when I made the transition every
single line of code came from me.
LXF: How big was it at the time?
RH: Maybe 20,000 lines of code.
LXF: So people who submit code
for SQLite now sign something?
RH: I’ve got a form that they have to
fill out, dedicating it to the public
64 LXF73 DECEMBER 2005
LXF73.hipp 64
domain. It just goes on and on about
how they renounce all future rights to
this and all of that. It’s based on the
Creative Commons licence.
LXF: Do you get many contributors
to the project?
RH: I went in to the CVS tree two
weeks ago and did a study to see how
much code had been contributed by
whom. I counted lines of code, and
69% of the code came from me.
Thirty per cent came from Dan
Kennedy. The other 1% was divided
among I think six other contributors,
the largest of which was half a per
cent. So it’s really mostly me there.
LXF: Would you say that perhaps
the public domain licence turns
people off?
RH: From contributing? I don’t know. I
can see how the BSD licence seems
to really hurt OpenBSD. My theory is
that Linux took off when OpenBSD did
not simply because Linux is GPL and
people feel free to contribute to it with
the certainty that no one was going to
abuse their work.
But I think the real reason that not
many people have contributed to
SQLite is because I’m so incredibly
fussy about the code that goes into
the core. And that’s more to do with
me than anything. People send me
code all the time and I really do
www.linuxformat.co.uk
appreciate it, I really do, but I have
very, very specific ideas of what the
code should look like, how it should be
commented, how it should be
indented... And it’s more than how it
looks, it’s how it’s structured. A lot of
people send me patches and things,
and I appreciate the idea, I really do,
but I think, “You haven’t really thought
this through, you haven’t simplified the
code to the extreme, you haven’t
refactored it three or four times...” I like
to make code as simple as possible.
LXF: That’s one of the things that
people always talk about with
SQLite. It’s one of the few open
source projects that has very
good documentation and that is
actually very well formatted and
commented. You can actually
jump into the code and understand
it, which is a real surprise in
today’s world.
RH: That is my goal…
LXF: And does that have anything
to do with it being public domain?
I mean, if your code wasn’t
commented, there’d be a lot more
reliance on you, but by making the
code completely stand by itself –
RH: People should not and do not
have to call on me. Yes, that’s true. I
don’t know that it was a conscious
choice when I started doing it, but I’m
aware of that now.
LXF: You’re just a picky
programmer, are you?
RH: The reason I’m so fussy about the
code and insist on commenting on it
so well and making it so simple is to
compensate for my own mental deficit
– I have a notoriously bad memory
and if I don’t structure the code very
carefully and make it very easy to
read, I’ll forget what it does in a couple
of weeks!
But also, when I was first looking at
databases I was curious about how
databases and database engines work.
I went reading about things like
PostgreSQL on the internet, and it
takes six or nine months of study to
really understand what this does. I
wanted my program to be something
that somebody could just pick up, who
was just curious about how these
things work, read through it and say,
“Now I understand better how a
database engine is supposed to work.”
LXF: Which is I think one of the
goals of free software: to be able
to learn from other source code.
RH: Yes, I try to keep the comments
helpful. If the comment exists as a
huge hack I’m thinking, “Well, why is it
in here? Take it out.”
LXF: What kind of new features
are you working on for SQLite 4.0?
RH: Actually I’m really hoping that I
never get to 4.0. The only reason I’d go
16/9/05 5:01:21 pm

to 4.0 is if we had an incompatible file
format change.
LXF: So in ten years’ time it will
be 3.9.9.9.6.3...
RH: Or whatever, yes! Right now I’m
making some changes to the query
optimiser. It’s picky about choosing
indices; it needs to do a better job of
making good choices. Traditionally
SQLite has been pretty weak in that
area. If you write the query correctly
you’ll do a real good job with it but it
doesn’t spend a lot of time analysing
the query and trying to find the best
way to do it. Whereas I hope that the
big commercial client–server enterprise
database systems spend a lot of time
trying to optimise these queries.
I’m actually getting a bit of funding
from a company and that helps me to
work on things without having to worry
about making a living otherwise.
Once I get [the query optimisers]
going well, I want to get in some
semblance of referential integrity and
support. That’s just stuff I’m doing, I’m
not getting paid for it.
LXF: Do you ever see yourself
supporting SQL 92 queries; heck,
99 – let’s go the whole hog!
RH: I’m not trying to make a product
out of this. I sell services, not products.
And the people who are coming to me
for services are using SQLite in
embedded devices. Their primary
concerns are low memory usage and
low power consumption. So I’m really,
really focused on maintaining both of
these because they’re good customers
– they keep coming back to me for
consulting help and I want to make
sure they stay happy. A lot of the SQL
99, 2003 – whatever the latest entity
is – would cause [the device] to bloat
up and run slower so I’m really going
to resist that as much as I can.
LXF: SQLite is really, really fast,
and it’s also really, really small. If
it came down to it, which would
win out?
RH: I don’t know. The embedded
customers want both. Fast in the sense
of low power consumption. It does its
work quickly and then it finishes and
stops draining the battery. Small,
because they’ve got small memories,
both Flash and RAM, and they’re
accounting for every byte. “Oh Richard,
can you save us 12 bytes here?” “Sure,
why not?”
LXF73.hipp 65
LXF: Presumably, memory capacity
is going up a lot faster than
battery power.
RH: Probably so. But as memory
capacity goes up they start putting
[SQLite] into smaller and smaller
devices. So now it’s no problem to put
it in a set-top box, but to go into your
cellphone, wristwatch or smartcard,
which is what people are doing with it
now, you’re back to square one again.
LXF: That simplicity comes across
in the way that SQLite handles
stuff. For example, you say, “This is
an integer, this is a big integer, this
is a string”, and it says, “Yeah,
whatever.” As far as I’m aware it
ignores the lot. Is that a feature or
something you want to change?
RH: I think of it as a feature. I like it
very much because I do most of my
development work for clients in
scripting languages, mostly Tcl/Tk, and
it does the same kind of thing.
LXF: Just to clarify: it treats
everything as a string?
RH: Version 2.0 treats everything as a
string. Version 3.0 makes the
distinction between integers, floating
point numbers, strings and blobs
[binary data objects]. But it allows you
to put any of those four things in
different rows of the same column. In
a traditional SQL database you declare
a column to be ‘short int’ and can only
put a short int there. With SQLite you
can declare a column ‘short int’ and
put anything you like in there.
LXF: If you can put a string into an
integer field, is it really an integer
field? That’s the question.
RH: Well, to my mind the whole
business of all the strong type in SQL
is a bug in the design of the language.
I contend that it was put there initially
so that implementers could determine
in advance how much disk space they
needed for a single row. Because they
were unwilling to go to the extra
trouble of writing disk access routines
that could have variable record sizes;
it’s much harder to do. So it’s a part of
the implementation showing through –
LXF: Outdated, really.
RH: It’s a database: it’s designed to
store data, not to tell you how to
format your data. That’s my view. There
are opposing points of view, to put it
nicely. But I come from a scripting
www.linuxformat.co.uk
INTERVIEW RICHARD HIPP
language background and it made a
lot of sense to me just to allow you to
put anything you want into any
variable, because that’s what you do in
a scripting language.
LXF: So that ‘feature’ is by design.
RH: People keep coming to me and
saying, “When are you going to fix
that?” Well, it’s not broken! I think a lot
of it is just institutional resistance. SQL
has always been done this other way,
and now I’ve sort of extended the
design to allow you to do these other
things and it’s “We’ve never done it
that way, this isn’t really SQL”.
LXF: Where do you see your role in
SQLite being?
RH: I want to keep doing what I’m
doing right now – just kind of
maintaining it. SQLite doesn’t take a lot
of maintenance work because it
doesn’t give a lot of problems.
I’m trying to build up a lot of other
people to support it so that if I get hit
by a bus on the way home someone
else can take it over. Dan Kennedy is
more than capable of doing anything
that needs to be done, he’s sort of my
backup at the moment, but it would
be nice if there were a bunch of
people that understood the code well
enough that they’re really comfortable
just going in and fixing problems.
LXF: Is that what your goal is?
RH: I’d love to find a bunch of Dan
Kennedys. I guess I’m just really fussy
and I’m looking for the very best talent
that’s out there, and the very best
people are already working on other
things. It would be great if I could
develop some more people, and get
some people to contribute. You know,
my name appears nowhere in the
source code.
LXF: “If you have complaints, just
email dan.kennedy@...”
RH: Putting the code in the public
domain, that was a big step. It was sort
of like marrying off your firstborn. It
was really hard to do because I put so
much time into that and I really did
release it, I intentionally went through
and removed my name. It’s more than
just a legal thing: I really let it go. And
I’m trying to continue to do that, to let
it go. And that sounds incompatible
with this pickyness I have in terms of
who can contribute, but it’s not,
because I’m not trying to protect my
code, I’m trying to protect the public’s
code. LXF
Go to
www.linuxformat.
co.uk/mag/hipp.html
for more of Richard’s
thoughts on XML, MySQL,
and whether SQLite has a
place in enterprise.
LXF73 DECEMBER 2005 65
16/9/05 5:01:29 pm
 FREE VS OPEN
Free
freedom
as in
Is open source really the same as free software?
Paul Hudson doesn’t think so...
G
NU, which stands for GNU’s
Not Unix, is the name for the
complete Unix-compatible
software system I’m writing so
that I can give it away free to
everyone who can use it.”
And so begins the GNU Manifesto,
written by Richard Stallman to ask for
help and support at the beginning of
the project. That was 20 years ago,
and although we’re still without the
GNU/Hurd OS he embarked on, free
software is more widespread than ever
before. But things are changing.
THE FOUR PILLARS
OF FREEDOM
The four things that a licence must
grant to be considered free:
■ You must be free to use it freely as
you see fit.
■ You must be free to read the source
code and accompanying resources.
■ You must be free to be able to
change the source code as you want.
■ You must be free to distribute the
source code and binaries to others.
It should be clear from reading this list
that free software is inherently also open
source software, because you must be
able to read, change and distribute the
source code as you please. However,
not all open source software is free
software, because you may not have
rights to modify or distribute it.
66 LXF73 DECEMBER 2005
LXF73.versus 66
The movement for open source
software, a term that many people
interchange unthinkingly with the
phrase free software, can be traced
back to 1997. That’s when Bruce
Perens and the Debian Project ratified
the Debian Social Contract, which
included the Debian Free Software
“THE OSI BELIEVES THAT THE TERM ‘FREE
SOFTWARE’ IS TOO AMBIGUOUS TO BE
USEFUL, AND IS IN LONG-TERM DECLINE.”
Guidelines (DFSG). These guidelines
detail what is acceptable for use in the
Debian distro – software must include
source code, must not discriminate
against persons or groups, must allow
modification and so on.
Freeware Summit 98
The hacker and writer Eric Raymond
soon crystallised the essence of the
DFSG into the infamous ‘Cathedral
and Bazaar’ presentation he gave at
the Linux Kongress in May of the
same year. Then, in January of 1998,
Netscape released the source code to
Netscape Navigator, citing Raymond’s
presentation as a key driver.
With all this buzz about free
software, in April 1998 O’Reilly Media
set up a special Freeware Summit in
www.linuxformat.co.uk
Palo Alto, California.
The attendees
included Larry Wall (Perl), Brian
Behlendorf (Apache), Guido van
Rossum (Python), Eric Allman
(Sendmail), Paul Vixie (Bind), plus
Linus Torvalds, Eric Raymond, and Tim
O’Reilly himself – all notable hackers
and speakers who had released work
under free licences. The result of this
meeting, and the others preceding it,
was the creation of the Open Source
Initiative (OSI), a group that aimed to
promote and expand open source
development.
Having the terms open source and
free software bandied around in the
same discussion certainly leads to
confusion, so you may well wonder
why the group needed to coin the
term open source at all. Though in its
narrowest sense it meant code being
open to view, it was taken on to give
the movement a less ambiguous,
more business-friendly face. In the
OSI’s own words, “the Open Source
Initiative is a marketing program for
free software. It’s a pitch for ‘free
software’ on solid pragmatic grounds
rather than ideological tub-thumping.
The winning substance has not
changed, the losing attitude and
symbolism have.” One of the central
beliefs of the OSI was and still is that
the term free software is too
ambiguous to be useful, and they will
happily tell you that free software is in
long-term decline.
The “winning substance” has
indeed changed, because it turns out
that the attitude and ideological tub-
thumping are what made our
movement great in the first place.
Without them, open source is just
another kind of software development.
Frantic semantics
It is increasingly apparent now that
free software and open source are not
the same thing. Not convinced? Well,
first you need to forget the OSI’s FUD
that free software is confusing or
ambiguous, because if you think about
it you’ll see that open source is just as
bad – as is just about every other
combination of words that try to get
22/9/05 10:46:27 am

ON LICENCES
Although licence propagation is a
serious problem, equally we don’t
believe that everything can be boiled
down to just two licences: we love the
BSD licence as much as the next geek.
our point across. This is why you find
things like FLOSS, which stands for
Free/Libre Open Source Software, as
people keep on trying (and failing) to
nail down exactly what we mean. The
Windows source code got leaked over
the web last year, effectively making its
source code open to whoever wants it.
But download it, read it, modify it or
even think about doing any of the
above and you’ll probably wind up in
jail. I don’t just want open source code:
I want the freedom that ought to
come with it. In Richard Stallman’s
words, “a person who has grasped the
idea of free speech, not free beer will
not get it wrong again.”
With petty things like naming out
of the way, let’s focus on the actual
substantial differences between open
source and free software: the attitude
and the symbolism.
Attitude isn’t an easy thing to discuss,
but we could put it succinctly by saying
that open source is a great way to
develop good software quickly, whereas
free software is a great way to build a
long-term, free computing environment.
Linus’s Law, as defined by Eric Raymond,
is: “Given enough eyeballs, all bugs are
shallow.” That’s quite true, and happens
to be the central tenet of open source:
let everyone who wants to take part in a
project do so, and presto chango you’ve
got a development team raring to go.
Much to Microsoft’s annoyance,
open source happens to work
remarkably well as a development
model. But given that Microsoft largely
ignored the internet until it had taken
off, it should come as no surprise that
That’s a recorder, not a bong.
LXF73.versus 67
it has a hard time adapting to the new
paradigm: open source has really hit
the big time thanks to the widespread
rise of the internet.
However, open source has brought
with it a plague of truly biblical
proportions: licence proliferation.
Whereas the Free Software movement
aimed to centralise the community
around one or two software licences
(the GPL and the LGPL), the OSI now
recognises approximately 60 open-
source licences. When something is
licensed under the GPL, most people
will know what freedoms that gives
them without needing to read the
licence file. But what freedoms does
the Intel Open Source licence grant
you? How about the Apple Public
Source License, or the IBM Public
License, the Sun Public License or the
Sun Industry Standards Source
“LICENCE FEES FOR COMPANIES USING
‘LINUX’ IN THEIR PRODUCT NAMES:
IS THIS THE FREEDOM WE WANTED?”
License? Does one company really
need two licences of its own if the
definition of Open Source is so simple?
Fortunately for all of us, HP is
taking a strong stand against licence
proliferation. Speaking at LinuxWorld
San Francisco this summer, Martin Fink
(VP Linux at HP) said: “If you’re out
there and you’re a vendor and you’re
planning to create a new license, stop.
Please don’t.” Quite right: we need less
of the licensing obfuscation, and more
of the attitude back.
Creeping corporates
As Chumbawumba will tell you, tub-
thumping can be a good thing, despite
what the OSI might say. In the case of
free software we can define tub-
thumping as the
FSF’s powerful
ability to keep
topics at the
forefront of
everyone’s
attention, year in,
year out. People
make so much
noise about free
software because
they really, really
care: they want
the source code,
www.linuxformat.co.uk
sure, but they also want the overriding
freedom to use computers however
they want to, without companies
“laying the smack down” with the long
arm of the law.
Part of this tub-thumping is the
belief that one should use free
software rather than proprietary
software, even if the proprietary
software is better. When Linus migrated
to BitKeeper, he did so because he
thought it the best source code
manager available. But with BitKeeper
being proprietary, when people started
looking at creating a free version of it
we got this choice quote from Larry
McVoy, the founder of BitMover, which
produces BitKeeper:
“If you are trying to copy BK, give
it up. We’ll simply follow in the
footsteps of every other
company faced with this sort of
thing and change the protocol
every six months. Since you
would be chasing us you can
never catch up. If you managed
to stay close then we’d put
digital signatures into the
protocol to prevent your clone
from interoperating with BK.”
Could anyone using BitKeeper ever be
free with that attitude?
Sometimes the use of proprietary
software is a little more cunning. For
example, many hardware
manufacturers now provide custom
drivers for Linux that require binary
modules in your kernel. Now, the GPL
outlaws non-free software linking to
GPL software, so why is this allowed?
The free software folks explicitly
disallow it – but the open source guys
don’t seem to bee to bothered by it.
Eben Moglen of the Free Software
Foundation said in our interview this
month: “I’m not sure whether the
kernel is licensed under pure GPL or
something else. But when I look at the
historical behaviour of the kernel
developers and the product that they
distribute I have the clear impression
that its licence is not GPL and only
GPL.” So even though the developers
claim that Linux is licensed under the
GPL, it’s not being enforced – to the
FREE VS OPEN
point where onlookers doubt whether
it really is GPLed.
Another hurdle we’re now just
starting to see is that of intellectual
property. Here at Linux Format we’ve
struggled with trademarks for some
time as we try to ship distros (most
notably Red Hat and SUSE) on our
coverdiscs, but now the wider world is
having to sit up and take notice
because of licence fees for companies
using Linux in their product names. Is
this the freedom we wanted? In the
open source world, several of the
OSDL’s biggest sponsors are huge
proponents of software patents. Are
those the bedfellows we wanted?
The future of freedom
It should be clear to everyone that
once we give up freedoms, they are
tricky to claw back. People who are
shortsighted enough to think that
open source is really just a nice way to
write code need to think about the
future. Should we really let the
contamination of proprietary code into
our free ecosphere to satisfy a short-
term need? Should we be forced to
pay money to call software ‘Linux’?
Irritating as some people seem to
find him, Richard Stallman is at least
reliable: you know he’ll always stand
up for free software, no matter what
his friends or sponsors might think.
Perhaps if Ben Franklin lived today,
he might say that people willing to
trade their freedom for temporary
functionality deserve neither and will
get none. Which is, surprisingly for
some, precisely what the FSF has
been saying for a long time… LXF
HOW WELL-VERSED
ARE YOU?
Do you know the difference between
common licences?
GPL
LGPL
BSD
IBM Public License
Apache
Mozilla Public License
Artistic License
LXF73 DECEMBER 2005 67
22/9/05 10:47:05 am
 INTERVIEW EBEN MOGLEN

THE LXF

INTERVIEW

Super Linux evangelist! Software freedom fighter!

