NEWS

Newsdesk
■ Patents victory in Europe ■ Apple abandons PowerPC ■ Mandriva buys Lycoris ■ Gentoo founder
moves to Microsoft ■ Knoppix 4 announced ■ New UK body for FOSS testing ■ PS3 to run Linux ■

Europe says ‘NO’ to WHAT DOES

IT MEAN?

software patents
● No ‘common
position’ or
europewide unific
ation
of patents
● EU Commission
undertake not
to re-introduce sim
ilar directive
Landmark decision removes immediate threat of US-style software patents ● European Pa
tents Office to
continue as norm
al
On Wednesday July 6, The
European Parliament
rejected the idea of
patenting software. The Computer
Implemented Inventions (CII) directive
was rejected by an overwhelming
majority: 648 MEPS voted for
rejection, with just 14 votes in favour of
recommending the directive and 18
abstentions. Strasbourg became the
epicentre of a major open source
celebration as demonstrators who had
come from across the continent to
campaign against the (CII) directive
were left stunned, then jubilant when
they discovered the decision had gone
in their favour.
The ruling means an end (for now)
to the threat of US-style software
patents, which many people in the
free software world believe would have Feelings were running high in the
stifled innovation and led to greater lead up to the vote.
problems for developers.
The overwhelming rejection belies
the real machinations in the EU
Parliament though. Previously the
“I HOPE THIS GIVES PEOPLE FAITH IN THE
Parliament had recommended
amendments to the CII which
EUROPEAN DECISION MAKING PROCESS”
effectively ruled out patents for simple The decision effectively removes the
JUST A CLICK AWAY software features. The amendment short-term danger of Europe-wide
U.S. Patent 5960411 process would have meant more software patents, as the European
voting, with an unpredictable result for Commission had already said that
Software patents have been granted in information and uses this to enable an
the final legislation. Rather than there would be no plans to re-
the US already, and much has been instant buying function. Many people are
support a directive that could, by later introduce the directive if this attempt
made of how simplistic some of them incredulous that such a simple idea was
are. One of the best examples of this is granted a patent, but Amazon
amendment, rule out patents for was unsuccessful.
Amazon’s ‘One-click’ ordering system, has already used it in the courts against software across Europe, pro-patent The result was welcomed by all the
where the website stores customer rivals such as Barnes&Noble. parties also voted to reject the anti-patent campaigners, who ranged
directive, creating a huge majority. from concerned individuals to

6 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.news 6 11/7/05 3:46:33 pm

 NEWS

multinationals such as Red Hat and
Sun Microsystems.
Mark Webbink, Red Hat’s Deputy
General Counsel, was pleased that the
decision provided “a stable state for
Europe, and one that provides us with
the opportunity to now legally
challenge the practice of issuing
software patents by the European
Patent Office”.
Effective grassroots opposition to
the CII directive was largely co-
ordinated by groups such as the
Foundation for a Free Information
Infrastructure (www.ffii.org), whose
campaign included advocacy, physical
demonstrations and plenty of
information for those who wanted to
contact their MEPs. Jonas Maebe of
the FFII said after the landmark ruling:
“I hope this turn of events can give
people new faith in the European
decision-making process. I also hope
that it will encourage the Council and
Commission to model after the
European Parliament in terms of
transparency and the ability of
stakeholders to participate in the
decision-making process irrespective
of their size.”
Amid the celebrations, there was
still cause for concern. The rejection of
this directive leaves the status quo,
where the EPO and individual member
states make interpretations of existing
patent laws. And while the immediate
threat is gone, it seems unlikely that
UK government launches
open source testing lab
A chance to prove Linux to the public sector
The UK government has opened a
testing centre where local authorities
and other public service bodies can
try out open source and proprietary
solutions without the expense of a
full-scale trial.
The test laboratory is part of the
Open Source Academy, which is based
at the National Computing Centre in
Manchester and was launched in the
spring. “Many organisations have heard
about the benefits of open source but
are wary about implementing OS
The Open Source Laboratory is housed at the National Computing Centre
in Manchester, though remote access is also possible.
Miguel
de Icaza
Ximian’s co-founder
is hard at work on the
.NET migration project
Mono at Novell.
COMMENT
solutions due to technical and
compatibility considerations,” said
project manager Ed Downs. “The
Open Source Laboratory allows
Desktop
potential users to test configurations,
evaluate the benefits and build
confidence in OS technologies,
without risk.”
dreams
First to make use of the centre is Obviously, I often get
Cheshire County Council, which is asked about the Mono
evaluating a Linux/Citrix solution for project. I’m hoping this column
possible roll-out across the authority’s will give me a chance to give
desktops next year. everybody an update on the
project’s progress, and reflect on
the challenges that lie ahead, as
well as explain some of the
decisions we have made.
We started the Mono project
for two reasons: to have better
development tools for desktop
applications, and to assist
developers targeting applications
for the Windows platform to
move their apps easily to Linux.
On the desktop, our hope
that Mono would spur innovation
has been realised. Novell is using
it for its own products like
iFolder3, and we have seen a
spurt of new activity among

well-funded organisations
campaigning in favour of software
patents will simply let the matter rest.
Webbink is upbeat about the future
though: “the issues have now been
better laid on the table. I think it’s time
for a broader debate, one that involves
industry, individual developers and the
public interest... it ought to be a global
debate to discuss how this subject
matter is best protected to reward
invention and encourage development.”
PATENTS AND
COPYRIGHT
Traditionally there have been two
methods of giving legal protection to
intellectual property - patents and
copyright. Copyright is immediate and
inalienable from any created work, as
the expression of an idea.
Patents seek to provide protection
for the idea itself, and must be granted
by a relevant patent office. The area of
software is, in some places, covered
both by copyright and patents,
something that was never envisaged
when patents were first granted.
Mandriva + Lycoris = ??
Lycoris founder to help development of new desktop product
Mandriva’s shopping spree
continues with the purchase of US
Linux vendor Lycoris. This should boost
Mandriva’s desktop offerings, as
technology from Lycoris’s Desktop/LX
distro will be rolled into Mandriva’s
home and home office product lines.
Lycoris was founded in 2000 by
Joseph Cheek, a former Microsoft
staffer, as Redmond Linux. The name
change occurred in 2002 when the
company launched its updated
Desktop/LX product. Lycoris made an
attempt to put the familiar desktop
product on to PDAs and tablet PCs
(which may be part of Mandriva’s
thinking), but these two products were
discontinued in a shakeup last year.
Financial details are thin on the
ground, but Mandriva’s intention is
clear: to pull together the entry-level
Mandriva Discovery distribution and
Desktop/LX to create a class-leading
Linux desktop product. Cheek has
moved to Mandriva and will oversee
the integration of some core Lycoris
technologies into Discovery as well as
the documentation effort. As yet, it’s
not clear whether this will include the
AI2 package management and
Bitstream-licensed font management
system that were such a large part of
Desktop/LX’s appeal.
Lycoris’s 20,000 paying users will
be offered access to parts of the
MandrivaClub’s forums and continued
use of the Iris Software Gallery.
Linux desktop developers. Since
our release last year we’ve
increased the number of
architectures supported by Mono,
and upgraded our C# compiler.
Open source development
peaks during the summer as
contributors spend some of their
free time working on new ideas.
This summer will be particularly
exciting as various students are
being funded by Google to work
on Mono projects. In the autumn,
Novell will release a new version
of Mono that incorporates many
new libraries and will also see
the debut of the Mono Debugger.
Developers who may never
have thought it possible are
porting their software with little
or no effort. For me, this is a
win-win situation: independent
software vendors can use a
single code base, and our
beloved free software platform
gets a lot of software that would
not have been made available
otherwise.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 7