Defender of the GNU and the Open Source Way!
No, it’s not Linux Lass – it’s Eben Moglen…

A
s professor of law and legal history at
Columbia Law School, the general
counsel for the Free Software
Foundation and the founder of the
Software Freedom Law Center, Eben
Moglen is a very busy man indeed. Add
to that his commitment to oversee the creation of v3
of the GNU General Public License, and you’ll
understand why it was that we counted ourselves lucky
to get a tiny timeslice of his schedule in which to talk
about GPL 2, GPL 3 and other licences upon which
the free software movement stands.

68 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.moglen 68 19/9/05 5:54:35 pm


ON THE LINUX COMMUNITY:
“The Ubuntu users group of Soweto, and Nokia, IBM, the
Brazilian minister of culture, the Chinese Communist
Party... What links all these people? Only the GPL.”
LXF: You’re here at LinuxWorld
San Francisco to talk about GPL
version three. What are you
going to talk about exactly?
Eben Moglen: I want to tell people
what’s about to happen, because
it’s time to tell people it’s coming,
there’s a schedule and here’s how
we’re going to start building the
process to make it work...
LXF: So there is a schedule?
EM: There will be a schedule, yes –
mind you, there’s a world out there
that could interfere.
LXF: On the one hand this, on
the other hand that...
EM: As presently advised I believe
that the Free Software Foundation
will release the first discussion draft
of GPL 3 in the first days of 2006;
I think we will have about a year of
public discussion and comment;
and we (I hope) will be in a position
to formally promulgate the new
licence at the beginning of 2007.
LXF: Are you expecting lots of
feedback?
EM: Yes. We’ve been studying some
examples of large-scale
e-deliberation internet [discussions]
on W3 patent process, the next
kernel mailing list, the Debian
process... On the basis of that, and
some other extrapolations I’ve
done, I think that there may be as
many as 150,000 people who want
to express an opinion on GPL 3.
LXF: And of those, how many do
you expect will be flames?
EM: That’s a good question, and
that’s up to the people themselves.
LXF73.moglen 69
You can’t tell, right? We have
identified as many as 8,000
organisations around the world that
may also have an opinion about
GPL 3. I think it’s going to be a very
large conversation. And one of the
things I think characterises that
conversation is that the parties who
will come to that conversation know
in some abstract way that it’s a big
diverse community we live in, but
they’ve no idea how big: they’ve
really never all met before. And
when the Ubuntu users group of
Soweto, and the Kazakhstan
Developers Association, and Nokia,
IBM, the Brazilian minister of culture,
the Chinese Communist Party... What
links all these people? Only the GPL.
LXF: So you’re expecting it to be a
big deal for the community?
EM: It will turn out to be a socially
transformative fact that the
stakeholders of the GPL are as large
and diverse a collection of people as
anywhere on earth, in any enterprise
or activity they share. We talk all the
time about community, but when you
begin to talk about something on
that scale I think the correct word is
society. It’s a free software society.
And to have a society-wide
conversation, to make a society-wide
political decision – in the true sense
of politics – that’s a unique event.
I think when it’s all over people are
going to say: “It’s a better licence,
but in a way the licence was the least
of it. It was the coming together and
seeing how big and powerful and
diverse we are that was the real
experience.” So, I think actually it’s
possible that the quantity of flaming
will go down over time, because
www.linuxformat.co.uk
INTERVIEW EBEN MOGLEN
people will suddenly begin to
understand – we live here together.
LXF: I guess you already have a
rough idea of what you’d like to
see. I think a lot of people see it
as being GPL 2 with patents.
EM: Ah, well... How about GPL 2 with
the globe? How about GPL 2 with a
big world of people speaking 85
languages contributing to GPLed
projects? How about GPL 2 with a
hovering difficulty of “treacherous
computing”, as Mr Stallman calls it.
How about GPL 2 with a load of web
servers? I mean, there are a lot of
different areas in which different
people have different concerns.
Yes, the patent problem is
obviously a very serious problem and
a lot of people will look to the licence
discussion to fix it. Although they’ve
also been reading Mr Stallman for
years telling them nothing will fix the
patent problem except changing the
patent system. And to the extent that
people are hoping the licence
contains magic, I think it’s important
to point out that magic isn’t what we
did. Good social design is what GPL 2
did; it didn’t do magic, and GPL 3
won’t do magic either.
LXF: But GPL 2 is also very short
and easy to read...
EM: I would certainly expect GPL 3
to be not much longer and not any
harder to read. That must not
happen. This is a licence whose
characteristic was that it was
addressed to developers, laypeople.
Individual software developers
around the world, here is a licence
that you can use. It’s got to stay that
way. Of course it’s longer than the
BSD licence or the MIT X11 licence.
If you just want to say, “Here, take
it, do whatever you want”, you can
write it even more briefly. The fact
that GPL has to have a little more
stiffness in it and a little more
design means it has to use a few
more words; but we’re not going to
end up with a 15-page licence,
that’s not going to happen.
LXF: Do you forsee any confusion
surrounding the transition from
2 to 3? Version two of the GPL
does say “any future version”,
but some projects – notably the
Linux kernel – have opted out.
EM: I think there will be minimal
confusion, because there will have
been a long public conversation
and there will be – I hope – no
reseller or developer or distributor
of GPL software who doesn’t know
that licence change is going on. I
think your question’s predicate is
right: the automatic re-versioning
clause will take care of the bulk of
projects. Those projects that were
labelled GPL 2 only will have to
make decisions about whether to
switch. And depending on how that
project is internally governed and
constructed that may be easy or it
may be difficult.
We will work with developers of
GPL projects, and in particular
developers of GPL 2-only projects.
We will include them in the
conversation by design, and we will
help them to make the best
possible decision about what
licence to use. Those are individual
developers and they have freedom
to choose, and GPL 2 is a licence >>
they could choose to use. But I
LXF73 DECEMBER 2005 69
19/9/05 5:55:00 pm
 INTERVIEW EBEN MOGLEN
>> think when it is all over that we will be
able to convince people that there is
nothing that they gain by staying with
the old licence, and much to gain from
moving to the new one.
LXF: So if we take the kernel, or
another project that’s stuck with a
version 2-only licence: if they
decide they’re going to re-licence,
does it become a matter of
contacting every contributor and
asking their permission?
EM: Well, that depends on the terms
of contribution agreements or other
arrangements inside the project, the
licence doesn’t tell you the answer to
that question.
I would say that the question about
the Linux kernel is a question that
should be put to the Linux developers:
“How do you choose what licence to
use and what’s your governance
process?” And I wouldn’t speak for
them. Mr Stallman, were he here,
would want me to emphasise that they
have entire freedom to do whatever
they want – it’s up to them.
There’s nothing in the FSF that
gives them any power to affect that
decision at all. I would say, as an
onlooker, that I believe that even the
Linux kernel (which, it is true, has a
very large and complex internal
process of developers) knows how to
make decisions of this kind with pretty
good grace. The kernel developers
make harder decisions every year than
70 LXF73 DECEMBER 2005
LXF73.moglen 70
what licence to apply to the kernel,
and they make them well. I must say
I’m always impressed by the degree to
which the kernel developers seem to
be able to thrash things out, and do so
with good faith and in an efficient way.
I think that the re-licensing of the
kernel (if that’s what they decide to
do) will occur in a pretty smooth and
simple form also.
LXF: It’s been a testament to good
work, surely, if version 2 lasted for
15 years.
EM: I agree with that. Not only 15
years, but six orders of magnitude
difference in market value of goods
ON GPL 3:
“There is nothing to gain by
staying with the old licence”.
being moved, four orders of
magnitude difference in the number of
people involved, three orders of
magnitude difference in how many
projects – yeah, I think it’s a pretty
sound testament to its design.
Moreover, I think it’s important for
us to recognise that it will retire ‘with
the title’. It’s not that GPL 2 will be
taken out of service because it failed –
it will never fail. It will just be replaced
fruitfully before it gets too old to
www.linuxformat.co.uk
reflect what’s happening now.
LXF: Well the other big licence
that the FSF stands for is the Free
Documentation licence, which was
determined as being ‘not free
enough’ by the Debian project.
EM: Yes. I should say that we have
refrained from pointing out too often –
or too publicly – that we tabled some
proposals to Debian, quite a long time
ago, which we thought addressed their
concerns fully.
I still believe that we will fully
address the concerns of the Debian
project with respect to the GFDL as
part of a general process of updating.
The GFDL now licenses Wikipedia,
which means that it is being used for a
form of communication rather
different from the technical manuals
for which it was designed. And as
Wikipedia grows to be the largest and
most frequently consulted reference
library on earth, which it will achieve
(it’s already more used than the Wall
Street Journal, which tells you
something), some changes to the
GFDL for the benefit of wiki-like
applications are probably a good idea.
LXF: Who’s going to do that?
EM: Jimmy Wales is a member of the
Free Software Foundation board, and
he and Richard [Stallman] and I have
been engaged in discussions recently
about what GFDL changes might be
valuable for the Wikipedia kind of
application. So I think that a revision of
GFDL is coming, which I think will
address outstanding concerns – all
outstanding concerns.
I regret that some flaming went on
that made it difficult to make progress
for a while. All I can say is that as
usual I think there’s no need for that. It
may be that some time needed to
pass in order for tempers to cool
before people could see their desire
to get to a common outcome clearly.
But I hope that the GFDL lesson
has perhaps been misinterpreted as a
suggestion that you couldn’t get
agreement, or that apparent
agreement about GPL was more
fragile than it is or something of that
sort. Without speaking out of school I
do think that the GFDL isn’t as
beautifully elegant and well designed
a licence as the GPL. I suspect that
I’m not the only person in the world
who thinks that. That means that
there might be some benefit in the
long run to a comprehensive redesign
of the GFDL in a way that I don’t think
a comprehensive redesign of the GPL
is necessary in order to make it
function well in the kernel. The GFDL’s
a more problematic licence. And yet,
at the same time, [there’s] a smaller
community of people who need to be
consulted. I think we’ve learned a lot
and will do a better job in future.
LXF: The cover text and
invariant section provisions
seemed quite unpopular…
EM: The cover text provisions
of the GFDL are there to make
it possible for Tim O’Reilly and
other such publishers to make
a commercial business out of
publishing free software manuals
under GFDL.
Those people needed to be
assured that if they published a book
in covers of their own branding that
nobody could simply put their
commercially published book into a
photocopier, bind it in the same cover
and send it out into the world. Tim and
others made clear that was what they
needed, and the licence was designed
to accommodate them in that.
19/9/05 5:55:20 pm

With respect to the invariant
sections provision of the GFDL, I note
– as maybe others have – that the only
invariant section ever used by the FSF
was to put the GNU manifesto in the
GNU Emacs manual.
But perhaps that showed not that
that was the purpose for which it was
put there, but rather that Mr
Stallman’s design expected more use
of a feature than it actually got. When
– whether it’s executable code or a
licence – you find you included a
feature that has very little actual use
in the field and which disturbs the
design in other respects, one thing you
can do is redesign to remove the
feature and accomplish the function in
some other way.
LXF: One thing I’m not sure of – is
it OK to use lines of code that
link to the kernel and tie in the
ON CHANGE:
“We need to move
by persuasion rather
than force.”
binary kernel module in closed
source projects? I wondered if
that would be illegal, or at least
against the thinking of the GPL.
EM: I’m not sure whether the kernel is
licensed under pure GPL or something
else. But when I look at the historical
behaviour of the kernel developers
and the product that they distribute I
have the clear impression that its
licence is not GPL and only GPL. I
think that for two reasons.
First, because I know that over the
years there have been a lot of
different methods for putting shims
[fixes] and binary modules in the
kernel and the kernel developers have
chosen not to enforce GPL. And so I
take it that they are tolerating activity
that they might, under the pure face
of the licence, have objected to.
Moreover I notice the GPL-only
symbols and the taint mechanism in
the kernel, which makes it appear to
me that there is a technical API for
distinguishing between permissible
and non-permissible forms of non-
free modules.
The implication that I draw is that
some non-free modules are permitted
LXF73.moglen 71
and some forms of linkage from
kernel to non-free code are permitted,
which the pure language of GPL might
not permit. Until the kernel developers
tell me, in writing that I as a lawyer
can read and understand, I don’t know.
But I look at their behaviour and I
think that they want some people to
be allowed to do some things, and it’s
their kernel, so they have a perfect
right to allow people to do whatever
they want.
LXF: But wouldn’t it be to
everyone’s benefit if that were
to be spelled out?
EM: It would be, I think, a helpful
contribution to the stability of the
industry for them to say in a parsable
and readable way what it is that they
want – if they know. With respect to
the particulars of what you are asking
about, I can’t give a legal opinion
without looking at the facts and
looking at the code and all sorts of
other things. But there are a lot of
people who want to combine unfree
code with free software operating
systems and APIs. And under
appropriate circumstances that’s
permitted and even designed for.
LXF: Shouldn’t you be keeping a
closer eye on what sort of
licensing agreements the kernel
developers have in place?
EM: The Free Software Foundation’s
position was made very clear from the
outset that the FSF represented
people who believed it was unethical
to distribute software that users
couldn’t understand, modify and share.
And I think that’s still very clearly the
FSF’s position every single day.
At the same time (and here people
who tend to regard the FSF as sellouts
or loonies somehow should look
again) even with respect to behaviour
that the FSF made very clear it didn’t
think was ethical, it certainly worked to
make it possible.
The very presence of LGPL from
the beginning was designed to allow
INTERVIEW EBEN MOGLEN
people to do that which the FSF also
thought they shouldn’t do.
But it is not the sign of a zealot
loony to say: “I think people shouldn’t
do this, but I’m going to design my
operating system to make it possible
for them to do it.” Because we need to
move by persuasion rather than force.
I would contrast that with what’s on
the other side of the story – the
people who have made sharing
difficult, every way they know how, all
the time, as though there were
something not ethical about sharing.
Who’s the zealot now?
LXF: What are your views on the
Creative Commons licence set?
EM: [Laughs] The Creative Commons
licence set includes many free
licences and a couple of unfree
licences. And I wish it didn’t include
any unfree licences.
LXF: That simple?
EM: I think that the basic question
that Creative Commons has to answer
is a powerful, complex question. We
are trying to change the way that
artists and writers and producers of
culture all around the world use the
copyright system, to encourage more
remixing and more sharing. And how
fully that change of mind can happen,
and how fast, and how best to induce
creators to change their mindset
about these questions, is a major
strategic issue.
I think Creative Commons has
chosen, understandably but not
necessarily wisely, to bring the
creative community along a
step at a time; if necessary
by accommodating the
community’s concerns
at any given moment
by offering less
freedom in licences.
Many musicians, for
example, would feel it is
perfectly OK to release
their music under share-
and-share-alike non-
commercial-use licensing but
would be wary of releasing under
a pure copyleft licence, because
they think: “If Jaguar puts my
music in a commercial
and doesn’t pay me I
will suffer.” They
might be right,
and they might
be wrong.
LXF: The rub is whether it’s
commercial exploitation.
EM: In the long run, perhaps it is safe
for musicians to permit even
commercial appropriation of their
music on pure share-and-share-alike
terms, as long as attribution is given.
Because maybe it’s true that in that
system, musicians don’t lose. But if
they think they do, why not afford
them a non-commercial use licence?
Yet in the indie media part of the free
software community there are people
who object even to the non-
commercial use elements of the
Creative Commons licences as being
too big a compromise with unfreedom.
Those people, and yet another
group of people including Mr Stallman
from what I read, go further and
object very vehemently to the
sampling licence and the developing
nations licence and other non-free
licences in the commons, which were
also devised in the Creative Commons
community, in order to make an
outcome more likely. Namely, a more
freely licensed or even non-freely
licensed but sharably licensed culture.
The question of how incremental
as opposed to how fundamental this
should be is a characteristic dispute
among reform and revolution
movements all through history in every
context, and I see this
one as a constructive Go to
disagreement over www.linuxformat.
co.uk/mag/moglen.html
strategy and long-
to read more of Eben’s
term goals. LXF
thoughts on patent reform,
politics and the OSDL
law school.
LXF73 DECEMBER 2005 71
19/9/05 5:55:31 pm
 TUTORIAL Online chat

FIRST STEPS LINUX BEGINNERS SERIES

Online chat How to use Skype,

IRC and Google Talk via Gaim
The launch of Google Talk and eBay’s purchase of Skype have pitched Voice over IP back and instant
messaging back into the headlines. High time for Andy Channelle to introduce you to online chat, then.

In the internet’s infancy, Yahoo, AOL and Microsoft all inventive hackers on various platforms created applications that

LAST launched their own versions of instant messaging (IM)

TIME
The LXF72 First Steps tutorial
showed you how to set up a
– the little smiley (or otherwise) faces used to express emotions.
printer from a Linux box and
discussed CUPS. If you missed
the issue, call 0870 837
4773 or +44 1858 438795
for overseas orders.
software. These packages – which were free to
download – could be used to conduct typed conversations over
the net. As the applications matured, new features were added
such as the ability to send voice messages, files and emoticons
However, despite rumours of a compromise, these three
companies refused to allow customers of other services to talk
to their customers. Each client used its own protocol, so AOL
Instant Messenger users could talk to other AOL users, but not
people chatting on Yahoo Messenger or MSN. Imagine if BT
refused to let its users phone Telewest customers! Fortunately,
could work with the various protocols. The Jabber project made
an early attempt at providing an open protocol for IM, but while
this has been a success in some circles, the big players were
still determined to retain control of their networks.
In August Google stepped in with a (sadly) Windows-only IM
and VoIP client built on Jabber called Google Talk. Fortunately,
as it uses Jabber, Linux users are not prevented from chatting
with their Google buddies since some of the ‘standard’ Linux IM
clients are capable of connecting to Jabber networks.
And this is what we’ll look at first, before we go through a
brief Skype tutorial and set up with IRC. By the end you’ll have
a range of options for chatting online from your Linux box.