LXF70.news 7 11/7/05 3:46:36 pm

 NEWS
Can Apple and Intel kill
Linux on the desktop?
Apple’s migration to Intel signals a change in the industry status quo...
Steve Jobs’s recent decision
to standardise the next
generation of Apple
computers on Intel chips – creating so-
called ‘Mactel’ machines – is causing
consternation among industry
commentators, PPC vendors and
others in the Linux community. The
effect of this move on the nascent
Linux desktop is of particular concern;
though the extent of the danger is
somewhat contested.
Apple’s CEO made the
announcement at the annual Apple
Worldwide Developers Conference in
June, suggesting that Apple computers,
probably starting with the Mac Mini,
will be migrated to Intel CPUs within
the next 12 months. These will be
standard x86/X64 processors and will
most likely use commodity chipsets as
far as possible to reduce costs. It is
expected that Apple will try to come
up with some method to make it
difficult to run OS X on anything but
the hardware it sells – in the past the
high premiums paid for Apple’s
hardware have contributed significantly
to the company’s bottom line. Linux
and Windows, though, will be
installable on Apple’s x86 hardware,
making a triple-boot system a distinct
future prospect..
Squeezed out?
The fear is that with an alternative to
Microsoft on Intel chips, a great
swathe of potential Linux users will
switch their allegiance to the Mac.
Journalist Eugenia Loli-Queru said on
OSnews.com that when buyers
realised they could try out OS X
without necessarily losing access to
their favourite Windows applications,
they would see this as the natural
“LINUX AND WINDOWS
WILL BE INSTALLABLE ON
APPLE’S X86 HARDWARE.”
8 LXF70 SEPTEMBER 2005
LXF70.news 8
“intriguing” opportunities for the future.
Apple’s apparent change in strategy will
have huge implications for Linux.
alternative to Windows. And as the
computer industry tends to work on
the principle of dominant player and
underdog, Linux, Loli-Queru suggested,
would be relegated from its current
www.linuxformat.co.uk
PHOTO NEWSTEAM.CO.UK
underdog position to outsider status.
Jobs described the decision to
supplant IBM’s PowerPC processors
with Intel as Apple’s third “transition”,
after 68K to PowerPC and OS 9 to
OS X. The change may cause obvious
problems for Terra Soft and its Yellow
Dog Linux distribution, which is one of
the better-known and well-respected
distributions to run on PowerPC
architecture. However, chief executive
Kai Staats is putting a brave face on
things, saying the move offers
In a post on the Terra Soft website, he
wrote: “I want to make clear that
PowerPC is NOT dead. Apple is one of
many PowerPC manufacturers ... Terra
Soft has been working for several
years to broaden the opportunity for
Power Linux solutions. With Apple’s
announcement, those plans have been
accelerated ... Terra Soft and others
are rallying to fill the void created by
Apple’s departure and expand the
Power Architecture marketplace.”
Terra Soft is to continue for the
foreseeable future as an Apple reseller,
but Staats could not confirm whether
the company would offer branded
Intel boxes with Linux pre-installed.
Staats did, however, suggest that with
Apple concentrating on x86, there was
an opportunity for Y-HPC, the version
of Yellow Dog for high-performance
computing, to become the logical
upgrade for current Apple X-Serve
customers who may no longer have
the option of OS X PPC.
PHOTO COURTESY APPLE
11/7/05 3:46:38 pm

Gentoo founder finds gainful
employment at Microsoft
Did lack of funding push Robbins to Redmond?
Daniel Robbins has had Gentoo
fans sharpening their Star Wars
metaphors by going over to the dark
side and taking a job at Microsoft.
Robbins, MS executives claim, will
allow the company to “understand the
open source community” and its
development model.
Robbins faced a barrage of
invective on many community websites
for helping Microsoft in what many
claim is its continuing attempt to
“embrace, extend and extinguish” Linux.
But others were more sympathetic,
one user writing that “even open
source developers have to put food on
the table.”
Hans Reiser, author of the Reiser
filesystem, explained Robbins’s
predicament: “I think everyone should
understand that he had large credit-
card debts, that he tried everything he
could to make it work fiscally, and that
the community failed to provide the
finances that would make it work. His
Is this the face of GNU/Longhorn? No, it’s a Gentoo screenshot!
US Navy may go open source
Report highlights a confusion of operating systems
After a long, arduous trawl
through its network, the US Navy has
discovered that it’s running just about
every OS ever invented and has
decided, for the sake of efficiency, to
standardise its IT infrastructure.
The Navy began a six-month ‘asset
discovery program’ in January to
identify the 250,000 computer
systems it has installed throughout the
US mainland. It is expected that the
second, overseas section of this
LXF70.news 9
approach was technically superior to
the other distros in its fundamental
approach, and funding could have
cured any detail problems. It was the
right approach. He went broke, and we
should all be sad at this.”
Robbins built Gentoo – the
distribution and company – in 1998
around his innovative Portage software
management system, which uses
source code to build optimised
binaries for individual systems. He
resigned his position as chief architect
of the project in April last year, but
stayed on at the company to ensure
its transition to a non-profit making
organisation went smoothly.
Once Robbins had transferred all
Gentoo-related ‘intellectual property’
to the Gentoo Foundation, he quit
for good, and moved over to Microsoft
in May.
Reports that the move would bring
balance to the source code were
unconfirmed at time of writing..
process will reveal a similar figure,
making a total of 500,000 systems.
One of the surprises is that there
are a large number of Linux systems
patched into the network, but these
rely on a disparate collection of distros.
A spokesman said that it would be
wise to standardise on a particular
distribution to lower costs and make it
more efficient – a prospect that will
have the likes of Red Hat and Novell
itching to pitch for the business.
www.linuxformat.co.uk
RealBASIC 2005 offers Visual Basic-like
tools for cross-platform development.
RealBASIC for Linux is here
Cross-platform RAD tool arriving soon on a desktop near you
Real Software has begun the
testing cycle for the long-awaited
release of RealBASIC 2005 for Linux, a
cross-platform Visual Basic-like rapid
application development tool. Once
the beta period has ended, the
company will continue to make the full
standard edition of the software for
Linux available as a free download
when it’s released in August.
RealBASIC’s chief executive Geoff
Perlman believes the application
addresses two important issues within
the Linux market. “First is the lack of a
good Visual Basic-like development
environment and the second is the
lack of desktop software for Linux.”
In addition to the free (as in beer)
standard version, a Professional edition
will allow developers to write and
compile cross-platform applications
suitable for Linux, Windows and OS X.
The standard version will cross-
compile, but the application will only
★ FAVE Open Source Creativity,
August 20, Trinity Community and Arts
Centre, Bristol, England
Creative types are being invited to
scratch their open source itch at this
event in the centre of Bristol. The
one-day program includes
performances, installations and
workshops covering music production,
video art and 3D graphics. Cory
Doctorow, author of Down And Out In
The Magic Kingdom, will also be talking
about the Creative Commons and the
potential issues surrounding the
“European Broadcast Flag”.
www.fave.org.uk
NEWS
run for five minutes on a non-native
system, though users who upgrade to
the Professional package will be able
to edit and compile applications
created on the Standard edition.
A spokeswoman for Real Software
(not to be confused with
RealNetworks) said that the company
had registered a lot of interest in the
Linux version. “When we announced
our Linux strategy in July 2003, it was
the biggest announcement the
company had ever had, in terms of
press coverage and in terms of
sustainable product download rates,”
she said.
The spokeswoman added that there
were no plans to charge for the
Standard edition in the near future
and that there were no licence
restrictions to prevent software written
with the application being released
under the GPL.
www.realsoftware.com
★ UKUUG Linux Technical Conference,
August 4–7, Swansea University, Wales
The 2005 presentation line-up for this
annual event has finally been
confirmed. Speakers including Mark
Taylor from the Open Source
Consortium and Bdale Garbee from
Debian/HP and will cover a lot of
ground, from music recording,
production and distribution to Mono and
ASP.NET – with, obviously, a bias toward
the technical aspects of Linux and open
source software. Michael Meeks from
Novell may also offer a glimpse of the
company’s desktop plans.
www.ukuug.org/events/linux2005
LXF70 SEPTEMBER 2005 9
11/7/05 3:46:42 pm