74 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.tut_begin 74 22/9/05 12:04:55 pm


PART 1 – CONNECTING TO GOOGLE TALK WITH GAIM
Being a multi-protocol IM client, Gaim can communicate
over a number of different networks.
Gaim is an open source, multi-protocol instant messaging
client. It was one of the big winners in Google’s Summer of
Code competition, which sponsored students’ work on open
source projects, so it’s not surprising that if you’re using Gaim
you’ll be able to connect to Google Talk with little bother.
From Gaim’s Buddy List window, select Tools > Accounts or
do Ctrl+A. This will open up the Accounts window, which should
display every active account you have. Click on the Add button
to open the New Account dialog box. Ensure that Show More
Options is selected. Choose the Jabber protocol from the drop-
down list, then add the following:
Screen Name: <Gmail user name>
Server: gmail.com
The Gmail username should be entered without the trailing
‘@gmail.com’ element.
These details should translate to other IM clients.
You can put your Gmail password into the password field for
automatic logins, but this isn’t a very secure way of working. It’s
probably safer to have the application request a password on
launch: that way no one can go online and insult people in your
name! Leave the Alias field blank, then choose whether to have
the password stored in Gaim. Choosing the auto-login option will
cause Gaim to connect to Google Talk every time you launch it.
In the lower section of the dialog box, make sure that Use
TLS If Available is selected and that Force Old SSL is
deselected. The port should be set to 5222 and the Connect
www.linuxformat.co.uk
LXF73.tut_begin 75
TUTORIAL Online chat
server to talk.google.com (Google’s servers run on port 5222).
Hit the Save button to go back to the Accounts window. The QUICK TIP
new account should be added to the list, and you’ll now be able ■ Only Gmail users
to connect by hitting the Online radio button. can access the Google
Talk service. If you don’t
Adding contacts have a Gmail account,
Adding a new contact follows the same process as with most email andy.
other instant messaging accounts. In Gaim’s Buddy List window, channelle@gmail.com for
select Buddies > Add Buddy or hit Ctrl+B to open the Add an invite. This is a first
Buddy dialog. The Screen Name is the first part of your come, first served offer
contact’s Gmail address, while the Alias is the name you want and is restricted to the
displayed on your side of the chat. You might, for example, want first 50 responses.
a contact with the Gmail address ‘bob.torvalds’ to be displayed
as Uncle Bob. The Group option can be used to keep work and
leisure contacts apart (you can create new contact groups under
the Buddies menu), and the Account option should be set to
the newly created Google Talk account.
Once created, you can select a Buddy from the list and, if
they’re online, hit the IM button to open the Chat window. You’re
ready to begin talking.
>>
Add Google buddies to your Contact book.
GOOGLE TALK ON KOPETE
Kopete is a full-featured alternative to Gaim, coming (you won’t be suprised to read) from the
K-tastic KDE stable. Setting up Kopete involves using the same information in a different way. For
example, the Jabber ID and server, separate in Gaim, should be input into the Jabber ID field in
Kopete’s Basic Setup tab. Under the Connection tab, make sure that Use Protocol Encryption (SSL) is
selected and that the app is set to Override Default Server Information with the talk.google.com
server. Kopete should then connect without difficulty.
In Kopete, new buddies can be added by naming them <username>@talk.google.com.
Although Google doesn’t
list it as supported, it’s
also quite simple to set up
a Google Talk account with
KDE’s Kopete.
LXF73 DECEMBER 2005 75
22/9/05 12:05:05 pm
 TUTORIAL Online chat

PART 2 – VOIP CALLS WITH SKYPE

>> Google Talk is the latest in a line of Google apps that are
only available for Windows. However, the world of free internet
calls is not walled off from Linux users: there are two
applications available to us that will do the job.
The most successful application of its kind, and the most
widely available, is Skype, while a new app called Gizmo –
funded by Linspire’s Michael Robertson – is attempting to
muscle in on the same territory. Both of these are available for
Linux, though the latter is limited to Debian binaries.
The latest version of the Skype software is 1.2.0.11, and is
available free of charge in formats including RPMs, Debs and
static binaries from www.skype.com. Getting it on to your Skype is closer to an IM client than a telephone.
machine should just be a case of downloading the required
package and installing it either through your package manager area. Click the Next button to move on to the invite screen –
or from the command line. one of the ways in which VoIP systems trump their phone
If in doubt, download the static binary with the .tar.gz file network rivals. When you add a contact to your system, a
extension and decompress it. You should now be able to open a message is sent to the person requesting their permission for
file manager, navigate to the directory you unpacked into and you to see when they are logged in and available for calls. You,
double-click on the Skype icon to launch the application. In case in turn, give them permission to see when you’re online. A
you’re not sure how to create desktop links, right-click anywhere contact that is available will then be highlighted in the
on the desktop and select Create New > Link To Application... in application’s Contacts tab. If it’s greyed out, it’s not available.
KDE (in Gnome you would select Create Launcher), and enter There’s a handy Search For Skype Users option (also under
the path to the Skype executable (where you unpacked the the Tools menu), which provides facilities for searching for
.tar.gz file earlier). This is likely to be something like names through the entire Skype username database. Searches
/home/username/skype-1.2.0.11/skype. can be made by username, real name, location and a host of
Alternatively, you can simply open the directory and drag the other details, though you’ll need to hit the Advanced button to
icon labelled skype.desktop on to the desktop. Double-click access these options.
on it to launch the application. Once you have all your contacts added and sorted (you can
create new groups by right-clicking on a contact name and
Do you expect me to talk? selecting Groups > Create New) you can start a call by right-
We’ll assume now that you have a working installation, have
created an account and that your PC is equipped with working
microphone and speakers. It’s important when setting the
microphone and speaker volume (do this using the KMix
application or Gnome’s Volume Control) to ensure the speaker
volume is not too sensitive, as this could cause feedback when it
comes to making a call. It also makes sense to ensure that both
mic and speakers are facing you, and that the mic is close
enough to capture a clear signal from wherever you’re sitting.
There are three ways to use this application: for Skype to
Skype calls (which are free); to call a normal land line through
the SkypeOut service; or as an instant messaging client to chat
to other Skype users. The cheapest way to use the application is
to encourage your friends to download and install the software
too. That way you can add them as contacts and phone them Hmmm. Right name, wrong location!
for free whenever they are online and logged in.
clicking on a name and selecting Call This Contact. By default,
double-clicking on contact name will launch an IM session (see
“IF YOU’RE HOUSE-SITTING FOR below), but you can change this to make a call in the
Preferences section of the Tools > Options dialog box. You can
UNCLE BOB, YOU COULD USE YOUR also use the right-click Service menu to send files to contacts or

SKYPE ACCOUNT FROM HIS PC.”
It’s best to think of Skype in terms of an instant messaging
client rather than a phone service, as computer-to-computer
calls are made on the basis of usernames rather than numbers.
The great thing about this is that usernames are portable and
not tied to a single location. If you were house-sitting for Uncle
Bob, for example, you could log into your account on his PC
and not have to miss any vital calls.
Adding contacts when you know their username is extremely
easy. Do Tools > Add A Contact and type the name into the text
block calls from particular users.
You don’t have to restrict your calls to individuals. Skype
allows users to make conference calls, and these can include
both Skype contacts and regular land line users. The quality of
conference calls will vary depending on the speed of your
internet connection.
To start a conference call, select Tools > Create Conference
to launch the appropriate dialog box. You can now select
Contacts from the list on the left of the window and click the
Add button – conferences can consist of four contacts plus the
host. Once the list is complete, hit the Start button to begin. As
people answer their phones (Skype or regular) they will be

76 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.tut_begin 76 22/9/05 12:05:09 pm


added to the conference. Remember that you will be paying for
any landline participants (see below) in the conference.
It’s also possible to add a new participant to an ongoing
conference call by right-clicking on their name and selecting
Invite To Conference. The conference host is the only person
who can do this.
Calling land lines
While calling other Skype users is all well and good, not
everyone has the required hardware or a decent internet
connection. Fortunately, it’s possible to use the Skype
application to call regular phone numbers. To add a number
you follow the same procedure, inputting the number where
before you put the Skype username. When putting in a number,
PART 3 – USING IRC
Internet Relay Chat (IRC) is the dowdy underdog of the IM
world. It’s been around since August 1988 and yet hasn’t
achieved the user base of any of the three main IM clients, nor
the press attention of Jabber. But don’t let this put you off: one
of the chief benefits of IRC is that you don’t need to belong to
any sort of Yahoo, MSN or Google Talk-like club in order to use
it. All you need is some client software, an internet connection
and a little bit of information. Both of the main Linux IM clients
– Gaim and Kopete – are perfectly happy with IRC chats.
Open up the Accounts window in Gaim and, in the Protocol
drop-down, select the IRC entry. Your screen name can be
anything you choose and can change every time you log in, but
it can’t contain any spaces. Gaim automatically sets the server
to irc.freenode.net and the port to 6667. These are standard
options, but often you may be attempting to contact a group of
people on a different network – IRC networks have sprung up
to cater to different interests and locations. If this is the case,
simply change these options. Again, set whether you want the
system to automatically log you in, and hit the OK button.
On each network is a series of channels (which are often
called chat rooms). Joining one of these channels means you
will be able to chat with other people in the ‘room’, but
remember that these are public and every member of the
channel can see all conversations.
Under Gaim’s Buddies menu is an entry called Join A Chat,
which you can also access by hitting Ctrl+C. This launches a
dialog box. In the drop-down list, select the IRC account
configured earlier (it will be named with the format
username@irc-server), then add the channel name in the
second text box.
Find a channel
Many people come across IRC for the first time in relation to a
particular piece of software or interest – someone on a forum
may casually type, “If you need more I’m on #linuxformat most
nights”, assuming that this is enough information. In this case,
we could deduce from the # prefix that this person is talking
Add the channel name plus # prefix to the Channel box.
www.linuxformat.co.uk
LXF73.tut_begin 77
TUTORIAL Online chat
you are also given the option of adding a label, which makes
the contact list more legible.
One thing to remember is that phone numbers have to be
prefixed with the international dialling code. The other thing to
remember is that calls of this nature have to be paid for – you
buy call time through the website’s account screen. Calls to UK
landlines fall into the company’s most popular tariff of just over
a penny a minute – as do calls to most other European
countries, the US, Canada, Australia and China.
Finally, you can also use Skype to send instant messages to
other Skype users, in the same way as you do with Gaim or
Kopete. Right-clicking on a contact and selecting Check The
Text will launch a new messaging window where you can input
text. Hit the Enter key to make it appear on your friend’s screen.
IRC MANNERS
Some IRC channels are very newbie-unfriendly. A people, you’ll find out soon enough.
question that, to an expert, might seem trivial The operators are the people you really don’t
may receive a harsh “Read The Friendly Manual” want to annoy: they have the power to kick you
(RTFM) response. out of a channel and ban you from returning.
IRC has a well-evolved and often strongly Habitual IRC users will also complain about users
enforced etiquette, and reaction to inexperienced overusing capital letters and text formats such as
users asking ‘stooped questions’ can sometimes bold, italic and blink (which should never be
be off-putting. So, before you ask something, try used), and clothing their messages in a rainbow
to read up on a subject and use the other of colours. Not everyone, though, is so militant.
avenues available such as Google searches and To gauge the atmosphere of a room, go in and
LXF forums. Once you’re confident that you’re in monitoring the flow of conversation for a little
the right place, don’t ask if you can ask a while. Always be sure to introduce yourself and
question; just ask. If you’re asking the wrong say “Hi”, though: nobody likes a lurker.
IRC running in Gaim. Type
into the bottom part of
the window and your
conversation appears like
magic in the top!
about an IRC channel dedicated to LXF. So in the channel
section of the Join A Chat we would just add ‘#linuxformat’ and
hit the OK button to launch the standard IM window ready for
scintillating conversation.
If you don’t know a channel name, visit
SearchIRC (http://searchirc.com) to comb through
all the different channels. This tool enables you to
search individual servers or the entire IRC community
NEXT
using names or keywords. Some of the more
popular channels have been reviewed by fellow
MONTH
The best ways to organise the stuff
users, which makes sorting the wheat from the
that’s littered all over your hard
chaff a little easier. disk. We’ll take a look at
So you’ve searched the servers and, frustratingly, search tools, too.
there’s no channel dedicated to 19th Century ergodic
literature. Well, the only option is to start your own! But
that’s for another time... LXF
LXF73 DECEMBER 2005 77
22/9/05 12:05:12 pm
 TUTORIAL Gimp
ILLUSTRATION & IMAGE EDITING
Gimp Animated fractals
Michael J Hammel just wants to have fun, so he takes a trip with one of Gimp’s creative filters.
In last month’s tutorial we
learned how to make a
metallic bordered emblem
with a recessed logo. If you
missed the issue, call 0870
8374773 or +44 1858
438795.
78 LXF73 DECEMBER 2005
LXF73.tut_gimp 78
When I started work on this month’s column I
realised that I’ve been talking about using Gimp for
real-world projects for quite a few issues: logos,
posters, retouching… But sometimes you just need to have a
little fun at work, and have to be creative just for the sake of
being creative. After all, what’s the point of being a graphic
artist if you can’t be artistic?
So, I got to playing with a few of the artistic plugins in Gimp
and stumbled once again upon the Flame filter. This is a plugin
based on open source code from the computer artist and
programmer Scott Draves. It uses long strings of numbers –
similar to a genetic code, according to Scott – to generate
random shapes, or flame fractals. The shapes can be coloured
using standard and custom Gimp gradients and the results from
the Flame filter, and other tools based on the same open
source code, can be extremely impressive. Several websites
exhibit galleries of these flames (see Links box on page 81),
many of which have space or fantasy themes.
The Flames filter
Still, on the surface this filter does nothing more than generate
random fractal designs. You can set some initial parameters and
tweak the size and position of the image, but in the end you
really don’t have as much control with this filter as you do with
the IFS Fractal filter (Filters > Render > Nature > IFS Fractal).
But if you do a little digging you’ll find the Flame filter holds a
little secret: animation. Scott has expanded on the original
source from which Flame originated to produce a few more
tools that work with the same type of designs, and the Gimp
Flame filter combines with these other tools to produce a real
creative outlet. As we’ll see, the shapes generated with the
Flames filter can be used as input to another Draves program
called Flam3. This program produces a series of images based
on an initial version created in the Gimp, and this series can
www.linuxformat.co.uk
then be turned into an animation using FFmpeg or
mpeg2encode. And that’s exactly what I’m going to do in this
month’s tutorial.
A brief tour of the Flames filter is all that’s needed to get
started with this project. It’s located in the Filter > Render >
Nature menu. Opening the filter, you’ll find a preview window
with the current flame design. A two-tabbed page provides
sliders for adjusting the rendering options and camera position,
and a menu at the bottom lets us set the colour map for the
rendering. Any gradient can be used, but if the Custom gradient
setting is used, any changes in the Gradient Editor will take
effect when any other rendering option causes the preview to
be updated.
Click on the Edit button to edit the current design. A second
dialog window opens with an interface very similar to the QBist
filter (Filters > Render > Pattern > QBist), where three rows of
three previews each are displayed. The central preview is the
current design. Clicking on any other button makes that preview
the current design and updates all the other previews. Below
the previews you’ll find a slider, which adjusts the speed at
which the previews are rendered. Generally speaking, the faster
the render the less complex the design, but the definition of
complexity with this filter is really a matter of style.
Below the slider is a randomise button, which will generate a
completely new design based on the variation specified in the
menu to the button’s right. There are nine variations we can use,
and switching from one to the other will usually generate a
design completely different from the current design. This means
you can’t make minor modifications to the current design using
the Variations menu. It’s mostly trial and error to determine how
to use the previews to generate something you’re happy with.
As I said previously, you don’t have a huge amount of control
with this filter, but that isn’t really the point of using it. Find a
good design – something appropriately bizarre – and save it.
21/9/05 3:53:08 pm
 TUTORIAL Gimp

CREATE A STANDALONE IMAGE

To get a feel for using the Flame filter, I’m going to walk you through a
simple space-themed scene. The majority of this work will be in creating the
space background. The flame itself is so easy it only takes one step.

1 2

Prepare your canvas Add some stars

Type Ctrl+N in the Toolbox to open a new canvas window. Use the default
400x320 size. If this isn’t set in the New Image dialog, just click on the Reset
button. Then click on OK to open the canvas. Type ‘X’ in the Toolbox to reset
the foreground and background colours, then drag the foreground (black) into
the canvas.
Open the Hurl filter (Filters > Noise > Hurl). Set the Random Seed to 10, the
Randomization to 70% and the Repeat to 11. Click the OK button to apply this to
the background layer. Desaturate the image (Layer > Colors > Desaturate), then
open the Levels dialog (Layer > Colors > Levels) and adjust the Input levels as
shown here.