■ Fast Rabbit Software has ported
two of its desktop applications to
Linux. Dolce Music Flash Cards helps
users learn to read music notation,
while Mathwords is a system aimed
at improving the mathematical word
skills of younger children. We'll be
reviewing both in LXF71.
■ Open Country, which provides
systems management solutions to
the likes of Cisco and Intel, has
become the latest member of the
Open Source Development Labs
(OSDL). Co-founder Laurent Gharda
said the group hoped to tackle the
“misperception” that Linux has a
higher total cost of ownership than
other OSs.
■ Electronic Arts has officially
released the Linux server software
for its Battlefield 2 software. The
game itself – an enormous tactical
military affair – is currently only
available in demo form.
■ The Fedora project, sweaty from
the effort of getting Fedora Core 4
out of the door, has published a road
map of ideas to be implemented in
Fedora Core 5. Top of the list is
reducing the bulk and providing a
distribution that comes on two discs;
options include having the core on a
single disc and then adding one for
KDE and one for Gnome, or splitting
the distribution into core and extras
and ensuring everything for general
use fits on a pair of CDs. Paul
Nasrat and Jeremy Katz have taken
on the task of overhauling Fedora’s
package management and creating
something to replace the system-
config-packages tool with recycled
bits of the Anaconda installer and
the Pup updater.
■ A period of consultation is set
to begin on the next version of
the General Public License
(GPL), the licence under which
Linux and lots of Linux software is
released. In a speech to attendees
at the LinuxWorld Conference in
New York in June, Eben Moglen,
the Free Software Foundation’s legal
adviser, said that the first public draft
of the new licence was imminent. A
major addition, Moglen said, were
specific clauses designed to deal
with “patent defence”.
LXF70.news Sec2:11
Knoppix 4 makes its
debut in Karlsruhe
It’s all work, work, work for Klaus Knopper’s team
Barely taking a breath after the
release of Knoppix 3.9 in early June,
developers of the Live CD rushed out
the release of version 4 in time for the
LinuxTag 2005 event in Germany
three weeks later.
Knoppix 4 is built on the
foundations of Debian Sarge but
updates a number of its applications,
including OpenOffice.org to 2.0. At the
time of writing the developers were
planning the usual CD release and a
new DVD version with over 5,000
applications. The DVD also contains
the text of two books: Knoppix
Kompakt by Rainer Hattenhauer and
Knowing Knoppix by Phil Jones.
Ahh… weaponry, death and
a little light destruction
Top-notch deathmatch game arrives on the net
After a wait in development hell
that surpasses even that of Debian
Sarge, the network-enabled first
person shoot-’em-up (FPS) Nexuiz
has been released. It took three and a
half years, but the development team,
led by Lee Vermeulen, have finally
made binaries for Linux, Windows and
Mac available at www.nexuiz.com.
Vermeulen says the inspiration
behind the game was a desire to take
things back to basics with a simple 3D
deathmatch played with “perfect
weapon balancing and fast-paced
action, keeping itself away from the
current trend of realistic shooters.” The
Run, hide, hide, fire
blindly, run, jump,
die. What a
wholesome workout.
www.linuxformat.co.uk
NEWS
David
Cartwright
David Cartwright is an IT
consultant who specialises
in providing Linux Systems
and solutions.
COMMENT
Options in this latest version include
an installation script to put a full
Debian system on a computer’s hard
disk, and support for the UnionFS,
A disc or 14
which places a persistent RAM disk on
the system, allowing any element of
the distribution to be altered.
too far…
I just went to download
the latest release of
Debian Linux. Oh dear, oh dear.
Back in the old days, I was a
Red Hat user – in fact I’ve used
everything from 6.2 right up
until they killed it off at release
9. In those days you just
downloaded two or three CD
images and you were sorted.
Since then, I’ve developed a
fondness for Debian – until I
went to get the new version,
that is.
OK, it’s packed with useful
stuff. But quite frankly, 14 CDs is
ever so slightly bonkers. Yes, 14.
game ships with 17 maps and quite a I could, of course, download the
number of servers are already hosting DVD images (there are only two
some punishing action. files, thanks to the extra capacity
Both the content and source of the of DVDs) but as it happens the
game have been released under the server I use for my Linux
GPL, which makes it possible for other installations doesn’t have a DVD
designers to take the basics, which are drive (one of the great things
built on the Darkplaces engine, and about Linux is that it works just
create a wide range of other games, great on my slightly geriatric
mods or just new maps for 1GHz PIII server).
deathmatch games. The game has I don’t care about
also been designed to run respectably bandwidth, I just can’t be
on low-end hardware, while retaining bothered to sit and burn 14 CDs.
features such as real-time lighting and Perhaps I’ll get the DVD version
bump-mapping for those with the and the network-install boot CD
grunt to use it. image and make the DVDs
available via a share from my
(gasp) Windows server.
Please Linux distributors,
let’s make this vast wad of CD
images an exception, not the
norm. I realise that you want to
cram as much useful stuff in
there as possible, so that people
like me can be confident of
having everything we could
possibly want, but bear in mind
that Microsoft and Apple
manage a single-CD installation
and even Solaris has a modest
media kit. Red Hat Enterprise 4
and Fedora manage four CDs –
and as far as I’m
concerned, that’s plenty.
LXF70 SEPTEMBER 2005 11
11/7/05 3:46:45 pm
 NEWS
PS3 hard disk
add-on to run Linux
A supercomputer wouldn’t be a supercomputer if it didn’t
Rumours abound that Sony’s
next-generation Playstation 3 will be
equipped with a hard disk option to
turn the games console (or media hub,
in media parlance) into a fully-
functional Linux computer.
Ken Kutaragi, president of Sony
Computer Entertainment, has given a
number of interviews in the aftermath
of May’s E3 show berating his rivals at
Microsoft and Nintendo for pushing
consoles as “toys”. In contrast, he claims,
the PS3 will not only be capable of
running a variety of operating systems
but the recently announced PS3 hard
drive attachment will turn the console
into a “Linux supercomputer”.
In an interview with Japan’s PC
Impress Watch, Kutaragi suggested
that the PS3’s Cell processor would
EMBEDDED LINUX NEWS
● Concept Distribution has begun shipping
one of the smallest general purpose PCs in
the world. The MiniPC is based on an AMD
Geode processor running at 400MHz,
features 256MB RAM and either a 40GB
or 80GB disk. There are also four USB
ports, front-mounted stereo connections,
100MB/s Ethernet and a display port
capable of driving a 1600x1200 monitor
at 85kHz. The basic model costs €347,
while a ‘thin client’ model starts at €265.
www.concept-distribution.com
● A curious device called the Unication
Magpie Linux PDA fetched up at this year’s
Computex show in Taipei. The device
features a physical and an onscreen
keyboard (for tablet-style access) and the
usual array of Wi-Fi and media player
facilities. It’s built on a Samsung RISC
12 LXF70 SEPTEMBER 2005
LXF70.news Sec1:12
be capable of running Windows, OS X
or Linux and made favourable noises
about Linspire’s Linux offering,
suggesting this – or something like it –
will form part of the final package.
He said: “We’re positioning the PS3
as a supercomputer, but people won’t
recognise it as a computer unless we
call it a computer, so we’re going to
run an OS on it. In fact, the Cell can
run multiple operating systems. In
order to run the OSs, we need a hard
disk. So in order to declare that the
PS3 is a computer, I think we’ll have
the hard disk preinstalled with Linux.”
On the launch of the Playstation 2,
Sony attempted to get the device
classified as a computer rather than a
games console to lower the import
duty entering the European Union.
processor and features a couple of
innovative features including integrated
VoIP support via 802.11b and a
short-range FM transmitter capable of
sending stored music to any nearby radio.
www.uni.com.tw/magpie_pda.html
● Matchbox Window Manager, a smart,
small-form window manager designed for
embedded applications, has been updated.
The latest release includes support for an
onscreen keyboard, better Gnome
integration and more remote options. The
project has been under the radar, but is
due to go mainstream as it has been
adopted as the platform of choice for
Nokia’s upcoming N770 Internet Tablet.
Nokia has been funding development of
the window manager for the last few years.
http://projects.o-hand.com/matchbox
www.linuxformat.co.uk
■ The new ‘open’ Office formats
from Microsoft have fallen foul
of the GNU project, which claims
the royalty-free licence under
which the formats are distributed
is incompatible with the GPL. This
may lead to problems implementing
document conversion in open source
applications. However, Dan Ravicher
of the Public Patent Foundation
suggested that, at present, there
is no reason for any developer to
fear the MS licence until the verity
of their patent claims has been
demonstrated.
■ Microsoft’s new Spaces blogging
service has also come under
intense scrutiny after it was revealed
that filters built in to the software
prevent users in China from typing
certain words. High on the list of
proscribed content are ‘freedom’,
‘democracy’, ‘human rights’, ‘Tibet’
and ‘Tiananmen’. Users attempting to
enter these words in a blog subject
line shown a warning that says, “This
topic contains forbidden words.
Please delete them”. Where do you
want to go today?
■ HP is blowing its own trumpet
after setting the “industry milestone”
of selling its one millionth Linux
server since getting into the business
in 1998. The company claims to have
shipped 45 per cent more systems
than its nearest rival and also chalked
up its 29th consecutive quarter at
the top of the Linux server pile. In
the first quarter of 2005, shipments
grew at 2.5 per cent more than the
industry average.
■ Meanwhile, HP’s Linux VP, Martin
Fink, has been fuelling rumours that
Linux will be ported to its NonStop
server platform. At a recent Red Hat
event, Fink said: “Maybe one day
you’ll actually see Red Hat Linux
running native on NonStop,” and then
he stopped.
■ Applicants to Google’s Summer
of Code project should know by
now whether the search giant has
approved their $4,500 grant to work
on one of a series of open source
projects. Participants will be working
on projects such as Mono, Apache,
Blender, Fedora and many others.
Two hundred lucky individuals will
be supported in their endeavours
throughout the summer.
■ Linux goes corporate! The BBC
has switched its weather-forecasting
presentation tools on to a Linux box.
Its new, controversial 3D map of
the UK, and the job of collating all
that data from satellites and wet
bits of seaweed are handled by a
bog-standard(ish) PC running Linux
with MySQL.
Richard
Cobbett
Florentine telescope
builder, physics pioneer
and church worrier. Or
was that Galileo?
COMMENT
Catching
the web
It’s great to see web
services finally taking off.
It’s a good thing for all of us.
Most feature a Linux base under
all the glitz, using that old
‘services and support’ model of
making money, and require no
experience or expertise from the
user. Just a few years after
everyone started talking about
how great such things would be,
we finally have them on our
screens. In a few years, we may
even have more than a handful
that are worth using.
The two big problems are cost
and interoperability, even on
most of the better services. Take
a group-calendaring program
like Trumba (www.trumba.com),
which costs $40 a year. That’s
great, until you realise every
member of your group has to
cough up the same to get any
real benefit out of it – and that
with limited support for iCal and
other standards, half of them
will still wind up doubling their
workload in the struggle to stay
organised. And this is considered
a really good application.
Support for RSS feeds, XML,
iCal and the rest is important in
desktop applications, but
essential on the net if users are
to get the best applications
around, rather than rely on all-in-
one jobs that may not offer the
exact service they need.
The PIM tool Backpack
(www.backpackit.com) has got
the model right. Because it
mixes its email reminders and
iCal feeds, the absence of a
dedicated calendar isn’t a
problem – although at $60 a
year for just 20 pages worth of
data, it’s clear that even the best
services still have some way to
go. Give it time. They say
it heals most things.
11/7/05 3:46:49 pm
 REVIEWS OpenLaszlo 3.0

WEB INTERFACE TOOLKIT

OpenLaszlo 3.0
Flash is everyone’s favourite proprietary format – now it’s everyone’s
favourite open format. Paul Hudson investigates…

KEY STUFF
Welcome Flash GUI-building
environment. See also: Ming.
■ SUPPLIER Laszlo Systems
■ WEB www.openlaszlo.org
■ PRICE Free under IBM CPL

Macromedia’s Flash format is

officially cool: it lets you
control exactly where things
go on a web page and works on every
platform under the sun. What’s more,
its use of vector graphics means that
its animation look pretty without
chewing up bandwidth.
Various projects have tried to
reverse-engineer Flash, but until now
open source has had no powerful
working environment equal to the
Flash MX 2004 Studio on Windows
and Mac OS X.
Before you get your hopes up,
OpenLaszlo is not a Flash editing
studio. But it is the next best thing: a
library of reusable page components
that helps you build Flash-based user
interfaces, without needing to sign any
odious licence agreements.
Placing it even higher on the
buzzwordtasticometer scale is the fact
that you design your interfaces using
XML, which OpenLaszlo uses to
generate the finished Flash movies.
Pre-fab web pages
Every web developer knows that Flash
is easy to use when you want to
animate little shapes or make simple
movies, but creating advanced user
interfaces sounds like a nightmare.
However, OpenLaszlo provides a
comprehensive set of GUI components
Beautiful interfaces galore await those who dare combine XML with Flash and JavaScript.
that come loaded with functionality. To
create a text box component, use the
<text> XML element; to create a
movable dialog, use <window>; to
create a button, use <button>; and so
on – you could almost guess your way
into OpenLaszlo programming.
Each component comes with a
certain amount of built-in
configurability. So, buttons can be
clicked on, text boxes can be typed
into, password fields type asterisks
rather than characters, and windows
The OpenLaszlo
documentation
is packed with
examples,
source code,
walkthroughs
and other help
– this is as good
as it gets in the
open-source
world.
can be moved. You can also customise
a component’s attributes: width and
height, colours, position and even its
actions. These actions are written in
JavaScript, using events that will be
familiar to any HTML programmer:
onClick(), onFocus() and
onMouseOver(). You can also use
basic variables inside your XML code –
something as simple as
width=”${immediateparent.
width}” gets properly translated into
the Flash movie.
From XML to Flash
All this magic works through Java. As
your page is requested, the XML gets
parsed and compiled by the
OpenLaszlo engine, which turns it into
a Flash movie. The reference
implementation for this is the servlet
container Apache Tomcat, but you
could feasibly use OpenLaszlo to
generate your Flash movies, then
move them to a static Apache server.
There’s little point though: the Java
compiler can be used to create
optimised static files that Tomcat will
serve up just as they are rather than
re-parse your XML, which makes it
just as fast as moving files elsewhere,
but without the inconvenience.
If you’re not impressed yet,
OpenLaszlo is also capable of
XML-RPC and SOAP. The first lets you
run remote procedure calls (there’s
your RPC) on other machines over
HTTP. It sounds complicated, but is
really just a way to call functions with
parameters and receive return values.
SOAP – the Simple Object Access
Protocol – transfers complex data
types to and from other servers, again
enabling you to link your Flash movie
to the wider world.
This support for XML goes even
deeper: OpenLaszlo can download and
parse an XML file from a local or
remote source, then allow you to run
XPath queries on it to present the
data. This kind of thing can be done
with normal Flash movies, but you
won’t see it often because it’s pretty
fiddly – unless you use OpenLaszlo.

20 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.rev_laszlo 20 7/7/05 5:20:21 pm


The selection of components on
offer is extremely generous. Along with
mundane things such as text boxes,
buttons, check boxes and lists, you
also get composite objects such as
date pickers, trees, tab groups, menus,
and data grids. Each of these can send
and receive data with the rest of the
Flash movie, but they can also be
used to send data over HTTP to a
form processor, like a HTML form.
Flash is known for its multimedia
capabilities, and OpenLaszlo doesn’t
fall short: you can place pictures,
movies, MP3s and even other Flash
movies in your XML, and control them
through JavaScript.
Double-take tasks
It is also possible create your own
code classes that add new
functionality or customisation –
OpenLaszlo lets you define a new
component as a class, then use that
class as a template that you can
stamp multiple times over a page. You
can then inherit from that class (and
inherit from the inherited class, and so
on), building up your own libraries of
complex objects. This seems hard to
believe at first – XML doesn’t have the
shiniest of reputations for taking on
tasks like this among veteran
programmers. But OpenLaszlo pulls it
off – and we’re impressed.
Of course, all this power makes it
a Herculean task to become fluent in
OpenLaszlo’s syntax, but to help you
along there’s a giant collection of
documentation covering every possible
aspect of system configuration and
development. Yes, you read that
correctly: this is a properly
documented open-source project that
guides you through installation,
learning to program, samples, a
The Eclipse OpenLaszlo plugin prompts you to fill in properties, complete
your XML statements and validate your code.
LXF70.rev_laszlo 21
REVIEWS OpenLaszlo 3.0
YOUR COMPONENTS TOOLKIT
We’re like children set loose in an eye candy store…
If a value can be only
true or false, checkboxes
are ideal. Users can add new Don’t write your own
Set the text, colour, and entries to combo boxes, calculations: the date
icon for each of them. or just use a fixed list. picker does it for you.
Sliders help your users
visualise numeric
quantities easily.
Use menus to give basic
program options, or add
Multi-line or single line? submenus for more
Password or normal You can make your complicated work.
text? Text edits let you list single-select or
choose. multi-select.
Tab sliders help you
store groups of Trees hold nested Windows are
Groupable items are information in limited information - they can independent dialogs:
perfect for radio group screen acreage. even hold GUI elements! move them, close them,
resize them and more.
components.
reference guide, an administration All this complexity might This product beats Java at its own
guide and more. This is surprising until overwhelm; were it not for the fact game (at least in the web browser
you realise that OpenLaszlo was open that IBM has created a plugin for space). It’s also completely free to use
sourced only recently – its company Eclipse that eases the task of under IBM’s Common Public License