3 4

Blur the background Make some stars sparkle

Duplicate the background layer (Layer > Duplicate Layer). Click on the original
background layer in the Layers dialog to make it the active layer. Open the
Gaussian Blur filter (Filters > Blur > Gaussian Blur) and set the Blur Radius to
2.5 for both the Horizontal and Vertical directions, then click on OK to apply the
settings to the background layer. Click on the duplicate layer to make it active,
and set that layer’s blend mode to Addition. Finally, merge the two layers
together (Layer > Merge Visible Layers).
Make several oval selections (hold down the Shift key to add a new oval to
existing selections). Feather the selections by 10–35 pixels (Select > Feather).
Open the Sparkle filter (Filters > Light Effects > Sparkle). Set the sliders to
something similar to what is shown here. The Spike Density should be set low – >>
that slider determines how many stars get enhanced.

www.linuxformat.co.uk LXF73 DECEMBER 2005 79

LXF73.tut_gimp 79 21/9/05 3:53:12 pm

 TUTORIAL Gimp
>> 5 6
Design your flame
Now we’re ready for the flame. Since Flame will overwrite the contents of the
current layer, you should start by adding a new transparent layer (Layer > New
Layer). Open the filter. You can use the default design or create one of your own.
Again, the nine previews in the Edit dialog allow you to change the shape.
QUICK TIPS
■ It’s easiest to create a
flame in the Gimp filter at
a high resolution to start
with, then save that
Flame file. This is better
than scaling the results
with the Flam3 tools or
with FFmpeg or MPlayer.
■ You may need to adjust
the brightness when
rendering at higher
resolutions with the Gimp
Flame filter. You’ll
definitely need to adjust
the zoom.
■ Also for higher
resolutions, try increasing
the Sample Density. This
slider doesn’t affect the
preview, but will change
the amount detail in the
image rendered to the
Gimp canvas.
■ Centre any Flame filter
file you save on the
preview using the Camera
options. If you don’t do
this any animation
generated from that file
will be shifted off-centre.
■ The easiest way to
affect the playback speed
is to keep the default
setting under Flam3 and
FFmpeg and use MPlayer
or Xine to set the frames
per second setting.
80 LXF73 DECEMBER 2005
LXF73.tut_gimp 80
Save the result
Once you’ve chosen a suitable shape and colour map, click on OK to apply the
effect to the background layer. If you find you really like the design, use the Save
button and save to any directory. The saved designs are what I’ll be using in the
next part of this project: creating a series of images using Flam3.
A FLAM3 ANIMATION
Once you’ve saved at least one design to a file you’re
ready to generate your flame series. The saved file is first
converted to an XML file by Flam3-convert, which is then fed to
Flam3 to produce a series of images. You’ll need to download
and compile the Flam3 software for this part of the project. The
software is available for Linux from the Flam3 website (see Links
box, below right). Unpack the software with
tar xvzf Flam3-2.6.tar.gz
Change into the Flam3 directory. The software has no
prerequisites so compiling it is extremely simple, thus:
cd Flam3-2.6
./configure
make
This produces a set of programs: Flam3-render, Flam3-
convert, Flam3-animate and Flam3-genome. Flam3-render can
be used to generate a single image but we’ve already done that
with Gimp’s Flame filter so there’s no need to concern ourselves
with that program. You can optionally install the software too,
with make install. But this requires root access unless you’re
familiar with setting the installation directory using Autoconf.
You can just as easily run the programs from the same
directory in which you compiled them. A note of warning before
you dive in to these programs: they aren’t documented for the
average user. There are numerous command line options with
some being undocumented and their use and meaning is not
always clear. But keeping in mind you’re trying to be creative,
don’t be afraid to just experiment. At worst, you’ll end up with
several JPEG images you don’t like. Just try, try again.
With the programs compiled, the next step is to convert the
saved Flame filter files to Flam3 XML files. This is easily
accomplished with the following command:
cat ~/.gimp-2.2/flame/fractured-glass |./Flam3-convert >
glass.xml
The Flame filter saves its configurations in an ordinary text
file. In this example the file was saved to a directory called
www.linuxformat.co.uk
flame in the default GIMP 2.2 user’s directory. The converter
program, Flam3-convert, reads the filter file via standard input
and outputs an XML file called glass.xml. The converted XML
file will be used to generate your image series. Unlike many
Linux command line programs, Flam3 programs get their
options from environment variables. The most obvious way to
set these is on a single command line, such as
template=vidres.Flam3 repeat=10 Flam3-genome > new.
Flam3
The environment variables template and repeat are
options for the Flam3-genome program. For a full list of
options, use the -? command line option with any of the
programs mentioned.
Creating a series
Now that you have the basic background for using the Flam3
tools, the image series can be created. The first step here is to
process the original configuration into a set of control points.
This is done using the Flam3-genome program. The genome
program mutates the design in a manner, according to Draves,
similar to genetic algorithms. The command format will be
similar to this:
verbose=1 aspect_ratio=1.6 gs=3.5 ss=2.5 template=article.
xml repeat=20 ../Flam3-genome > control.Flam3
The verbose option is useful to watch that something is
actually happening when the command is run, otherwise you
won’t see anything. The aspect ratio is set for my display, which
is 1280x800. Leaving this option out won’t change the
resolution you set in the original Gimp filter (420x300). The gs
and ss options increase the quality and size of the images that
will be rendered. The amount chosen is arbitrary for this
example – I was just shooting for enlarging the images and
improving the quality of the rendering to compensate.
The only two options you can’t live without here are the
template and repeat options. The template option specifies
21/9/05 3:53:16 pm

Move over Jean-Michel Jarre: here’s a collection of stills taken from a short movie made by rendering an image from one of 20 control points,
rotating it and morphing it for a number of frames until it becomes the next control point. If you watch the movie closely you’ll see each of these
four control points show up, though only briefly as they get morphed into the next control point. You need to use your imagination if all you’re
doing is looking at still images – go to www.graphics-muse.org/source/lxfi73.avi to see it as it should be.
the XML input file for the Flam3-genome program. This is the
XML file created earlier by Flam3-convert. The repeat option
specifies how many control points to generate. Each control
point is a variation on the original design.
The control points define a set of designs in the animation,
each based on the original design from Gimp’s Flame filter.
Rendering one image for each design and putting that into a
movie would produce a very short movie that moved wildly
between images. A better solution would be to produce a set of
variations on each design that move smoothly into the next
design. This is done by filtering the control points file through
Flam3-genome again to generate a sequence of rotations and
transitions. The command format is very similar to the last one:
verbose=1 sequence=control.Flam3 nframes=120 ../Flam3-
genome > seq.Flam3
The new options are sequence, which is the input file to
use, and nframes, which is the number of frames to generate
for each control point. Each frame is a small rotation of the
design along with a slow mutation into the next design. The
output from this command will be the input into the animation
program, Flam3-animate:
verbose=1 ../Flam3-animate < seq.Flam3
When this command is run, prepare yourself: it will generate
a huge number of images. Each image will be a frame in the
movie, but at 30–60 frames per second that isn’t a very long
movie. Changing the nframes option in the second Flam3-
genome command will reduce the number of images, as will
changing the number of control points you generated. Again,
playing with the tools is the best way to learn them.
Generating a movie
Once the huge set of frames is generated you’re ready to
produce your fractals movie. This is almost a no-brainer with the
use of FFmpeg, the Swiss Army knife of video conversion. The
simplest command to get from a directory of thousands of
JPEG images to an AVI-formatted movie suitable for playback
by MPlayer or Xine is
ffmpeg -i %04d.jpg movie.avi
The 04 in this code means that each JPEG image file is
prefixed with a four-digit number. If we had used the prefix
www.linuxformat.co.uk
LXF73.tut_gimp 81
TUTORIAL Gimp
option to animate the series we could have added some other
prefix to the filenames, but for this example a four-digit number
is sufficient. Generating the movie from a thousand 420x300
JPEG images doesn’t take long – a few minutes at most on my
laptop (Intel Pentium M 1.5G and 512MB memory).
Playback is simple enough as well:
mplayer -fs -vo x11
The -fs option puts the playback in full screen mode and
the -vo option just provides a fairly standard video interface that
doesn’t require hardware acceleration (which this example
doesn’t really need).
“KEEPING IN MIND THAT YOU’RE
TRYING TO BE CREATIVE, DON’T
BE AFRAID TO EXPERIMENT.”
You can see the movie of my series at my own website (see
Links box, below), and I hope you go on to produce some
colourful, trippy movies of your own. And what good is all this
work with fractals? Well, not much, really. Some of Draves’s
work has moved into the video DJ world, where walls of images
just like these throb to the sounds of modern music marvels.
But it’s something that only a few artists take to a commercial
level – for most of us it’s just a creative outlet, a way to have
fun. And before too long, we’re back to replacing teeth and
generating logos. The work must go on, after all. LXF
LINKS
NEXT
■ Fractal galleries: www.arcanefractals.com/index.shtml
www.renwebdesign.com/fractals/index2.html
http://occipital.net/main.php
MONTH
I’ll take a look at the best image
■ Flam3 website: http://Flam3.com viewers and content managers for
Flam3 Linux source code: http://Flam3.com/Flam3-2.6.tar.gz cataloguing your Gimp
Scott Draves’s personal website: http://draves.org projects..
Sample movie generated for this article:
www.graphics-muse.org/source/lxfi73.avi
LXF73 DECEMBER 2005 81
21/9/05 3:53:18 pm
 TUTORIAL USB auto-mounting
HARDWARE SETUP
HAL Auto-mounting USB devices
Resisting any reference to “I’m sorry, Dave”, Neil Bothwick instructs HAL to get his iRiver flowing.
It seems that everything has a USB connector these
days: cameras, music players, key-rings, watches. It
won’t be long before we have USB-connected fridges
(we already have internet-enabled microwave ovens after all).
Fortunately, USB is one of the few areas where sanity has
prevailed over manufacturers’ desire for lock-in, and the vast
majority of devices adhere to well-defined standards. This
means you can connect most of these devices to a computer
running Linux and have them appear as removable disk drives.
Mounting methods
In the beginning was /etc/fstab. If you wanted to mount a
device as a user, you had to first have the root user add an
entry for it to the /etc/fstab file then type the relevant mount
and umount commands whenever you wanted to add or
82 LXF73 DECEMBER 2005 www.linuxformat.co.uk
LXF73.tut_usb 82
remove a device. This is clearly not the most user-friendly of
approaches, so a number of alternatives were introduced.
Supermount, subfs and autofs all provide methods for
automatically mounting a device. You will find more information
on them in the Other Mounting Systems box on page 83.
The current favourite method uses the Hardware Abstraction
Layer (HAL) and the D-Bus message bus system from
freedesktop.org. These two pieces of software provide a
means for the OS to notify the desktop when a new piece of
hardware is connected. It is then up to the desktop to handle
mounting this hardware, either through a separate piece of
software like Ivman, or the built-in mounting in KDE and
Gnome. It may seem like a lot of complication to use three or
four pieces of software to handle what appears, at first sight, to
be a fairly trivial task, but it works well.
19/9/05 3:50:18 pm

Use your distro’s services manager to ensure that D-Bus
and HAL are started automatically. Note that some distros
list D-Bus as ‘messagebus’.
There are a couple of prerequisites to using HAL and D-Bus.
You need a recent kernel, at least 2.6.8 but 2.6.10 or later is
ideal. You also need to be using Udev to handle your /dev
directory. Most recent distros already use these, but check your
kernel version by typing
uname -r
in a terminal. To see if you are using udev, type
ps ax | grep udevd
and you should see something along the lines of 895 ? S<s
0:00 udevd. If you get no output from this command, you are
not running Udev and need to check your distro’s website,
documentation, wiki or forums to find out how to switch over.
Once you have established that the basic requirements are
present, see whether you have D-Bus and HAL installed by
checking in your distro’s package manager. Once again, the
latest versions of most popular distros have these installed and
enabled by default. If they are not available for your distro, we
have the source code on the coverdisc. The software packages
on the disc are the latest versions in the respective stable
distributions. There are later versions available but these are the
versions on which this tutorial was based.
If your distribution does not come with D-Bus and HAL, and
they are not available via your package manager, you can install
them from the tarballs on the CD. To install D-Bus, open a
terminal and type
su
<give root password>
tar xzf /path/to/files/on/cd/dbus-0.23.4.tar.gz
cd dbus-0.23.4
./configure && make && make install
Use KDE’s desktop configuration dialog to pick which icons
are displayed on your desktop.
www.linuxformat.co.uk
LXF73.tut_usb 83
TUTORIAL USB auto-mounting
To install HAL, repeat this process, replacing dbus-0.23.4
with hal-0.4.8.
These are both services, so make sure you also have them
set to start when you boot. Gentoo users can install HAL and
D-Bus, add the appropriate run scripts and make sure all
programs support HAL by adding hal to the USE flags in /etc/
make.conf and running
emerge --update --ask --verbose --deep --newuse world (or
emerge -uavDN world)
rc-update add hal default
Power to the user
This whole process is supposed to run in user space, without
needing any root privileges. However, mounting devices that are
not listed in /etc/fstab can only be done by root. HAL gets
around this by running a program called fstab-sync when it
detects a device. fstab-sync will create a mount point in /media
and add a line to /etc/fstab, so that the mount call may be
“ETC/FSTAB IS A CRITICAL SYSTEM
FILE, SO IT’S A BAD IDEA TO MESS
WITH IT UNNECESSARILY.”
issued as the desktop user, and permissions to the device set
accordingly. It is also able to clean up after itself, removing the
mount points and /etc/fstab entries, usually when rebooting.
The main problem with this is that /etc/fstab is a critical
system file, so it is a bad idea to mess with it unnecessarily.
Imagine what would happen if your system crashed as fstab-
sync was writing to /etc/fstab and the file was unreadable when
it rebooted. There are also security concerns about allowing
/etc/fstab to be modified on the fly without user intervention.
There is an alternative to this in the form of pmount. This
program allows users to mount devices, subject to certain
restrictions. It will only mount devices in /media, and it will only
work with removable devices, although there are ways around
this in case your USB hard disk is not recognised as a
removable device. pmount will create the mount points as it
needs them. It is also on the coverdisc. This is not the latest >>
version, but it is the one most compatible with the version of
OTHER MOUNTING SYSTEMS
Supermount was originally intended for CD-ROM use. It is a kernel patch that adds a pseudo-
filesystem that detects when a data CD or DVD is inserted into the drive, and mounts it. When the disc
is ejected, supermount detects this, closes any open file handles and unmounts it. Although intended
for use with read-only optical discs, it can be used with USB mass storage devices too, although there
are some limitations. It still needs an entry in /etc/fstab, so the device must mount with a consistent
device name – not a problem if you use Udev with suitable rules for persistent device naming.
It also requires the device to be mounted with the sync option. This ensures that any data is written
to the drive immediately and not cached. As soon as the drive access light goes out, it should be safe
to remove it and let supermount unmount it. Without sync, doing this would corrupt the filesystem.
A suitable entry for /etc/fstab is
none /mnt/flash supermount fs=auto,dev=/dev/usb/flash,--,users,sync,noatime,umask=000 0 0
The umask option ensures that the device is readable and writable by all users. This is necessary
because the device is mounted as root. sync is vital, ensuring that data is written to the disk
immediately. As there is no way for the USB port to send a signal that you are about to unplug the
device, supermount can only unmount it after it is unplugged, so it is essential that data is written
without cacheing. This is a far from ideal alternative. Although it works reasonably well, HAL is a far
better option. Supermount does have one potential advantage, in that it works with 2.4 kernels.
Autofs is a kernel option, with associated user space programs, designed for automatically mounting
network devices on demand. It is not really suitable for use with removable devices. subfs or submount
are designed for removable devices, and once again has the advantage of working with 2.4 series
kernels, making it suitable for older systems. As with supermount, submount requires an entry in
/etc/fstab, so you need to know that your device will always be given the same name.
LXF73 DECEMBER 2005 83
19/9/05 3:50:48 pm
 TUTORIAL USB auto-mounting
Using the Konqueror side bar to browse any
connected devices and their contents.
>> HAL on the disc. Install pmount in the same way as D-Bus and
HAL: untar the archive and run ./configure && make &&
make install.
“GNOME CAN BE CONFIGURED
TO AUTOMATICALLY MOUNT THE
DEVICE WHEN YOU PLUG IT IN.”
Once it’s installed, you need to tell HAL to use pmount
instead of fstab-sync. Go back to your root terminal and type
cd /etc/hal/device.d/
rm 50-fstab-sync.hal
ln -s $(which pmount-hal) 60-pmount.hal
Using HAL with KDE
Once you have HAL and D-Bus installed and running, KDE
should require nothing more to make it all work, provided KDE
was built with HAL support. If you installed KDE from a binary
package, this should not be a problem, as these packages are
generally built with all features turned on. If you compiled from
source yourself, you need to have used the --with-hal option
when running ./configure, otherwise you may need to build it
again. If you are running Gentoo, provided you set the hal USE
flag and re-emerged as above, it should just work.
Now plug in a USB mass storage device, such as a USB
flash disc or a USB media player. Don’t use a camera for now
as not all cameras work as removable drives
(see Digital Cameras box, below). After a
DIGITAL CAMERAS
One factor that makes mounting so many want to download your
different USB devices with the same system photos, but does not let
possible is that they all work to the same USB you view them until
mass storage standard, handled by the afterwards – considering
usb-storage kernel module. Unfortunately, as the slow speed of the USB
with all standards, there are exceptions. Many connection to my Canon
digital cameras do use usb-storage and can be SLR, this is probably not a bad thing, as browsing
mounted as disc drives in exactly the same way over the USB link is painfully slow. Ivman has no
as a flash disc, MP3 player or mobile hard drive. default way of handling this, but can be
However, there are others that do not, such as configured with a rule in
Canon cameras. All is not lost though, as KDE, ~/.ivman/IvmConfigActions.xml like this:
Gnome and Ivman all have ways of dealing with <ivm:Match name=”hal.camera.libgphoto2.
this. KDE still pops up an icon and allows you to support” value=”true”>
browse your photos using Konqueror’s camera:/ <ivm:Option name=”exec” value=”digikam”>
kioslave. Gnome pops up a dialog asking if you </ivm:Match>
84 LXF73 DECEMBER 2005 www.linuxformat.co.uk
LXF73.tut_usb 84
few seconds you should see an icon appear on the desktop .
Hold the mouse pointer over this icon to get some
information on the device, which is currently unmounted. Click
on it, and KDE will mount it and open a Konqueror window or
tab (depending on your KDE settings) to show its contents. If
you do not see an icon, right-click on the desktop and select
Configure Desktop. Go into the Behaviour section, select the
Device Icons tab and ensure that Show Device Icons,
Mounted Removable Medium and Unmounted Removable
Medium are all selected.
You can also view devices from the Konqueror sidebar.
Select the Service tab (at the bottom), pick Devices from the list
in the sidebar and you should see a list of connected devices.
The Devices tab just below Bookmarks performed this function
in earlier versions of KDE. For 3.4, you can find this in the
Services tab, under Media. If you have upgraded from a
previous version, you can edit this tab to change the address
from devices:/ to media:/. If there is no Devices tab and you
would like a separate tab for your removable media, right-click
on the navigation bar, select Add New > Folder and set the
address to media:/.
You still need to unmount a device before removing it, or
you risk corrupting the data it contains. To do this in KDE, right-
click on the device’s icon and select Safely Remove from the
pop-up menu. Once the icon reverts to its unmounted image or
disappears, it is safe to remove the device.
Auto-mounting with Gnome
This is just as easy as using HAL with KDE (if not easier). You
need to have installed the Gnome-volume-manager package.
This has probably been handled by your distro, but the source
code is on the coverdisc if not. This is not the latest version, but
it matches the D-Bus and HAL code.
Gnome is more advanced than KDE in this respect, as it
allows you some control over how it handles hotplugged devices.
While KDE simply displays an icon and only mounts it when you
open it, Gnome can be configured to automatically mount the
device when you plug it in. Gnome Volume Manager takes
things a step further, by allowing you to configure actions to
take place automatically when a device is detected.
You do this by running Gnome-volume-properties. While its
main use is to launch an audio or video player when a CD or
DVD is inserted, it can also be used with removable devices. For
example, it can launch your favourite photo album software
when you connect your camera. KDE will apparently have such
a control in the next major release. For now, Gnome users get
much more configuration choice than KDE users in this
department. Whether Gnome users will welcome having
more options than KDE or be horrified at the
thought is open to debate.
Ivman for more control
While KDE and Gnome have built-in support for
mounting hot-pluggable devices via HAL, many
people prefer to use a lighter window manager
and only add the features they need. Ivman
provides similar features to Gnome’s Volume Manager,
but in a separate package, independent of any window
manager. It has more features than Gnome’s Volume Manager,
so KDE and Gnome users may also consider using it.
The latest Ivman source code is on the coverdisc, and
installed in the same way as HAL and D-Bus. In order for Ivman
to work, you must ensure that all three programs are started at
boot, in this order: dbus, hald, ivman (the ‘d’ after hal stands for
daemon). You should also have pmount installed, otherwise
Ivman will use fstab-sync with the risks that poses.
19/9/05 3:50:49 pm
 TUTORIAL USB auto-mounting