continues to develop the software, but developing with OpenLaszlo. Not only (CPL: think GPL plus patent granting).
sells support and services as well. This is it designed to help you write valid OpenLaszlo is a huge boon to the
was a brave move, and we’re lucky to and well-formed XML, it also has an open-source community. When you
have them aboard. integrated GUI designer that lets you combine OpenLaszlo with Apache
drag and drop components rather Tomcat and Eclipse you get an open-
than hand-code the XML. If you’re source trio that packs a knockout
new to Eclipse, it might take a bit of punch. If you’re experimenting with
time to get accustomed to it, but web-based GUIs (perhaps you’ve read
anyone who uses Eclipse already will our feature on page 66), we urge you
be right at home. to try OpenLaszlo first – we promise
you won’t be disappointed. LXF
Welcome to Mongo
Once you’ve tried OpenLaszlo, it’s hard
to go back. You can spend weeks
LINUX FORMAT VERDICT
hand-crafting complex website GUIs in FEATURES 8/10
Java, but why bother? OpenLaszlo has PERFORMANCE 9/10
all the tools you need to create EASE OF USE 9/10
powerful and cross-platform user DOCUMENTATION 10/10
interfaces that look immaculate. It
takes a lot of effort to make Java look Looks hard at first, but worth any effort it
takes to learn. The future of web GUIs.
this good, and it certainly takes much
more skill to get the same power. And RATING 9/10
the docs provided here are unequalled
as far as we’re concerned.
www.linuxformat.co.uk LXF70 SEPTEMBER 2005 21
7/7/05 5:20:23 pm
 REVIEWS Xandros Business Desktop
LINUX DISTRIBUTION
Xandros Business
Desktop OS 3.0
It’s migration-friendly but it might be best to hold on to MS Office for a while, finds Andy Channelle.
KEY STUFF
Enterprise desktop distribution based
on Debian. See also: Novell Desktop,
SUSE Professional and Sun JDS.
■ DEVELOPER Xandros
■ WEB www.xandros.com
■ PRICE £75 (five-user pack £275)
A Linux distro aimed at office
workers should include an
office suite, right? Wrong, at
least according to the people
responsible for Xandros Business
Desktop OS 3.0. They have decided
not to install either of the included
productivity packages by default so, if
you want OpenOffice.org or StarOffice 7,
it has to be manually added through
the (admittedly excellent) package
manager Xandros Networks. In a
distribution aimed at normal office
users, this seems strange.
There are a few reasons why this
could be: a) space is at a premium on
modern distributions; b) sysadmins
may simply want to install MS Office
using the included CrossOver
Professional; or c) Xandros supposes
large-scale buyers will use its xDMS
product to set up bespoke installations
across many desktops.
These are sensible reasons, but it
may have been smarter to forgo, for
example, one of the two installed PIM
suites (of the two we would suggest
keeping Evolution for its Exchange and
GroupWise functionality), the games
and the large selection of media
players and use this space to include
OOo on the first CD. As it is, installing
an office suite manually is an extra job
iTunes support is improved thanks
to CrossOver 4.2.
22 LXF70 SEPTEMBER 2005
LXF70.rev_xandros 22
Xandros BD includes support for Exchange and Groupwise via Evolution.
on the to-do list that could have been
avoided. Rant over, we can turn to the
rest of this fine distribution.
The installation routine for Xandros
BD is slick, involving little more than a
few clicks and a couple of questions.
The installer can use available
partitions or resize and create new
ones, and the whole process should
be completed within half an hour (plus
ten minutes to install an office suite!).
The First Run wizard, which helps
the user to set time zones, mouse
button behaviour and so on, has been
retained from the previous versions
and still does a great job.
Niggly networking
Another familiar part of the installation
process is the small issue of setting up
a network. This process does work, but
requires further configuration once the
desktop has been started; even more
so in the case of wireless networking.
Once configured, both wireless and
wired networking work almost as
expected, but the Actiontec 54G
wireless card on our test machine
(recognised and usable by SUSE and
Ubuntu among others) didn’t work: we
had to dust down an older Wi-Fi card.
Moving from office to home to public
access point is easy, but it would be
good to see the option of saving
www.linuxformat.co.uk
settings for different situations so that
encryption, DNS and other
information could be associated with
particular locations.
Desktop tools
The desktop is based on KDE 3.3,
whose software and file management
apps are complemented by Xandros’s
own proprietary tools. These are
uniformly good, and the networking
options available through Xandros File
Manager are excellent. Mounting a
Samba share, for example, simply
involves highlighting it in the file
manager (it will have joined available
networks automatically) and selecting
the Mount option in the side panel.
Setting up a printer is just as easy.
New to this version are an upgrade
of CrossOver Office Professional to
version 4.2, an anti-virus package
developed by Xandros, support for
connections to MS Exchange Server
and Novell GroupWise, and
improvements to the VPN and firewall
configuration options. It also includes
the latest version of Skype’s VoIP
software and a free upgrade to
StarOffice 8 when it’s finally available.
The virus package is an interesting
proposition and will be handy if the
doom-mongers are right about a
future packed with Linux viruses.
Xandros BD’s combination of
user-friendly, cross-platform
applications such as Firefox, OOo and
the Outlook-alike Evolution and the
adherence to many Windows
conventions means it could probably
drop into the average user’s working
day with little disruption.
The ability to run legacy
applications on it if necessary (and
CrossOver runs the likes of Word, Excel,
Photoshop exceptionally well) means
that the process of migration can be
undertaken at quite a leisurely pace.
The high-quality networking tools
mean Xandros BD will be able to fit
right in to a heterogeneous
environment, providing seamless
access to servers across architectures.
The case of the missing office suite
is annoying, but it’s not drastic enough
to prevent us recommending you
consider this alongside the more
established business desktops. The
inclusion of CrossOver Office in
particular will be a potential deal-
maker for many businesses that want
to move forward but don’t want to
leave the past behind. LXF
“An odd choice of packages
and sloppy Wi-Fi support
makes Xandros more
trouble than it’s worth.
When you buy a distro, you
expect the very best – and
this isn’t it.”
LINUX FORMAT VERDICT
FEATURES 7/10
PERFORMANCE 6/10
EASE OF USE 7/10
VALUE FOR MONEY 7/10
This distribution could form the backbone
to either large or small-scale office
migration plans.
RATING 7/10
11/7/05 12:01:04 pm
 REVIEWS Studio to Go!