Ivman is capable of a lot more than simply auto-mounting

devices, but to do this, you need to have two instances running.
The first instance runs as root at startup, usually from /etc/init.
d, and handles auto-mounting. This is already set up in the
default configuration files in /etc/ivman. The second instance is
run as your normal user, via whatever method your window
manager uses to run programs when it starts up. If in doubt, put
it in ~/.xinitrc. This instance of Ivman reads its configuration
from ~/.ivman. If this directory does not exist the first time you
run Ivman, it will be created and filled with default configuration
files. These defaults will add no extra functionality until you edit
them. By doing so, you can add features such as automatically
running a program when a particular device is connected. For
example, you can have it automatically play a DVD with MPlayer
or Xine by adding
<ivm:Match name=”hal.volume.disc.type” value=”dvd_rom”>
<ivm:Option name=”execdvd” value=”umount $hal.block.
device$ &amp;&amp; mplayer dvd://1 -really-quiet -fs” />
</ivm:Match>
or
<ivm:Match name=”hal.volume.disc.type” value=”dvd_rom”>
Even when not recognised as a removable drive,
<ivm:Option name=”execdvd” value=”umount $hal.block.
HAL and Konqueror can still show the contents of my
device$ &amp;&amp; xine -f -g dvd:$hal.block.device$” /> digital camera.
</ivm:Match>
to ~/.ivman/IvmConfigActions.xml.
This, along with several other ideas, is already in the default
drive as non-removable. The solution is to add the device name
config file created in ~/.ivman, but commented out. Nothing (not the mount point) to /etc/pmount.allow.
runs by default from here. The Custom Rules page of the Ivman When HAL detects a device, it uses the device name when
wiki at http://ivman.sourceforge.net/wiki/Custom_Rules has mounting it, so a USB disc detected as /dev/sda1 would be
several more examples, such as a rule to automatically sync mounted at /media/sda1. If you want more meaningful names,
directories from your home directory to a specific USB flash disc add rules to /etc/udev/rules.d/10-udev.rules to give suitable S
names. HAL uses the actual name, ignoring any symlinks, so a
as soon as that disc is inserted. The possibilities are limited only ■ Should a new device not
by your imagination. rule like show up when you first
BUS=”usb”, KERNEL=”sd?1”, SYSFS{product}=”iRiver H300 connect it, you can check the
A few final points Series”, NAME=”%k”, SYMLINK=”iriver” system log to see how it is
For security reasons, pmount is only intended to mount would still mount it at /dev/sda1, or whatever drive name the identified. With the device
removable devices. If it thinks a device is a hard disk, it will kernel allocated, whereas switching things around to unplugged, type tail -f /var/
log/messages in a root
refuse to mount it. This can be a problem if you have a USB BUS=”usb”, KERNEL=”sd?1”, SYSFS{product}=”iRiver H300
terminal. Then watch the
hard disk that pmount identifies as non-removable. You can test Series”, NAME=”iriver”, SYMLINK=”%k”
kernel messages when you
is this is the cause of a drive not mounting by running pmount means the device is mounted at /media/iriver. This is the Udev
plug it in. It will take a few
from a terminal. If the following code: rule I use for my iRiver MP3/Ogg player. If you want to know seconds for the device to be
pmount /dev/sda1 #(or whatever node the device uses) more about writing udev rules, see the tutorial in LXF66 or visit recognised, then you can see
returns a ‘device is not removable’ error, HAL has identified this www.reactivated.net/udevrules.php. how the kernel sees it and the
If the device’s filesystem (usually MSDOS device node (if any) it is given.
FAT) has been written with a volume name, Depending on your system
this will be displayed as the icon text. If there logger, the path to the log file
is no volume name, the device name will be may be /var/log/everything or
/var/log/current.
used instead. This is another reason to make
sure that each device is given a meaningful ■ IEEE1394 (otherwise
name, especially if you have several USB known as FireWire), devices
also work well with HAL and
devices. You can add or change the volume
the other software covered
name of an existing partition without
here. If you build your own
reformatting the filesystem with Mlabel, part
kernel, ensure that
of the Mtools package (which was created to IEEE1934_SBP2 support is
access MS-DOS disks from Unix without the included. If you are using the
need for mounting them). kernel supplied for your
If you want to rely on HAL detecting and distro, this should already be
mounting a device for you, you should not done. This is the module that
have it listed in /etc/fstab. If it is listed there, handles IEEE1394 hard disks.
pmount will use the mount details given there, You may need to manually
load the sbp2 module, either
but mounting as the current user, instead of
by tying modprobe -v sbp2
deriving the mount point from the device
as root or by adding it to
name. Although this normally means you
/etc/modules.conf (or
Gnome Volume Manager’s preferences, where you choose whether a new should not have the device listed in /etc/modules.autoload for
device is mounted automatically or waits for you to select it. It handles /etc/fstab, it does give you a way to override Gentoo users).
removable media (CDs and DVDs) as well as removable drives. the defaults. LXF

www.linuxformat.co.uk LXF73 DECEMBER 2005 85

LXF73.tut_usb 85 19/9/05 3:50:54 pm

 TUTORIAL Gambas

BASIC PROGRAMMING

Gambas Application building

PART 3 Edison was wrong. Forget about perspiration, genius is 100% delegation in Dr Mark Alexander
Bain’s book. So sit back, put your feet up and let a Gambas GUI do all the work for you.
I’ve probably already mentioned this, but it bears

LAST repeating: all good programmers should aim to be as

TIME
We had a look at using
Gambas to program in a
professional manner – using
variables and reusing code in
modules and classes. Oh, stop
yawning; we’re doing some
fun stuff again this month.
If you missed the issue, call
0870 8374773 or +44 1858
438795 for overseas orders.
idle as possible. That’s not to say that you should be
doing nothing – completely the opposite. But it’s your brains
you should be using, not your fingers. We started addressing
this last month when we looked at modules and classes. We’ll
continue our quest to become the perfect idle programmer by
having a look at what Gambas can do to make us look good
with the minimum of effort.
The first thing to remember is that there is usually
someone who has already done what you’re trying to do;
especially when you’re dealing with Linux. Let’s say, for instance,
that you want to be able to look up the owner of a website. It’s
a simple task – just go to the command line and type whois
linuxformat.co.uk. You will be given a long list of data
including the domain name, registrant, registrant’s physical
address, date of registration and the IP addresses of Linux
Format’s registered servers.

How can you do this with Gambas?

■ Gambas 1.0.11 If you’ve answered that you’d look for domain databases or start
(stable)
searching for NIC handles then stop: you’re not listening to what
■ Gambas 1.9.20
(development) I’m telling you. If, on the other hand, you’ve answered “I’d cheat”,
■ Project code buy yourself a beer – you’ve got the right mentality to be a
from the tutorial good programmer. The real question is: “How do you cheat?”
The answer is to use Gambas’s SHELL statement to get
Linux to do the hard work for you, then use the result in your
program. Start off by creating a form that contains a text box, a
text area and a couple of buttons, and don’t forget to name the
objects sensibly. In this case I’ve called the text box txtUrl, the
text area txtResult, and one of the buttons btnWhois. The code
for the button click is
PUBLIC SUB btnWhois_Click()
DIM filename AS String
filename = “/tmp/whois.tmp”
IF (txtUrl.Text) THEN
Every Linux command (in this case the command is whois)
can be turned into a GUI.
SHELL “whois “ & txtUrl.Text & “>” & filename WAIT
txtResult.Text = File.Load(filename)
ELSE
txtResult.Text = “”
END IF
END
The code is quite simple: it defines a filename (/tmp/whois.
tmp) and runs a shell command, taking the input from txtUrl
and running whois on it, then outputting the result into a text
file. So if the input were linuxformat.co.uk, the actual
command being run would be
whois linuxformat.co.uk > /tmp/whois.tmp
The contents of the file /tmp/whois.tmp are then put into the

86 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.tut_gambas 86 21/9/05 1:49:21 pm

 TUTORIAL Gambas

text area by using the File.Load command. The IF statement

simply ensures that an error cannot occur if the input box TCDUMP
remains blank. This way of writing code ensures that incorrect PRIVILEGES
data cannot be entered – and so prevents crashes.
Normally tcpdump can only
The other way is to handle errors as they occur: be run as root. If you want to
PUBLIC SUB btnWhois_Click() run this application as any
DIM filename AS String other user you’ll need to
filename = “/tmp/whois.tmp” update the sudo file
TRY SHELL “whois “ & txtUrl.Text & “>” & filename WAIT (/etc/sudoers) with the user
IF ERROR THEN details, adding a line like:
txtResult.Text = “” bainm ALL=(ALL)
NOPASSWD: ALL
END IF
txtResult.Text = File.Load(filename)
END
This time the whois command is sent regardless of the input,
and the results are handled according to whether an error
occurs or not. Which way you do it comes down to personal The SHELL command
choice: there are pros and cons to both. The first method enables you to produce
carries the drawback that you need to be sure that you know all a network monitor using
possible error situations, but it does ensure that the wrong tcpdump.
things cannot be entered. The second method guarantees that
all errors will be caught, but needs you to be able to output sudo /usr/sbin/tcpdump > /tmp/bainm_tcpdump.
sensible error messages. Either way, the important thing is that tmp
the program does not crash. fi
The other button is (it’s obvious if you’re thinking along the You might ask why I’m not just running
right lines) a close button – every form needs one. So if you sudo /usr/sbin/tcpdump > /tmp/bainm_tcpdump.tmp
name the button btnClose the code will be: The answer is that this would be run every time the button
PUBLIC SUB btnClose_Click() btnUpdate is pressed. That would mean that you’d end up
ME.Close running multiple versions of tcpdump. The script checks to see if
END tcpdump is already running before it tries submitting it again.
One thing to take notice of is the WAIT option for SHELL. If You may also wonder what the grep -v grep bit is all about.
you use WAIT the program will wait for the command to finish This simply removes the ID of any extra processes that grep
before it continues. If WAIT is not used, the program will starts itself.
continue with the command running in the background. The
background mode can be useful when you want to set a shell Bored with pressing buttons yet?
program running and to leave it running. As always, it’s good to The application works well as it stands – starting tcpdump and
look at a real life (whatever that is) example. then viewing its output – but as a user you have to click a button
Imagine that you need to build a GUI application to monitor continually in order to see the result. We can automate this
network traffic. In particular, you want to read all of the headers
of any packets of data flying about on the network. Hopefully
you have realised that I’m not asking, “How would you do that?”.
Instead I’m asking, “Is there an existing Linux application that
“IF YOU THOUGHT OF CHEATING,
already does what I want?”. In this case, there is: tcpdump. BUY YOURSELF A BEER – YOU’VE
Normally you will set tcpdump running and monitor its
output in a console window. We’ll do that, but use Gambas GOT THE RIGHT MENTALITY.”
instead. As before, you’ll need a form, but this time only add a
text area and three buttons (name them txtResult, btnUpdate process by adding a timer. Select the timer (the icon that looks
and btnClose). like a clock) from the toolbox, and draw it on your form. It
Now we can add the code: doesn’t matter where, because it becomes invisible when you
PUBLIC SUB tcpdump_update() run the form. Remember to name it something useful (eg
DIM command AS String tcpdumpTimer). If you double-click on the new object Gambas
command = will take you to the code window and you should see:
“if [ ! \”$(ps -ef | grep /usr/sbin/tcpdump | grep -v grep)\” ]; PUBLIC SUB tcpdumpTimer_Timer()
then” & END >>
“ sudo /usr/sbin/tcpdump > /tmp/bainm_tcpdump.tmp;” & Edit your code so that it reads:
“ fi”
SHELL command SETTING OBJECT PROPERTIES
txtResult.Text = File.Load(“/tmp/bainm_tcpdump.tmp”)
END Objects can be set in two ways. The first is to use the Gambas
PUBLIC SUB btnUpdate_Click() property screen. The second is to set them from within the
tcpdump_update program. For example, the following code switches on the timer
and sets the time between firings:
END
PUBLIC SUB Form_Open()
In this example the commands to be run on Linux are built up
tcpdumpTimer.Enabled = TRUE
into the command variable, so the script to be run is actually tcpdumpTimer.Delay = 5000
if [ ! “$(ps -ef | grep /usr/sbin/tcpdump | grep -v grep)” ] END
then

www.linuxformat.co.uk LXF73 DECEMBER 2005 87

LXF73.tut_gambas 87 21/9/05 1:49:24 pm

 TUTORIAL Gambas
>> THE STARTUP
CLASS
If you have multiple forms in
a project you can change
which one opens when you
run the application by going
to the project window, right-
clicking on the form that you
want, then selecting Startup
Class. Alternatively you can
create a module, set it as the
startup class and then add
the code:
PUBLIC SUB Main()
frmWebBrowser.Show
END
Obviously you would
change the name of the form
to the one that you want to
start with.
“WE’VE GOT A NICE CLEAN FORM
THAT’S ACTUALLY DOING SOME
QUITE COMPLICATED TASKS.”
The Gambas Menu Editor
helps you create a tidy and
professional-looking form.
88 LXF73 DECEMBER 2005
LXF73.tut_gambas 88
PUBLIC SUB tcpdumpTimer_Timer()
tcpdump_update
END
If you run the program now the timer will do nothing – by
default it is turned off. To turn it on, go to its property screen and
change Enable from False to True. You may also notice that
there is a Delay parameter. This is the time in milliseconds
between each firing of the timer subroutine. Setting this to
5,000 (five seconds) will give you a lively display without
overloading your processor. You can even remove the btnUpdate
button if you want.
Cleaning up when you leave
If you close your application down and then type the following
into a console:
ps -ef|grep tcpdump
you will probably get result something like this:
root 1443 1437 0 15:37 ? 00:00:00 /usr/sbin/tcpdump
This is because our application has set tcpdump running and left
it in the background. Just to be nice we should really turn if off
again when we’re finished.
This can be done in the Form_Close subroutine:
PUBLIC SUB Form_Close()
DIM command AS String
DIM filename AS String
DIM tmp_process_id AS String
filename = “/tmp/bainm_tcpdump_process.tmp”
command =
“ps -ef | grep /usr/sbin/tcpdump|grep -v grep” &
“| awk ‘{print $2}’ > “ & filename
SHELL command WAIT
tmp_process_id = file.Load(filename)
SHELL “sudo kill -9 “ & tmp_process_id WAIT
END
Two shell commands are run in the subroutine – the first
identifies the process ID for tcpdump. The second kills it. The
beauty of this technique is that you can use any Linux command.
www.linuxformat.co.uk
For instance, you could design a GUI for top, using the timer to
run top -n1 > /tmp/top.tmp for single iterations, and then read
in the result using file.load.
Having looked at some of the things that Linux can do for us,
we can go on to look at the built-in Gambas functionality.
The Gambas menu
So far we’ve used buttons to carry out any work in the forms.
These work well, but do tend to clutter up the screen, and also
present the problem of maintaining the correct layout when the
form changes size. The answer is to use Gambas menus.
Open up a form in design mode, then right-click on it. Click
on Menu Editor (or you can press Ctrl+E). You can now build the
structure of your menu by adding items. As always, remember to
name them sensibly rather than the default Menu1, Menu2 etc.
You create sub-menus by indenting a menu item. When you
close the editor the new menu will be displayed on the form.
With the menu in place, the next stage is to add some
functionality. The first thing to add is the form close. Assuming
you’ve created a menu item labelled Close and named it
mnuClose, its code will be:
PUBLIC SUB mnuClose_Click()
ME.Close
END
So we’ve now got a nice, clean, professional-looking form that is
actually doing some quite complicated tasks (or at least it’s
managing the processes that are doing the hard work). We’ve
also seen how all of those Unix and Linux developers who’ve
gone before can help us look good. Now we can look further at
the ways in which the lovely Gambas developers can help us to
be idle programmers.
Gambas components
Back in LXF71 we saw how useful the Gambas components can
be, and how easy they make it to access a database. Just to
remind you, to use a component you must first enable it by
opening the Project Properties screen (from the project window
click on Project followed by Properties) and going to the
Components tab. This lists all the components available to you.
To see how useful these can be we’ll look at creating our own
web browser. Check the gb.qt.kde.html component. When you
go back to the toolbox you’ll find a new tab entitled KDE. If you
click on this tab you see the new objects that are available for
any forms.
Create a new form (name it frmWebBrowser) and add a text
box (txtUrl), a button (btnGo) and a web browser object
(webBrowser). You should also add either a Close button to the
form or a Close element to the menu. Once that’s done you can
add the code:
PUBLIC SUB btnGo_Click()
IF (txtUrl.text) THEN
webBrowser.Path = “http://” & txtUrl.Text
END IF
END
If you now run the application then you’ll find that you have a
working (albeit simple) web browser.
We can enhance our basic web browser by making use of
more of the Gambas components. Tick gb.net (the networking
component) and add a text label (txtIp) to the form. When
you’ve done that, modify the code so that it looks something
like this:
PRIVATE dns AS DnsClient
PUBLIC SUB btnGo_Click()
IF (txtUrl.text) THEN
webBrowser.Path = “http://” & txtUrl.Text
END IF
21/9/05 1:49:26 pm
 TUTORIAL Gambas