LIVE DISTRIBUTION FOR AUDIO

Studio to Go!
Graham Morrison looks at a Live distribution that aims to turn your beige box
into Abbey Road Studios. But is it more Beatles For Sale than Sergeant Pepper?

KEY STUFF
Suitable even for beginners and
flushed with an embarrassment of
Linux applications that aren’t all just
about audio. Alternatives include two
completely free Linux distributions
tailored to audio production: Apodio
and AGNULA.
■ DEVELOPER Fervent Software
■ WEB www.ferventsoftware.com
■ PRICE £49.99

If you followed any of our

recent audio tutorials, you’ll
no doubt be aware that
producing music using your Linux box
is becoming increasingly feasible.
There’s a wealth of audio applications,
from sequencers to multi-track
recorders, all tied together using one
of the most effective audio transport
layers available.
The downside to all this versatility,
as is often the case with Linux, is
complexity. And as we noodlers know
too well, complexity is the perfect
excuse for not getting around to
actually creating anything.
Fervent Software has a solution. It’s
called Studio.. to go! (hereafter called
STG to avoid that terrible punctuation),
and is a Live Linux CD stuffed full of
audio-centric software without any of
the associated hassle needed to set it
all up. For the uninitiated, this means
you don’t need to configure your own
kernel, set up audio drivers for your
hardware, divine the cryptic Jack
protocol or make animal sacrifices to
get virtual instruments to make a
sound. The idea with STG is that it will
just… go.
Based in England, Fervent
Software was co-founded by Chris
Cannam, one of the principle
FEATURES
AT A GLANCE
VST synthesizers and effects
work out of the box.
It really is a whole studio! Recording, composing, editing, synthesis and effects all off a Live CD.
developers of the Rosegarden audio
editor. Credentials like that are
important, and Fervent actively
promotes open software and Linux,
especially for use within education. In
fact, one of the main motivations for
creating STG was the difficulty some
people had in getting Rosegarden to
work satisfactorily. This isn’t the fault of
Rosegarden, but the result of using an
operating system tailored to the
desktop rather than audio production.
Su.. su.. studio
STG is based on the Knoppix live
distribution, but it’s been heavily
customised. You insert the CD into
your drive; reboot your PC and,
pending correct BIOS configuration, a
few short minutes later you’re
presented with a KDE 3.3.1 desktop. If
someone were to look over your
shoulder, they’d think you were a Linux
audio expert. There are a couple of
There are some
excellent tutorials
for getting the
most out of
the system.
prominent links to the Rosegarden
audio sequencer, and the toolbar at
the bottom is populated with large
CPU and memory meters, along with
a comprehensive audio mixer.
First impressions are impressive,
mainly thanks to the fully functional
audio system that has auto-configured
itself during the system boot. It’s low
latency, and this quick response time
between key press and sound is pretty
much vital for professional studios
today. Jack, the audio connectivity layer,
is already running, and everything
defaults to using it, even the venerable
XMMS player.
STG supports the same audio
hardware as a typical Linux distribution.
That means lots of Creative hardware,
most USB audio devices, and some
semi-professional products from the
likes of M-Audio and Hammerfall. It
also works very well with the onboard
laptop soundcard we used for testing.
LilyPond is
included for
excellent notation
rendering.
In common with other Live
distributions, your home directory
occupies volatile memory rather than
the security offered by a hard disk.
This is the whole point of a Live
distribution, but it needs special
attention when you’re creating and
working. Any problems with your
system, such as a power failure, and
your work will be lost.
Being based on Knoppix means
STG will install on to your hard drive
pretty painlessly, but Fervent Software
suggests using external storage, such
as a USB hard drive, for permanent
storage. USB devices appear on the
desktop automatically.
Run VST
Configuration is STG’s strongest card,
and most importantly after Jack, the
system comes complete with a
working VST system. This is for using
the vast library of high-quality effects