PRIVATE SUB loadFile()

webBrowser.Path = “file:///” & htmlfile S
END ■ Press F4 to display the
PUBLIC SUB Form_Open() properties screen.
http = NEW HttpClient Press F5 to run the
htmlfile = “/tmp/test.html” application.
END Press F6 to bring up the
Notice that this time the web page is downloaded before the Gambas toolbox.
browser reads it. ■ If you’re not sure what
You’ll need to put a text area (txtHtml) and a button methods are available when
(btnSave) on the Editor tab. Then you can add the code, thus: coding in Gambas,
PUBLIC SUB tbsEditor_Click() right-click on the object and
select Event. This will display
SELECT CASE tbsEditor.Text
a list of all of the
CASE “Viewer”
programmable events.
loadFile
■ Hide text in a textbox by
CASE “Editor”
setting ‘Password’ to True.
txtHtml.Text = File.Load(htmlfile)
END SELECT
END
The Gambas components are accessible through the PUBLIC SUB btnSave_Click()
Project Properties screen. file.Save(htmlfile,txtHtml.text)
END
txtIp.Text=”” We can now modify the Upload tab, adding three text boxes
dns.HostName = txtUrl.text (txtTarget, txtUsername and txtPassword) and a button
dns.GetHostIP (btnUpload). In the code we need to add a global parameter:
txtIp.Text = dns.HostIP PRIVATE ftp AS FtpClient
END and you’ll need to add the following to the form_open
PUBLIC SUB Form_Open() subroutine:
dns = NEW DnsClient ftp = NEW FtpClient
END Last of all, we’ll add the subroutine to upload the file.
When you run the application this time, the IP address of the PUBLIC SUB btnUpload_Click()
website will be displayed as well as the website itself. The great ftp.URL = txtTarget.Text
thing is that you don’t have to understand how the component ftp.User = txtUsername.Text
looks up the information – you just need to know how to use it. ftp.Password = txtPassword.Text
ftp.Put(htmlfile)
A more advanced web browser END
We’ve managed to make a simple web browser, but they’re ten a This is, of course, an extremely simple editor – there’s no syntax
penny anyway. I made it a bit more interesting by adding the IP highlighting or checking. You’ll also find that the application only
address of the web page. Now we can produce a simple HTML uploads a single filename: that’s something for you to think
editor. We’ll do this in a slightly different way from the simple web
browser: instead of just viewing the page, we’ll download it, view
about. This should, however, give you a good idea of what can be
achieved very quickly using the Gambas components. LXF NEXT
and edit the downloaded text, then upload it again when we’re
finished. We will, of course, use the Gambas components to do
MONTH
We’ll be looking at a roundup
all the hard work for us.
of some of the little tips and
Create a new form (I called mine frmHtmlEditor), and either
tricks that separate the great
make it the startup class or modify your module to start it. Don’t from the good.
forget to add a button or menu item to close it. Since there are
going to be three distinct activities we can use a tab strip (call it
tbsEditor) from the toolbar to divide these cleanly. You’ll find that
there is only a single tab (Tab 0) by default; just go to the
properties page to change this (change the count parameter).
Now you can click on each tab to change its text to something
more useful (such as Viewer, Editor and Upload).
The Viewer tab will look very similar to the web browser that
we’ve just looked at, but the code is quite different. Create the
objects txtUrl, btnGo and webBrowser (don’t forget to add the
gb.qt.kde.html component). You will also need to add the
gb.net.curl component to enable the use of HTTP and HTP
functionality. The code for the web browsing part of the form is
PRIVATE http AS HttpClient
PRIVATE htmlfile AS String
PUBLIC SUB btnGo_Click()
http.URL = “http://” & txtUrl.Text
http.Get(htmlfile)
loadFile Buttons, timers, tabs… Anyone can make a GUI application
END in minutes using Gambas.

www.linuxformat.co.uk LXF73 DECEMBER 2005 89

LXF73.tut_gambas 89 21/9/05 1:49:27 pm

 TUTORIAL PHP
SCRIPT PROGRAMMING
PHP Data objects in 5.1
The best things come to those who wait. Yes, the PHP update is here. Let Paul Hudson show you around.
LAST
TIME
In LXF72 we wrote a
pathfinding script, with a little
help from New York taxis. If
you missed the issue, call
0870 8374722 or +44
438794 for overseas orders.
S performance, and realpath() now
■ You can use a file as a DSN
by specifying file://path/to/
your/file. Inside that file you
need to put the DSN string.
■ Instead of PDO_FETCH_
LAZY you can use PDO_
FETCH_ASSOC to mimic
mysql_fetch_assoc().
■ If you prefer to execute
the query and loop over it
separately, use the fetch()
method on the object you get
back from $db->query().
■ Pressing Forward four
times in Unreal Tournament
uses your adrenaline to give
you extra speed. Mwahaha!
90 LXF73 DECEMBER 2005
LXF73.tut_php 90
A long time ago, in a magazine that’s probably
collecting dust on your shelf by now, I promised we’d
cover PHP 5.1. The big delay was the Sudoku
competition: we spent a few issues solving problems mysteriousy
similar to the Sudoku challenge that would give you an unfair
advantage over the crazy people who don’t read this tutorial. As I
write this we’re approaching the last days of the competition,
which means you’re probably all finalising your entry about now.
Anyway, on to the important stuff: this issue I want to look at
PHP 5.1, the new release from the 5.x family. This is the first
minor release of PHP 5, despite the original 5.0 release coming
out about 15 months ago. Still, it was worth waiting for: we’ve got
new functions, new performance gains and new extensions, of
which arguably the most important is PDO: yet another new
three-letter acronym that stands for PHP data objects.
Faster than a speeding PHP 5.0
A lot of work has gone into making PHP 5.1 more efficient at
even the most basic level. So, while basic compilation and
execution has been optimised, more advanced things such as
the switch() statement, variable fetching and magic method
invocation have also been tuned. scandir() is now ten times
faster, various array functions have been revisited with an eye to
Let’s see how
fast this new PHP
really is…
www.linuxformat.co.uk
has a cache to avoid multiple lookups of the same virtual path.
So, PHP 5.1 is fast, fast, fast, and you don’t actually need to
rewrite your scripts to take advantage of it – everything gets
faster as soon as you upgrade.
Alongside pure performance you can now also get your mitts
on some new functions: array_product() calculates the sum of
all array values; array_diff() and array_intersect() now have
key-based equivalents; htmlspecialchars() now has its inverse
in the form of htmlspecialchars_decode(); and there’s even a
cool new time_sleep_until() function that sleeps until the
provided timestamp has been reached. Some other functions
have added new parameters, but let’s face it: you want to know
what PDO is and why you should care about it. Alrighty then…
The what, why and how of PDO
Back in LXF48 we looked at how to use the Pear::DB package to
provide database-agnostic SQL querying in PHP scripts.
Pear::DB accomplishes this mighty feat by forcing you to use
plain SQL, and abstracting the database function calls so you
can move between servers easily. Its weakness is lack of speed:
Pear::DB itself is written in PHP, and needs to jump through
various hoops to call the right functions. PDO is designed to take
half of Pear::DB – the part that handles many databases
through combined function calls – and move it into the core of
PHP as a C extension. This is much faster, and still leaves
Pear::DB as a viable project to handle the SQL abstraction.
Confused yet? Bear with me. In PHP 5.0, if you wanted to
write SQL that would work on any database, using PHP that can
connect to any database, you needed to use Pear::DB. Pear::DB
handles the part of connecting to multiple databases by
exposing functions like query() and replacing them with mysql_
connect(), pgsql_connect() and so on depending on your
connection. This last part is very slow, so this is the part that
PDO takes over. Instead of changing query() for the correct
function then calling that instead, PDO calls the function directly.
Using PDO isn’t the end for Pear::DB: it still has a valuable life
ahead of it. But for the rest of us who just want a nice way to
move between the mysql and mysqli extensions, PDO is perfect.
There’s no point talking about PDO if we’re not going to put it
into practice, so here goes. Our first script will connect to a local
MySQL server with the username ‘lxf’ and the password
21/9/05 7:30:01 pm
 TUTORIAL PHP

‘r0xx0r5’, then open the database ‘filch’ and store the connection Database-
for later use. All set? independent
SQL
$db = new PDO(“mysql:dbname=filch;host=127.0.0.1”, “lxf”,
“r0xx0r5”);
If you’re expecting five more lines of code, I’m sorry to
disappoint: that’s all you need to connect and get ready for a Pear
DB
query. If you’re used to calling mysql_connect() and mysql_
select_db(), then you’ll notice it has all been amalgamated into
a Pear::DB-style data source name (DSN). A PDO DSN is made
up of various key=value fields, preceded by the type of database
you want to connect to. In the example above, mysql is the MySQL- Postgres- Oracle- Other DB-
database type, filch is the name of the database to select (the specific specific specific specific
SQL code SQL code SQL code SQL code
dbname key), 127.0.0.1 is the IP address to connect to (host),
lxf is the username and r0xx0r5 is the password.
You need the appropriate PDO driver installed to connect to
a given database type. In the example above we’re using ‘mysql’,
which means we need to have the MySQL driver installed – ie
PDO
compiling PHP using --with-pdo-mysql. The PDO extension is
compiled in by default, which means you only need to select the
drivers you want to add to it. Other drivers include --with-pdo-
pgsql for PostgreSQL, --with-pdo-oci for Oracle, and so on.

Quick queries MySQL Postgres Oracle etc

Of course, that’s such a simple script that it’s not surprising it’s
just one line, so let’s expand it to query the database. For this
we’ve created a very simple table called ut_results, which // eek! We can’t possibly let *that* get committed to the Pear::DB and PDO can
contains the names of players in Unreal Tournament games, plus table... happily live together: one
handles converting your
the winners. With that table, here’s the PHP code you need to $db->rollBack();
SQL, and the other links
query it: ?> multiple back-ends to one
foreach($db->query(“SELECT Winner from ut_results;”) as set of classes and methods.
$row) Catching errors
echo $row[“Winner”], “\n”; The last thing to look at is how to catch errors that get thrown
Yes, it only takes two lines to do. If you’re pining for the days when something doesn’t work as planned. PDO, being a modern
of mysql_query() and mysql_fetch_assoc() loops, then I’m PHP exception, can be made to use other PHP exceptions to
sorry: PDO is just easier to use. The nice thing about the above handle fatal errors, rather than just send back unusual return
code is that it encapsulates the entire query and extraction into values. By default it adopts a mix of things: fatal errors (‘your
just one line. At first sight you might think that it runs password was rejected’) use exceptions, whereas lesser errors
$db->query() multiple times, but it doesn’t – if you’d like to (‘that query didn’t work’) quietly send back false return values. I
confirm this for yourself, go into the MySQL monitor and run the don’t know about you, but I prefer it when extensions act as full
status command to see the query count. exception citizens: I want everything as an exception, darn it!
You can make your querying even easier by activating the Fortunately, PDO lets us tweak its behaviour to force exceptions
aptly named lazy mode. Here, rather than reading each row as for all errors, with the setAttribute() function. Here’s the code:
an array, you can have the whole thing converted into an object <?php
for you using the field names as the variable names. To do this, try {
just pass PDO_FETCH_LAZY as the second parameter to $db = new PDO(“mysql:dbname=filch;host=127.0.0.1”,
$db->query(), like this: “lxf”, “r0xx0r5”);
foreach($db->query(“SELECT Winner from ut_results;”, PDO_ $db->setAttribute( PDO_ATTR_ERRMODE, PDO_
FETCH_LAZY) as $row) ERRMODE_EXCEPTION );
echo $row->Winner, “\n”; try {
If your database supports transactions, these are also $result = $db->query(“SELECT Winner from ut_sresults;”,
available. You’ll need MySQL 4.1, PostgreSQL, Oracle or some PDO_FETCH_LAZY);
other compliant database; without one of these, the transaction foreach($result as $row) {
commands are ignored. To get started, run the echo $row->Winner, “\n”;
beginTransaction() method on your database object, then go }
ahead and run all the queries you want. As long as you don’t } catch (PDOException $ex) {
alter the structure of tables by adding or removing columns echo “Query failed: {$ex->getMessage()}\n”;
(some databases will flush the transaction in this situation) you’ll }
be fine. Once you’re done querying, run either commit() or } catch (PDOException $ex) { NEXT
MONTH
rollBack() depending on whether everything went OK or not. echo “Couldn’t connect: {$ex->getMessage()}\n”;
Here’s an example: }
<?php ?> Rich treasures await you
$db = new PDO(“mysql:dbname=filch;host=127.0.0.1”, “lxf”, Using two try/catch statements we can catch two when you know how to
“r0xx0r5”); different types of exceptions using the same interface with the
$db->beginTransaction(); PDOException class: the outer one catches problems Google API.
$db->query(“INSERT INTO great_bands (Name) VALUES connecting to the database, and the inner one catches
(‘Marillion’); bad SQL queries. Easy as pie. LXF

www.linuxformat.co.uk LXF73 DECEMBER 2005 91

LXF73.tut_php 91 21/9/05 7:30:06 pm

 TUTORIAL PHP
SCRIPT PROGRAMMING
PHP Website authentication
Paul Halfpenny explains how to set up authenticated areas of a website with PHP and MySQL.
THE AUTHOR
Paul Halfpenny is product
manager at Hostway UK,
where he coordinates R&D
between the hosting
company’s international
technology departments. He
is a keen proponent of open
source technologies and a
proficient PHP coder.
92 LXF73 DECEMBER 2005
LXF73.tut_authentic 92
There are many reasons to create secure areas on a
website that you can keep other people out of. For
instance, you may wish to create an administration
area where you can update your website or upload new
documents. This works both as source control and a way to allow
non-technical staff to carry out basic tasks.
But the most popular reason in the past few years has been
to create member’s areas – parts of your site that are only
accessible to visitors who have registered for content or pay a
subscription fee. In this model, the site owner needs to ensure
that the visitors entering the member’s areas are legitimate
users, and the best way to achieve this is to provide a username
and password for each visitor. But how do you ensure that the
username and password are correct?
Well, there are a couple of ways to do this. You could use
basic HTTP user authentication, hard-coding a specific
username/password combination into a script. Or you could
authenticate against a database. We will look at both here.
HTTP authentication with PHP
First, hard-coding. Basic HTTP authentication uses a challenge–
response scheme to authenticate users attempting to access a
password-protected page. The challenge process begins when
the user requests a file from a web server. If the file is in a
protected area, the server responds by sending out a 401
(unauthorised user) string in the header of the response. The
browser detects that response, and up pops the username and
password dialog box. The user enters their username and
password in the dialog box, then clicks OK to send the
information back to the server for authentication.
If the username and password pair is valid, the protected file
will be displayed to the user. The validation will carry through for
www.linuxformat.co.uk
as long as the now-authenticated user is within the protected
area. However, if the username and password typed into the
dialog box cannot be authenticated, the dialog box will again be
displayed, prompting the user to try again. This cycle will be
repeated until the proper username–password combination is
entered or the user gives up.
A simple PHP script can mimic the HTTP authentication
challenge–response system by sending the appropriate HTTP
headers that cause the automatic display of the username/
password dialog box. PHP stores the information entered in the
dialog box in three global variables ($PHP_AUTH_USER,
$PHP_AUTH_PW, and $PHP_AUTH_TYPE). Using these
variables, you can validate input against a list of login details kept
in a text file, database, or any other list you have lying about.
Let’s start slowly, by writing a PHP script that simply checks
for a value (any value) for $PHP_AUTH_USER. If no value exists,
the script will send a 401 unauthorised message in the header.
This header will cause the username–password dialog box to
appear, and execution of the script will halt. After the user enters
some values in the dialog box and presses the OK button, the
values will be sent and the page will reload. When a value like
$PHP_AUTH_USER is entered, the first section of the script will
be skipped and the header information will not be sent.
<?php
// File Name: auth01.php
// Check to see if $PHP_AUTH_USER contains info
if (!isset($PHP_AUTH_USER)) {
// If empty, send header causing dialog box to appear
header(‘WWW-Authenticate: Basic realm=”Secret Area “’);
header(‘HTTP/1.0 401 Unauthorized’);
echo ‘Authorization Required.’;
exit;
20/9/05 6:44:31 pm
 TUTORIAL PHP

} else { +-------------+----------+----------+
// If not empty, display values for variables | real_name | username | password |
echo “<P>You have entered this username: $PHP_AUTH_ +-------------+----------+----------+ Authentication with CGI
USER<br>You have entered this password: $PHP_AUTH_ | Joe Smith | joe | ai890d | The $PHP_AUTH_USER,
PW<br>The authorization type is: $PHP_AUTH_TYPE</p>“; +-------------+----------+----------+ $PHP_AUTH_PW, and
} | Jane Smith | jane | 29hj0jk | $PHP_AUTH_TYPE global
?> +-------------+----------+----------+ variables are available only
Now we’re going to validate a username–password pair | Mary Smith | mary | fsSS92 | when PHP is installed as a
against values hard-coded within the script. This technique is +-------------+----------+----------+ module within Apache. If
you’re using the CGI version
useful if there is only one username–password combination that | Bob Smith | bob | 2NNg8ed |
of PHP, you’re limited to
everyone uses (for example, the administrators of a website). +-------------+----------+----------+
.htaccess-based
This script is quite similar to the first example, but the main You’ll need to look at all the records in the database when a
authentication or database-
difference is that if a value has been entered, the script runs an visitor enters a username and password to find a match. To do driven authentication using
if/else statement: If an incorrect value has been entered for the this you can use an SQL statement, similar to this: HTML forms to input the
username or the password, display that dialog box again. But if SELECT * FROM users WHERE username=’$PHP_AUTH_ username and password, and
the username is valid (else), send a message to that effect. USER’ and password=’$PHP_AUTH_PW’ PHP to validate matches.
<?php With this SQL query, we are looking to get a result value that
// File Name: auth02.php we can act on. Counting the number of matched rows will be
// Check to see if $PHP_AUTH_USER contains info enough, since the number will be either zero (no matches) or
if (!isset($PHP_AUTH_USER)) { one (a record in our Users table with a match):
// If empty, send header causing dialog box to appear mysql_connect(“hostname”, “username”, “password”) or die
header(‘WWW-Authenticate: Basic realm=”Secret Area “’); (“Unable to connect to database.”);
header(‘HTTP/1.0 401 Unauthorized’); mysql_select_db(“dev_i2ii_com”) or die (“Unable to select
echo ‘Authorization Required.’; database.”);
exit; $sql = “SELECT * FROM users WHERE username=’$PHP_
} else if (isset($PHP_AUTH_USER)) { AUTH_USER’ and password=’$PHP_AUTH_PW’”;
if (($PHP_AUTH_USER != “user”) || ($PHP_AUTH_PW != $result = mysql_query($sql);
“password”)) { $num = mysql_numrows($result);
header(‘WWW-Authenticate: Basic realm=”Secret Area”’); Finally, you need to put the connection sequence in the
header(‘HTTP/1.0 401 Unauthorized’); context of your authentication script. This will give you a script
echo ‘Authorization Required.’; that works through the code flow in a structured format.
exit; <?php
} else { // File Name: auth04.php
echo “<P>You’re authorized!</p>“; if (!isset($PHP_AUTH_USER)) {
} header(‘WWW-Authenticate: Basic realm=”Secret Area “’);
} header(‘HTTP/1.0 401 Unauthorized’);
?> exit;
Try entering an incorrect username and a correct password, } else if (isset($PHP_AUTH_USER)) {
or a correct username and incorrect password, or an incorrect // If non-empty, check the database for matches
username and an incorrect password. You’ll notice that the mysql_connect(“hostname”, “username”, “password”) or
server will not validate the combinations. Now try using ‘user’ for die (“Unable to connect to database.”);
the username and ‘password’ for the password. This time you mysql_select_db(“dev_i2ii_com”) or die (“Unable to DB.”);
will be authorised. $sql = “SELECT * FROM users WHERE username=’$PHP_
The drawback with this method is that many sites will require AUTH_USER’ and password=’$PHP_AUTH_PW’”;
a unique username and password for each user, to ensure that $result = mysql_query($sql);
the site owner can track the number of visitors gaining access. // Get number of rows in $result. 0 if invalid, 1 if valid.
Of course, with information on the internet so easy to exchange, $num = mysql_numrows($result);
unique user–pass combinations also help to make sure that their if ($num != “0”) {
revenue model does not dry up after the first customer! echo “<P>You’re authorized!</p>”;
exit;
Authenticate against a database } else {
To avoid this you can keep the information in a text file or in a header(‘WWW-Authenticate: Basic realm=”Secret“’);
database, and match those values just as easily as you matched header(‘HTTP/1.0 401 Unauthorized’);
hard-coded values. Either is valid as a method, but with echo ‘Authorization Required.’;
increasing visitors to a site, it does make more sense to use a exit;
database that can handle multiple requests and can be updated }
on the fly. }
Here we’ll be exploring how to match usernames and ?>
passwords to a list residing in a database table. This example We’ve discussed two types of authentication, each as suitable
uses PHP’s functions to connect to a MySQL database, but you’re as the other for protecting secure areas of your website. The
free to substitute any of the built-in database connectivity choice of which to use will be largely dependent on the needs of
functions in order to make this work with any other database your site. This will become apparent when you decide how many
that PHP supports. users will need to be given login credentials. The principle is the
To store your usernames and passwords you will need to same – allow access if the values entered are the same as those
create a table in a database to hold the information. Our table is stored. The only decision you need to make now is who should
called ‘Users’ and looks like this: be allowed access. LXF

www.linuxformat.co.uk LXF73 DECEMBER 2005 93

LXF73.tut_authentic 93 20/9/05 6:44:39 pm

 ANSWERS

Answers
If you really are stuck and the HOWTOs yield no good result, why not write in?
Our resident experts will answer even your most complicated problems.

OUR EXPERTS Triple booting

Q
I have a 40GB hard drive
Whatever your question is, we can split into two 20GB
find an expert to answer it. From partitions for Windows
installation and modem woes to XP and Knoppix. I made
network administrations, we will get
a new 10GB partition and installed
the answer for you – just fire off a
letter or email and it'll all be taken Fedora Core 4 from the [LXF70]
care of. coverdisc, wanting a triple boot so
I could choose Windows, Knoppix
or Fedora Core. When I rebooted
Having run a small after installation, only Fedora Core
ISP and produced
and Windows were on the boot
coverdiscs for
Linux Format, menu, not Knoppix. I went into
Neil Bothwick Fedora Core, and Knoppix is still
describes himself there on hda6. I installed Knoppix
as a Linux jack-of-all-trades.
again hoping that it would add
Fedora to the boot menu, but it did
not. Now it only shows Windows XP
Nick Veitch is the or Knoppix. Adding extra distros to your PC’s boot menu is as easy as copying and
editor of the pasting, but you usually have to do it yourself.
magazine, and
Is there a way to add Fedora to
answers your easy the boot menu so I can pick XP,
questions! Or Knoppix, or Fedora Core? My setup mkdir -p /mnt/fedora options for triple (and more) booting
indeed anything to is as follows: mount /dev/hda3 /mnt/fedora with Linux, there is an excellent
do with Grub, Lilo, Netatalk, Vi...
hda1 Windows XP Then select System > More reference at http://home.planet.
hda3 Fedora Core Applications >File Manager – nl/~elst0093/motub/multboot.html.
hda6 Knoppix Superuser Mode from the KDE menu, NB
Hans Huberland From the LXF forums navigate to /boot/grub and load
is Rackspace This is a fairly common grub.conf. Now go to the same Server migrations
Managed Hosting’s
Linux expert and a
real-life system
administrator on
call for your questions.
A situation when installing a
second Linux distro. The
installers are good at
detecting an existing Windows system
and adding the relevant option to the
directory in your Fedora setup and do
the same. Highlight the three lines in
the Fedora file starting with ‘title
Fedora Core’ and copy them to the
Knoppix file. Save and reboot and you
Q
I have had a dedicated
Linux server with a
hosting company for
about six years or so. As
part of their customer-retention
Send your questions for our experts to:
boot menu, but very few will pick up should get your three way choice. strategy they have arranged for me
Linux Format, Future Publishing,
on other Linux installations. If the Fedora installer used LVM to upgrade to a newer server with
30 Monmouth Street, Bath BA1 2BW
or email: lxf.answers@futurenet.co.uk.
The result, as you have already (Linux Volume Manager) when better bandwidth allowance, SLA
Got a sysadmin query? Send it to Hans discovered, is that your previous distro partitioning the disk, you might not be [service level agreement] and so on.
at sysadminqa@rackspace.co.uk. is still installed, but there appears to able to access your Fedora files from The server is used by my very
be no way to boot into it. As a short- within Knoppix, as Knoppix doesn’t small business for email, web
term solution, you can usually use the support LVM. In this case, you need to hosting, a testing server and
installation disc in rescue mode to perform the process from Fedora, so managing DNS for domains that we
boot your installation. Pressing F1 after you’ll need to first reinstall the Fedora own. A few friends and family also
booting from the disc usually shows Core boot loader from the installation use the server for their email and
the options. disc. Start the installer as before and web hosting.
The long-term solution is to add an select the Upgrade An Existing Is there a best-practice way of
entry to your current boot menu for Installation option. Select the Update transferring DNS, websites, home
the hidden distro. This is made easier Bootloader Configuration option to directory files and email (Sendmail)
here because both Knoppix and reinstall the Fedora bootloader. to the new server with minimal
Fedora Core use the Grub bootloader. Now let the update finish, boot into disruption for myself and other
You need to load the Grub Fedora and copy the relevant section users? I cannot afford to pay the
configuration files from the two distros of the Knoppix boot menu to that of hosting company’s staff to do the
into a text editor. First you will need to Fedora core. transfer for me.
mount the other distro’s partition with If you are interested in the various Lee

94 LXF73 DECEMBER 2005 www.linuxformat.co.uk

LXF73.answr 94 23/9/05 10:30:44 am

 ANSWERS

Server migration, especially

A when changing distribution

and versions (MySQL,
Apache and the like), is
never a straightforward procedure.
However, it is a good opportunity to
Four starters

revise your current setup, and just as
when moving house, you will come
across a lot of stuff that can be thrown
out. Also, this is a perfect opportunity
to audit all your configurations and
create and test backup and disaster
recovery strategies.
On the new server, set up and test
all the zone files, email configurations,
user accounts and websites. You can
trick your workstation into believing
that the sites are on the new IP by
modifying your hosts file (/etc/hosts
on Linux, c:\windows\system32\
Q
Regarding booting from
an external USB or
CompactFlash device – I
have four questions:
ALL ABOUT RACKSPACE
LXF Sysadmin
Answers – in
association
with Rackspace
1. Is it possible to boot from a USB Managed
flash drive if your BIOS doesn’t Hosting
support USB booting?
2. Can a boot CD with Smart Boot
Manager pass off the process to
the USB?
3. Is it possible to fit an
uncompressed Knoppix or Kanotix
CD to a 2MB CompactFlash or USB? When it comes to managed hosting the outset and will remain on your
Rackspace is unique. We are the only account. All Rackspace employees are
4. What parts of the system would
hosting company to guarantee 100% committed to ensuring customer
be better off on the hard drive, eg
network uptime and we will even satisfaction, and they will not rest until
swap or logs? pledge to replace faulty hardware (let’s a client problem is resolved.
From the LXF forums face it, it does happen) within an hour. Since Rackspace was established in
I’ll answer each of your

drivers\etc\hosts on Windows XP).
DNS is the first target for migration.
Point NS records at the registrar to
the new name servers at least three
days ahead of the planned switchover.
Reduce the TTL (‘time to live’) on your
new records to a reasonable 35
minutes so that on IP switchover
visitors will clean their cached records
of your IPs and pick up the new ones
in no time at all. Set the TTLs back to
a more internet-friendly value, say 24
hours, a few days after the migration.
Depending on your budget and
how critically reliant your sites are on
a database back-end you may have to
take extra precautions on the way you
move over your databases.
Finally, tend to your log files and
web statistics. They’re often forgotten
during migrations. HH
A
questions in turn.
1. Yes, but you will need
a boot floppy to do this. For
example, Puppy Linux has a floppy
disc image on the website that can be
used to boot it from a USB flash disc.
See www.goosee.com/puppy.
2. No. Smart Boot Manager does
not support booting from USB devices.
3. It should be, although you
shouldn’t uncompress the image. USB
reading is quite slow, so any
reasonably powerful machine would
probably uncompress faster than it
reads, making the compressed image
faster. There are instructions for
installing Knoppix on to a 1GB (or
larger) flash disc at www.knoppix.
net/forum/viewtopic.
php?p=64999#64999. Alternatively,
Damn Small Linux can be installed on
As you would expect from Red Hat’s
only Advanced Hosting partner you will
receive Fanatical SupportTM and instant
emergency response from fully-
qualified level three technicians,
available 24 hours a day, 365 days a
year. No answering machines, no
diversions, no silly on-hold messages.
All our customer platforms are
housed in state-of-the-art, secure data
centres and can be fully customised to
meet any requirement. A dedicated
account manager is allocated to you at
a USB device, and DSL is a cut-down
Knoppix. So you should be able to put
a full Knoppix on to a 1GB flash disc
by booting DSL, running
sudo dsl-usbinstall to install it to the
USB device and then replacing the
knoppix/knoppix file with the
full-sized one from a Knoppix CD.
1998, it has specialised in managed
hosting – nothing else. We’re not an ISP,
we don’t sell domain names, and we
won’t offer website design – all we do
is managed hosting. As a result, we do it
very well and 97% of our customers
would happily recommend us.
For information on how we can
improve your web hosting, please
contact us at www.rackspace.co.uk
and ask any questions you have about
the ultimate managed hosting service.
See page 99 for Rackspace’s star letter.
4. Anything that involves writing,
especially frequent writing like swap
and log files, should not be placed on
the flash device. Flash memory has a
limited write lifetime, typically between
100,000 and 1,000,000 writes (and
as little as 10,000 writes for older >>
devices). If the system is continually

A QUICK REFERENCE TO: SU

Everyone knows you should never log
in to your desktop environment as
root – it’s not safe. So what should you
do when you need to run a command as
root? Some programs, such as the
KDE Control Centre, will ask for your root
password when you want to do
something that needs root access. At
other times, you can use the su
command, which we’ll explain here.
Su does not, as is often thought, stand
for super user, but for switch user. We
can use it to temporarily log in as
another user. Unless you give it the name
of the new user, it logs you in as root
after you give it the correct password.
Now you can do whatever you need to
do, before pressing Ctrl+D to log out and
revert to your normal user. Depending on
how your distro sets things up, you may
not be able to run X programs after
running su. For example, SUSE 9.3
allows you to run X programs, but
SUSE 9.2 does not. If this happens to
you, install Sux. It almost certainly comes
with your distro, but may not be installed
by default. Sux works like Su but also
enables X explicitly.
If you only want to run a single
command as root, you can pass it to su,
for example:
su -c “kwrite /etc/fstab”
The other option you may need to use
is ‘-’ on its own. The su command
normally gives you root privileges, but
you still have all the settings, such as
path and home directory, of your normal
user. In contrast, su - gives you a full root
login. The difference may seem subtle,
but it can be important. If you have
trouble running a system command with Switching user with the su command (lower window) adds a new entry
su, try su -. in the system log (upper window).

www.linuxformat.co.uk LXF73 DECEMBER 2005 95

LXF73.answr 95 23/9/05 10:30:48 am

 ANSWERS
FREQUENTLY ASKED QUESTIONS LIVE CDS
FAQ What is a Live CD?
It is a distro on a single CD (or
DVD) that runs directly from the disc.
FAQ Aren’t all Linux
distribution CDs
bootable? How are Live CDs
different?
A typical distro CD boots into an
installer, so that you can put the distro
on your hard disk and run it from
there. A Live CD distro runs direct
from the CD, booting straight into the
desktop in most cases. There is
nothing to install, just boot and run.
FAQ Can I save my work
using a Live CD?
Not to the CD of course, they are
read-only. However, most Live CD
distros make some provision for saving
work and settings. The favourite
approach is to keep the home
directory on a USB pen drive, so all
your work and settings go here.
Alternatives are floppy drives and hard
disk directories. Some Live CDs, such
as ATmission, use a clever trick to
make it look like you are writing data
back to the CD, when it is actually
being written to a special file on a
hard disk or pen drive. The advantage
>> could
writing to the same area of memory, it
fail in a few months.
Your choices are: to put these on
the hard drive, which limits portability;
to write logs to a RAMdisk and copy it
back to the flash drive on shutdown,
which would limit the writes to one per
session; or to use the JFFS2 filesystem
from http://sources.redhat.com/
jffs2. This is a special filesystem
designed to address this problem with
flash devices. If you want a
ready-made distribution for a flash
drive, try Flash Linux from
http://flashlinux.org.uk. NB
Ejected and rejected
Q
I have very recently
installed Mandriva 2005
LE from the [LXF67]
coverdisc. The install
went very well for a first time Linux
user. But I have a problem with
installing some of the additional
programs. Sometimes I am informed
96 LXF73 DECEMBER 2005
LXF73.answr 96
of this approach is that you can
effectively change any file on the CD,
not just your home directory. You can
even install new software.
FAQ What are the benefits of
Live CDs?
You can run them anywhere, on any
computer capable of booting from a
CD. Want to buy a new laptop? Boot it
from a Knoppix CD to see whether all
its hardware is recognised and
configured – an instant Linux
compatibility test. Live CDs are also
useful for disaster recovery and other
maintenance. If you made a mistake
editing a critical file and your
computer won’t start up, boot a Live
CD and fix the damage. If you want to
swap your hard disk for a larger model,
that Mandriva needs to have my
DVD in drive hda (my CD drive) and
I must then hit Enter. When I put it
in, the disc is immediately ejected –
not surprising as it is a DVD! If I try
putting the DVD in the DVD drive
instead and hit Enter, the disc is
again ejected. If I then simply try to
abandon that task I am forced to
end the session to do so.
My /etc/fstab contains
/dev/hda /mnt/cdrom auto umask=0,
user,iocharset=iso8859-15,codepage
=850,noauto,ro,exec,users 0 0
/dev/hdc /mnt/cdrom2 iso9660
user,iocharset=iso8859-
15,noauto,ro,exec 0 0
From the LXF forums
This occasionally happens
A with Mandriva installations.
Despite the installer running
from the DVD drive, it will
sometimes add the path to your CD-
ROM (or CD-RW) drive to its list of
software sources. In your case, it is
looking at /mnt/cdrom when your
www.linuxformat.co.uk
We cannot cover
Live CDs without
mentioning Knoppix.
This is the latest
version, now available
on DVD with even
more software.
connect both and boot from MEPIS to
prepare the new disk and copy over all
your data.
FAQ And the disadvantages?
There’s the limited write ability,
and the need for a second device to
hold any files you do write. The other
is that a Live CD distro is slower in
operation than a normal hard disk
installation. The equivalent of your
hard disk is held in a large,
compressed file on the CD. Usually
containing around 2GB of files, it’s
packed into less than 700MB.
CDs have longer seek times and
slower read speeds than hard disks,
and then the data has to be
uncompressed. Expect to see longer
program loading times, but once
DVD is mounted at /mnt/cdrom2.
Fortunately, the solution is dead
simple. Put the DVD in the correct
drive and start the Mandriva Control
Center. Go into the Software
Management section and select Media
Manager. This will show you a list of
software sources – possibly only the
If Mandriva forgets where your DVD drive is, it is easy to remind it.
loaded they should run as fast as any
other distro.
FAQ What hardware will
they run on?
Most Live CD distros are written for
x86 hardware, so they usually run on
AMD64 boxes too. There are a couple
of dedicated AMD64 Live CDs, and
Ubuntu does one for PPC. Support for
specific items of hardware, such as
graphics chips and modems, is the
same as for any other Linux distro.
Where the Live CDs are different is in
the way they automatically detect and
configure any hardware they find while
booting. The Knoppix hardware
detection system is exceptional.
FAQ Where can I find these
Live CDs?
On our coverdiscs, of course (see
Kanotix with LXF69 or MEPIS with
LXF65). You can also download ISO
images from the following websites:
■ Knoppix www.knopper.net/
knoppix/index-en.html
■ Ubuntu www.ubuntulinux.org
■ ATmission www.atconsultancy.nl/
atmission
A long list of Live CDs is at www.
linuxlinks.com/Distributions/LiveCD.
DVD. Click on the DVD entry, press
Edit and change the URL entry from
removable://mnt/cdrom/media/
main to removable://mnt/cdrom2/
media/main, that is, the path to your
DVD drive. Click on Save Changes and
all should be well. NB
23/9/05 10:30:50 am