24 LXF70 SEPTEBER 2005 www.linuxformat.co.uk

LXF70.rev_studio 24 11/7/05 11:35:01 am

 REVIEWS Studio to Go!

you may have installed on a local copy

STUDIO TO GO’S
AUDIO SOFTWARE
of Windows. By default these are read
only, but it only takes a right-click on “THE BEST NATIVE LINUX
Editing: Rosegarden
Audacity
the mounted folder to enable writing,
depending of course on the usual SYNTHESIZERS AND
ReZound
Sweep
Linux compatibility issues with
Microsoft’s filesystems. The other way
EFFECTS ARE INSTALLED.”
Ardour of getting to VST plugins is through
Effects: Creox Rosegarden itself. editing process. Audacity is the all- Studio ..to go! works best, though,
Jamin powerful behemoth, ReZound is great as a distribution for a solid state
Time Machine Tuned desktop for monitoring and Sweep is quick and recording box. It is an effective way of
Jack-Rack Both the desktop and the K menu are effective. Each has been built to use experimenting with Linux audio, and
FreqTweak
tailored for running audio applications. Jack for its sound input and output, effects and synths work seamlessly out
QMidiArp
Other than the folders linked to your which means you don’t need to worry of the box, a refreshing change from
DestroyFX (VST)
own partitions and VST plugins, the about conflicts in your audio hardware. dealing with dependencies.
Synths: FluidSynth
QSynth desktop features a folder full of This works especially well for Audacity, As to the question of whether it’s
Timidity tutorials, and another that contains which has always caused problems worth paying for something you can
Hydrogen example projects. In the tutorials folder sharing audio channels with get mostly for free, then the answer
Hexter you’ll find a six stage, step-by-step Rosegarden. Ardour is also included has to be ‘yes’. Fervent has
Aeolus guide to working with STG. There are for the more serious multi-track unquestionably invested some serious
AMS also more specific guides to using recording tasks. time into developing this package, and
ZynAddSubFX Rosegarden and the music typesetter The best native Linux effects and it wasn’t too long ago that £50 would
Cosmogirl2k (VST)
LilyPond, which are the two synthesizers are pre-installed. buy you nothing but the most primitive
FearKiller (VST)
applications that take centre stage Highlights of the effects rack are Creox audio software. With STG, you can use
within the STG package. and FreqTweak. Creox is a bank of the time that you’d normally spend
and synthesizers available on Windows Rosegarden, reviewed in LXF65, is distortion, phase, flange, tremelo and configuring your system on composing
and is no small feat. VST is by far the commonly regarded as the best audio delay effects, connected in series for music, and for that alone, we think it’s
most common format for effects and sequencer for Linux users. It’s a some excellent electric guitar worth it. LXF
synthesizers, so for STG to provide a mature tool used for MIDI composition processing. FreqTweak, on the other
working framework straight out of the and multi-track audio recording, hand, is a crazy spectral delay where
box is excellent. Configuring this featuring matrix and score editors, you can change the phase and delay
manually would involve delicately using a synth and effects plugin of any specific audio frequency band.
balancing Windows emulation with architecture. If you’ve used Cubase on Results vary from muddy caverns to
Linux low-level audio design. another platform, you’ll feel right at something that sounds like converted
VST instruments and effects can home in the Rosegarden. helium nuclei escaping from the core
be used in two ways. Firstly, there is a LilyPond is a brilliant tool for of the sun.
folder on the desktop called VST rendering notation, and Rosegarden Among the synths, the standout is
Plugins. Opening this reveals a can export compatible files natively. As an incredible synthesized pipe organ
selection of Windows VST instruments LilyPond is a command-line utility, it’s called Aeolus. It’s not much to look at,
in their raw library format. The surprise a little disappointing that there’s no just a window with some buttons, but
is that when you click on one, it loads provision for simplifying its use, it sounds sublime. Each button actually
the plugin and makes the connection especially when you consider Fervent represents various organ stops, for
to the audio system automatically. That is aiming STG at the Linux novice. changing the timbre of the sound, and
is impressive. When you need to edit audio files it feels incredibly realistic, especially
STG also provides links to any directly, STG provides three of the most with the built-in reverb. Other synths
Windows partitions on your local common packages available for Linux: worthy of a mention are the mighty
system, which means you can navigate Audacity, ReZound and Sweep. They Alsa Modular Synth, Hexter for DX7
to the location of any VST instruments each cater for a different aspect of the sounds, and FearKiller, a really rich-
sounding VST instrument.

Well-rounded package
Monitoring is provided for by
meterbridge. Usually this is a rather
unfriendly application started from the LINUX FORMAT VERDICT
shell, but in STG you just need to FEATURES 8/10
select the meter you want from the K
PERFORMANCE 7/10
menu, and it connects itself
automatically to Rosegarden’s outputs. EASE OF USE 8/10
Audio applications are rounded off VALUE FOR MONEY 7/10
with K3b for burning a copy of your
While you may argue that all this software
masterpiece to disc. In fact, STG has is available for free, it’s going to take you
plenty of other Linux applications many hours to configure a system that’s as
refined as Studio.. to go!.
unrelated to music-making, and
wouldn’t work too badly as a normal RATING 8/10
Linux distribution. It’s only missing an
Aeolus sounds like the kind of organ Captain Nemo would play. office suite.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 25

LXF70.rev_studio 25 11/7/05 11:35:15 am

 REVIEWS Soul Ride

SPORTS GAME

Soul Ride
Graham Morrison straps a virtual board to his feet, puts on his goggles and
makes his cool face – but still can’t quite get into this disappointing effort.