Desert foxed
Q
I have to SSH into three
Red Hat servers at one
desert outpost (I work for
an oil company). The
TCP/IP connection to the field
servers is an unreliable internet over
satellite link. Very often I find myself
losing connectivity halfway through
an operation, and if I leave a session
open for more than 15 minutes, the
satellite router (to which I have no
access) rudely drops my connection.
I know that I can run most
applications in the background but I
am looking for a solution to
maintain a persistent connection. Do
you know if there is a budget
solution that I can implement?
Greg
Yes, there is. The Nohup
A command runs a command
immune to hang ups, with
output to a non-TTY, while
Screen is a full-screen window
manager that multiplexes a physical
terminal between several processes
(typically interactive shells).
I am a big Screen fanatic. When
invoked as Screen, you can create new
windows through a Ctrl+A C with
Ctrl+A “ (note that is double quotes)
to list and select active windows.
Ctrl+A D detaches from Screen, to
which you can reattach by invoking
screen with the -r parameter. If your
session drops and you want to
reattach to a screen that hasn’t been
properly detached, invoke as
screen -x. Amazing! HH
Fstab in the dark
Q
I have an old but
dependable Compaq
Deskpro EN (733MHz,
488MB of RAM) which
I’ve set up with a 10GB hard drive
with Windows XP. I also have the
My Documents directory including
all my music files, and my SUSE
LXF73.answr 97
Linux 9.2 installation, on a 40GB
hard drive. This allows me to
completely max-blast Windows
when I want to start again without
affecting my documents or Linux.
My problem is that SUSE cannot see
either of my Windows directories –
when I look under storage devices it
shows only the Linux filesystem.
I’ve tried Mandriva and Fedora
Core 4, but the same thing happens.
I really want to be able to listen to
my music in Linux (and Windows)
without changing my
windows/documents partition to
FAT32. Any ideas?
Carl Draper
If you installed SUSE after
A Windows, the installer
should have picked up your
Windows directories and
added them to /etc/fstab, so they
would be mounted on boot. If this is
not the case, you need to add them to
/etc/fstab manually.
This has to be done as root, so
open a terminal, type sux - to log in
as root and give your password. Use
whatever editor you prefer to change
the file, for example:
sux -
<enter root password>
kwrite /etc/fstab
You need to add two lines to the
end of this file. This example assumes
your Windows NTFS partitions are
the first primary partition on each
drive. Otherwise, change the device
names accordingly.
/dev/hda1
ntfs
/windows/C
ro,users,gid=users,umask=0
002,nls=utf8 00
/dev/hdb1 /windows/D
ntfs ro,users,gid=users,umask=0
002,nls=utf8 00
Save the edited file, then create
the mount points. Mount the partitions
and check that they are mounted.
mkdir -p /windows/{C,D}
mount -a
df -h
SUSE’s
installer
should have
detected your
Windows
NTFS partition
and set up
read access. If
not, you only
have to add a
couple of lines
to /etc/fstab.
www.linuxformat.co.uk
Provided the mount command
gave no errors and df showed the
partitions, you should have full read
access to them. Write support for
NTFS is very limited in Linux, so
Windows partitions are usually
mounted read-only. If the mount
command gave an error along the
lines of ‘wrong fs type, bad option, bad
superblock...’, double-check the options
you typed into /etc/fstab. NB
No boot Ubuntu
Q
I wonder if you can help.
I am running Ubuntu and
during boot it says the
following:
‘fsck failed. Please repair
manually.
* CONTROL-D will exit from
this shell and continue
system startup.
root@(none)::~#’.
Ctrl+D does indeed continue the
boot process and brings me to the
login screen. Logging in gives the
error message:
‘Your home dir is listed as:
/home/john, but it does not
appear to exist. Do you want
to log in with the / (root) dir
as your home dir. It is unlikely
that anything will work unless
you use a failsafe session’.
From the LXF forums
Despite what it sounds like,
A Fsck is not a chain of
clothes shops, nor a strange
Linux curse. It is the
FileSystem Check program, which
performs a similar function to the likes
of Scandisk.
The first message means that the
boot process has detected an error on
one of your partitions that needs your
attention. It wants you to do this
before pressing Ctrl+D to continue.
By pressing Ctrl+D straight away you
have left it in its faulty condition, so
the partition could not be mounted.
Presumably, this partition is mounted
at /home, which explains why
/home/john cannot be found when it
fails to mount.
The error message should have
told you which partition was affected,
for example /dev/hda6. If not, typing
grep /home /etc/fstab
will tell you which it is. Now type the
following code, replacing ‘N’ with the
partition number:
fsck -f /dev/hdaN
After some disk activity and various
screen messages, Fsck should exit
without an error. To be safe, I prefer to
run the command a second time, to
ANSWERS
make sure things really are fixed. Now
you can press Ctrl+D to continue with
the boot process.
If the problem persists, it is likely
that your disk has a fault. Your first
action should be to back your data up,
now – not tomorrow. Then you should
install the Smartmontools package –
from http://smartmontools.
sourceforge.net or an Ubuntu or
Debian repository, and run
smartctl /dev/hda
for a report on your disk’s health. NB
Virtually gone
Q
I have just installed
Fedora Core 4 and have
one quick question –
where have all of the
virtual terminals gone in this
release? Pressing Ctrl+Alt+F1 to F6
just gives blank screens.
Marc McRae
A
This is a known bug. The
fault lies in the file
/usr/X11R6/lib/modules/
libvgahw.a so move this
somewhere safe (don’t delete it, just in
case) and replace it with the same file
from Fedora Core 3. If you do not
have FC3 installed, you can get the
file from http://rapidshare.de/
files/2399145/libvgahw.a.html. This
is advertising-financed web space, so
you will need to scroll down to the
bottom of the page, click
the Free button then
scroll to the bottom of
the next page for the
actual download link.
Copy the replacement
file to /usr/X11R6/lib/
modules/ and reboot.
Your virtual terminals
should now be back.
Zombie
fears
Q
I have
recently
been
humiliated
by my ISP for spamming.
It turned out that a forum
I had set up on my home
box had been hacked and I
was mass mailing the whole
world. That issue has been
fixed but now I’m losing
sleep fearing that my server is
a zombie. Do you recommend
tools or frameworks for quickly
and reliably telling if a machine
has been broken into? >>
Andy
LXF73 DECEMBER 2005 97
23/9/05 10:30:53 am
 ANSWERS
>> While there is no short
A answer to that, there are
some simple steps that can
be followed to reveal most
common scripted break-ins.
1. Use ls -lai to determine whether
there are any files in /tmp and
/var/tmp that shouldn’t be there. In
particular, watch out for executables,
scripts and text files that are full of
email addresses.
2. On RPM-based systems it is
possible to verify whether system
utilities such as Ls, Ps, Netstat and so
forth have been replaced with ones
that hide the hacker’s activity. On a
Red Hat-based system, the following
packages should be verified using
rpm -V <packagename>, for the
following packages: util-linux,
coreutils, net-tools, procps and lsof. computers running XP. I’m also
3. Check running processes with
ps -auxf.
4. Use netstant -tanp to find out
whether there are processes listening
on strange ports, or inexplicable
amounts of outgoing traffic. The -p
option shows which program is being
used. Very often this is named in a
way to make it look like a legitimate
program (such as httpd). The lsof
command can also list listening ports.
5. Review /etc/passwd to see if
any users have been added to the
system or have had their UID changed.
It’s a good idea to compare to a
Gtk-chtheme and the gtk-qt module for the KDE Control Centre can control the look of GTK apps in KDE.
98 LXF73 DECEMBER 2005
LXF73.answr 98
known clean copy of the password file.
6. Check Apache log files for
tell-tale signs of exploits where utilities
such as Wget were used to download
some form of malware. Check other
system log files for anything suspicious
– in particular, for log files that have
been redirected to /dev/null.
7. Finally, Chkrootkit (www.
chkrootkit.org) checks your server
for signs of rootkit presence. HH
SUSE networking
Q
I just installed SUSE 9.3,
and am having a couple
of networking problems.
I’ve got the system all
running smoothly so far but I’m
trying to share my internet
connection with two other
trying to have my Linux machine
identify my XP network for mapping
drivers and view shared folders and
so on. I have a problem with my
cable modem: now and again my
internet connection does not work
after booting up. Can you
recommend a good book that
covers issues like this without too
much technical stuff?
James Caven
You need two network
A interfaces on the computer
– one to the local network
and one to the internet. The
www.linuxformat.co.uk
Browsing Windows shares in KDE is as easy as typing smb:/ in the location
bar. Too much typing for you? you could always set up a bookmark.
former would be your local Ethernet,
the latter your cable modem, which
could be Ethernet or USB.
You can set up connection sharing
from SUSE’s Yast. First you need to
make sure your internet connection is
working properly on the SUSE
machine, then turn on routing by
going to Network Services > Routing
in Yast and ticking the Enable IP
Forwarding button at the bottom.
Press Finish and it’s all done.
Now you need to tell the other
computers where to look for their
internet connection. On each
computer, set the gateway address to
the IP address of the local interface
on the SUSE computer.
Browsing shared folders on your
windows machines is easy if you use
the default KDE desktop. Open a file
manager window and type smb:/
(that’s a single slash) in the location
bar. You’ll see a list of your workgroups
(usually only one) and you can browse
through here to access the various
shared folders.
If you want to share folders to be
accessed from the Windows
computers you will need to set up
Samba. Go to Internet & Network >
Samba in the KDE Control Centre, click
the Administrator button for root
access and set up any directories or
printers you wish to share. Make sure
your workgroup name, in the Base
Settings tab, is the same as on the
Windows boxes.
Finally, your intermittent cable
modem problem may be a timing
issue. Is this a USB modem? Does it
have an Ethernet option? If the
answer to both is yes, it would be best
to add another Ethernet card to your
computer and connect the modem
that way. If you are stuck with USB, it
is likely that the connection is not
coming up fast enough to be ready
when the computer boots. In this case,
the quick solution is to unplug and
reconnect the modem from your USB
port. This should force it to reconnect
to the ISP.
As for reading material: do you
have a boxed version of SUSE? The
SUSE manuals are some of the best
around, and have the advantage of
being specific to your distro. NB
23/9/05 10:30:57 am

WIN A NEUROS DIGITAL AUDIOCOMPUTER
with Rackspace Managed Hosting
www.rackspace.co.uk
Every month, the best question
related to systems administration
that a LXF reader sends in wins a
prize. This month you have the
chance to win the life-enhancing
Neuros MP3 digital audio computer.
With a Neuros, you can manage
all the music stored on your PC.
Create play lists. Delete songs. Get
new files, thanks to the Neuros
Synchronisation Manager. No more
getting stuck in front of the PC to
make changes – do it all on the
Neuros, and get on with it.
Then use its NeuroCast feature to
broadcast the music on your Neuros
device through any FM radio.
★ STAR QUESTION WINNER!
This issue's lucky winner is Dragan – your new Neuros digital audiocomputer will be with you shortly!
Sudo security
Q
My web developer has
been granted access to
FTP and SSH into a
dedicated server that
we are renting. He can upload
pages and manage MySQL
together with an Apache include
file for the server’s site-specific
configurations.
Since our company’s security
policies dictate that we cannot
disclose the root password to a
contractor, we are being called
by the developer to restart
Apache a number of times a day,
GTK on KDE
Q
I’ve been using Linux for
a while, but there’s
something I’ve been
puzzling over that I’ve
never worked out. I use KDE for my
desktop, but I still use some GTK
apps such as Gimp. Is it possible to
apply GTK themes to GTK apps
running under KDE? If so, how?
Mark
Yes, you can use Gtk-
A chtheme to preview and
select GTK themes. The
program is available from
LXF73.answr 99
NeuroCast automatically scans the
FM radio dial for an available
frequency and broadcasts using all-
digital stereo encoding, just like
broadcast towers used by
professional radio stations.
The Neuros has many recording
capabilities that make it a robust tool
for the recording hobbyist. The
device includes an on-board
microphone, perfect for the quick
memo or recorded lecture. It has
line-in recording capabilities and you
can record to MP3 or WAV, with a
choice of seven recording qualities.
Neuros also features a built-in
FM radio with five preset buttons so
which is not ideal. What do you
recommend?
Dragan
If you are running Webmin
A you will be able to create
a user that is restricted to
doing nothing but
stopping and starting Apache. First,
create a new user through
Webmin > Webmin Users and select
Apache Webserver. Click on the
Apache Webserver link to restrict
access specifically to whichever
aspects of Apache administration the
contractor needs.
Alternatively, if command line
access is preferred, Sudo becomes
http://plasmasturm.org/programs/
gtk-chtheme as source code or RPM
packages. There’s also a Debian
package available from the various
Debian repositories.
An alternative solution is a module
for the KDE Control Centre that adds
a panel for GTK Styles And Fonts to
the Appearance & Themes section.
You can get this from www.
freedesktop.org/Software/gtk-qt.
This module allows you to select a
theme in the same way that Gtk-
chtheme does, or apply your KDE
theme to GTK applications. I use both
programs, because the KDE Control
www.linuxformat.co.uk
WIN!
you can play your
favourite stations
with one click,
just like on your
car stereo.
Tempting, isn’t it?
You know what to
do: email
sysadminqa@
rackspace.co.uk.
the way to go. It is likely that a copy
of Sudo (www.sudo.ws) came
preinstalled with your distribution.
The sudo command allows
certain users or groups to execute a
number of commands as root or
otherwise specified. The
configuration file /etc/sudoers,
editable through visudo as root,
defines who can do what as who.
The configuration itself can be a bit
daunting, and time spent reading the
man pages is time well spent.
Here is a simplified configuration
that can be used to allow user
‘webman’ to execute the Apache and
MySQL startup files. The user will
Centre module has no preview facility.
I’d use Gtk-chtheme to browse newly
installed themes, or the KDE Control
Centre when I know which one I want.
You may as well install both – some
distros come with the KDE module
pre-installed – and make up your own
mind. NB
Debian desktop
Q
I installed Sarge from a
DVD distro. Everything
seemed to go OK. At the
end of the install, it asked
me to log in and then dumped me in
a full-screen Bash shell. I expected
ANSWERS
also be able to kill, as user ‘apache’,
any renegade process belonging to
user ‘apache’:
Cmnd_Alias HTTPD = /etc/rc.d/
init.d/httpd
Cmnd_Alias MYSQLD = /etc/
rc.d/init.d/mysqld
Cmnd_Alias
webman
KILL = /bin/kill
ALL = (root)
NOPASSWD: HTTPD, MYSQLD
webman ALL = (apache)
NOPASSWD: KILL
Usage:
$ sudo /etc/rc.d/etc/httpd stop
$ sudo -u apache kill 9982
$ sudo /etc/rc.d/etc/mysqld restart
This should set you straight. HH
to see a desktop environment. I
repeated the procedure using a net
load ISO from the Debian site. Same
result. What is going on? What do I
have to do to get a desktop? Why
didn’t the install create it for me?
From the LXF forums
Debian installs very little by
A default: just the basics to
get a core system working,
which does not include X.
During the second stage of the
installation, after the reboot, you are
asked to choose from software
collections. The first in this list is
Desktop Environment. It looks like this
>>
LXF73 DECEMBER 2005 99
23/9/05 10:30:59 am
 ANSWERS
The desktop environment collection has been selected – the web server collection has not.
>> is pre-selected, because the cursor is
in the selection box to the left of the
name, but it is not. Package groups
are only installed when there is a star
in the box (see screengrab above). You
need to explicitly select the groups
you want by moving the highlight bar
over them and pressing space. If you
simply press Enter at this stage
without selecting anything, you will get works over an encrypted SSH link.
exactly the system you describe.
All is not lost. There is no need to file with
reinstall. Log in as root and type
aptitude to load the package
manager. Highlight Tasks and press
Enter, move down to End-User and
press Enter, then highlight Desktop
Environment and press ‘+’ to select it.
Press G to see what will be installed
and G again to begin installation.
This will install both the KDE and
Gnome desktops – you will be able to
choose which you use when you log in. named id_dsa, to go in ~/.ssh, and a
There are a few basic configuration
questions to answer, but the defaults
are fine if you are unsure. You will also to create RSA instead of DSA keys.
be asked some questions to help
configure the graphical display. These
are the same as you would have been ~/.ssh on the server. Now SSH will use
asked during installation, had you
selected the desktop option.
Once installation has finished,
which will take several minutes, your
desktop should load the next time you
boot up. NB
Hiding passwords
Q
I have a simple shell
script that is scheduled
to download files from a
remote server by FTP. In
the shell script I have hard-coded
USERNAME and PASSWORD to
string variables to access the remote Ncftpput are part of the Ncftp package
server. How do I prevent the
USERNAME and PASSWORD being
100 LXF73 DECEMBER 2005
LXF73.answr 100
seen by others when they just open
up the shell script file?
Anonymous
The safest way to do this
A requires SSH access to the
server. If this is available,
you can use the scp
command to send the files. The
syntax for this is similar to cp, but it
For example, you would download a
scp -p user@server:/path/to/my/file
As it stands, this will still ask for a
password, but SSH has a means of
authenticating users by means of
keyfiles. If you do not already have a
keyfile pair, use ssh-keygen to
generate them. Full details are in the
man page, but ssh-keygen -t dsa will
create a pair with the default settings.
This generates two files, a private key
public key named id_dsa.pub. The
names will be different if you choose
Copy the public key to a file named
authorised_keys and put this in
the keys to authenticate and not
require a separate password.
If SSH is not an option, you will
have to use an FTP client to transfer
the files. Some of these have the
option to store passwords in a
configuration file, which you should
chmod to 600 so that only you and
the root user can read it. This is safer
than putting the password in a script
to be used when you run the
programs, because then the password
can be read with ps while the program
is running. For example, Ncftpget and
and accept a login definition file
instead of a URL. The file format is
www.linuxformat.co.uk
simple:
host ftp.host.com
user myuser
pass mypass
Then you can download the files
with a single line in your script
ncftpget -f login.def dest/dir path/to/
file1 path/to/file2 ...
where login.def is the file containing
the login information. Ncftp, the
interactive FTP client in this package,
is able to store encrypted passwords in
its bookmarks file, but this file is not
used by the non-interactive Get and
Put programs.
FTP is inherently insecure. Even if
your password is not stored anywhere,
it is still sent in plain text when logging
in. If security is important, you should
really look for an alternative means of
transferring the files.
Webministration
Q
We had a Fedora Core-
based solution set up as
an email, DNS and
firewall server for an
office of eight people.
We manage the server mostly
through Webmin, and we have our
phone system wired to the server’s
serial port where call activity gets
written to the log file /home/phone/
cdr.current. A Cron job emails the
previous day’s activity daily while
we review current telephone activity
over SSH a number of times a day.
cdr.current gets deleted
automatically at the end of each
month. After a week it is already
substantially long, and too long to
view entirely over SSH.
Is it possible to view an arbitrary
log file through Webmin to avoid
additional software being written or
installed? We are trying to avoid
both unnecessary SSH access and
additional software being installed.
David
The shell using tail -n
A would return the last ‘n’
lines of the log file. This in
turn could be conveniently
piped into tac to reverse the text,
effectively listing the latest entries at
the very top.
tail -n 200 /home/phone/cdr.current
| tac
Webmin, under the Options tab,
makes such commands easily
executable. Here is a possible
configuration option:
Definition = View CDR
Command = /usr/bin/tail -n 200 /
home/phone/cdr.current | /usr/bin/tac
Run in directory = /tmp
Run as user = Webmin User
Command outputs HTML = No
Maximum time to wait for command
= 5 seconds
Saving the configuration creates a
View CDR button, which should
provide the functionality. HH LXF
Submission advice
We are happy to answer all sorts of
Linux-related questions. If we don’t
know the answer, we’ll find out for
you! But in order for us to give you the
best service, it helps a lot if you read
the following submission advice.
● Please be sure to include any relevant
details of your system. “I can't get X to
work” doesn't really mean anything to us if
we don’t know things like what version of
X you are trying to run or what hardware
you are running on.
● Be specific about your problem. Things like
“it doesn't work” or “I get an error” aren’t
all that helpful. In what way does
something not work? What were you
expecting to happen? What does the error
message actually say?
● Please remember that the people who
write this magazine are NOT the authors
or developers of Linux, any particular
package or distro. Sometimes the people
responsible for software have more
information available on websites etc. Try
reading the documentation!
We will try to answer all questions. If we don't
answer yours specifically, you'll probably find
we've answered one just like it. We can't
really reply to all your questions.
23/9/05 10:31:01 am