KEY STUFF

Snowboarding game that emphasises

realistic slopes over impressive stunts.
Also try: Tux Racer.
■ PUBLISHER Linux Game
Publishing
■ DEVELOPER Slingshot
■ WEB www.linuxgamepublishing.com
■ PRICE £14.00

Snowboarding. It’s a strange

thing, but the feeling of
having both your feet
strapped to a plane of wood, sliding at
speed down a precipitous and snowy
slope is one of the best in the world.
It’s this peculiar mix of mountains,
snow and the force of gravity that
spurs thousands of people on to the
slopes each winter to experience the
exhilaration for themselves.
There have been plenty of
attempts to replicate this magic on the
PC, and console games like EA’s SSX 3
or even Microsoft’s Amped 2 have
genuinely captured some of the
excitement, but until now the only
option for Linux users has been Tux
Racer, and having wear a penguin
outfit on the slopes does reduce the
realism somewhat.
Wish you were here?
Let’s give a gnarly hand, then, to Soul
Ride, which was released several years
ago on Windows. The game’s engine
was then used to promote several real-
world ski and snowboard resorts:
Stratton, Jay Peak and Breckenridge.
Each resort offered a free download,
which contained the game engine and
a geographically correct, three-
dimensional model of the resort. The
YOU CAN GO YOUR OWN WAY
arder right
Slow your bo .
d the best path
down to fin
Commit to yo downhill.
poin
Bombing down the course in replay mode, the only way
we’ve found to get a third person view from the front.
idea was that you’d play the game, get
a feel for the terrain and splash out on
a holiday.
Things got interesting back in
January of 2003, when the game
engine was released under the GPL.
The project can still be found
languishing on sourceforge.net, but it
did result in a Linux version of the
client (as well as one for OS X).
Linux Game Publishing (thanks for
Hyperspace Delivery Boy and Majesty
Gold, chaps) has since resurrected
Soul Ride into a bona-fide Linux game,
including the three real resorts along
ur route and
t the board
with another two called Mammoth
and Tuckerman Ravine. It comes on a
single CD, works with kernel version
2.2.x or later and requires at least a
500MHz CPU and an OpenGL
graphics card. Basically, any machine
from the last few years should do.
Pure and simple
Soul Ride doesn’t feature quite the
same task-based events that you find
with other snowboarding titles. The
aim is still to get down the mountain in
the quickest time possible, performing
as many tricks as you can, but you’re
e Martens that pivot like an angler,
Hang on as th
ard pi cks up speed.
bo
not penalised for going too slowly, or
taking an unorthodox route.
This seems to be important to the
Soul Ride philosophy, where the game
engine focuses on the physical
interaction between rider, board and
snow rather than providing fireworks
and exaggerated terrain.
The graphics looked tired even
when the game was first released, and
four years on they look worse. Even in
the sparsely populated landscape of
Linux entertainment software, Soul
Ride’s graphics are poor.
The rider is particularly bad,
suffering from a severe lack of
polygons and a stylist that clearly has
no idea what the kidz are wearing at
Whistlers this season. There’s been no
effort made to illustrate the bindings
(the things that hold the border to the
board); instead it looks like he’s
wearing an oversize pair of Doc
hauling in the catch of a lifetime.
There is a first-person mode, but this
makes it extremely difficult to adjust
the angle of your boarder.

26 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.rev_soul 26 8/7/05 3:03:21 pm


“Think Tux Racer without
the herring. Graphical
glitches aside, this is a
cool game.”
ATHER
CHANGING WE
ea
It may seem lik
nice sunset…
change the
…but you can
weather…
a realistic
…to produce
white-out!
Happily, the game’s real-world
physics make up for any shortcomings
in the rendering of the rider – it really
does feel like real snowboarding.
As on a real piste, pointing the
board directly down the slope into the
fall-line results in sharp acceleration.
Control the speed of the board by
limiting the time you spend in the fall-
line, traversing the slope from one side
to another, letting the board slip under
your feet to change your angle of
attack. You’re balancing on either the
front or the back edge, and the hard
part comes when you reach the side
of the slope and need to turn. Let the
board slide into the fall line, and shift
your weight on to the opposite edge.
In fact, the actions for controlling
the boarder are similar to how you
ride in real life, and Soul Ride captures
this feeling very well. When the slope
becomes constrained, you’re forced to
spend more time in the fall-line, and
the board quickly starts to travel at an
uncomfortable speed. You then need
to try to stay in control until you’ve got
enough space to cut a wide enough
LXF70.rev_soul 27
curve into the slope and slow
yourself down.
The game physics handle all of this
with ease. The player’s centre of
gravity over the board has a profound
affect on how the board handles, and
you find yourself angling it to
accommodate the undulations in the
snow before the board touches down.
Control yourself!
Controlling the board takes a little
getting used to. You can use a joystick,
mouse or keyboard, but with the
keyboard you lose a lot of the subtle
degrees of control. For basic control,
you move either to the left or right to
change the angle of the board on
the slope – the equivalent of the
boarder shifting their weight from front
to back. You have to do this quite
gently, as sudden movements can end
in disaster. The left mouse button is
used to ollie (or ‘jump’ in standard
English). The boarder crouches while
the button is depressed, and springs
up when it’s released.
You need to time your jumps for
when the boarder reaches the crest of
a hill or top of a ramp, hopefully with
the result that he flies through the air.
You’re supposed to be able to perform
tricks at this point, but for this to be
effective you need to move your
controller in only one axis. For
example, to rotate left while in the air,
you would quickly move left as you
jump. The trouble is that isolating a
single axis with the mouse is
extremely difficult – with the result
that your rider is catapulted into
the air, limbs flailing in all
directions until being abruptly
dumped into a pile of snow.
In order to progress though
each of the predefined routes
down a mountain, you must
reach the bottom in one piece.
Speed isn’t important, but each
course has a par time. You don’t
need to be perfect, as your soul rider
has the equivalent of three lives – if
you wipe out prematurely, you get the
chance to rewind your descent to a
previous point, from which you can
make another attempt.
This removes some of the
frustration with the control system,
and adds a new dimension to the
game as you can attempt several
different lines across tricky terrain.
Unwanted freeze
The graphics for the mountains are
basic but passable. As we’ve
mentioned, they’re based on real data,
www.linuxformat.co.uk
The degree of
freedom Soul
amazing: you Ride gives you
can go down
any one of th
ese slopes.
and they do look realistic. The textures
are of a low resolution, as is the terrain
detail, but they provide enough visual
cues to be effective. The graphics are
fast, but we experienced some serious
issues where the game would freeze
at regular intervals. People
complained about this when the Linux
client was first developed, and it’s a
pity it’s still present.
For sound design, the developers
seem to have been inspired by a kettle.
The sound of the board bouncing
through the powder is like the noise
produced as the water reaches boiling
point, cooling slightly for slower
sections. This is augmented by a
steam-like spray that ejects from the
rear of the virtual board, presumably
to simulate the glistening cold powder
that your board throws into the air.
You can also take your boarder
off-piste using the Heli-Drop mode.
This presents you with a
three-dimensional view of the whole
mountain range, and you manually
select where you’d like your run to
start. Due to the slightly random
REVIEWS Soul Ride
is
nature of the snow at your
designated start point, it can be
difficult to get going, but a mode like
this strengthens Soul Ride’s credentials
in providing more of the feeling and
less of a game.
As you might expect, there’s an
excellent video-recorder mode where
you can save and replay your best
descents. You can pause and rewind
as well as slow the action down, and
the camera usually chooses a well-
placed third-person viewpoint for any
particularly neat tricks or jumps you
manage to pull off.
It’s a little difficult to be objective
when it comes to judging this release.
If you’re not a snowboarder, you’re
most likely to see it as another old
game that almost predates the
Playstation 2, with below-average
graphics, serious rendering issues
and a tendency to freeze.
But if you do love
snowboarding, you’ll like this
game. It really does evoke some
of the adrenaline-fuelled craziness
of the real thing. It’s the kind of
game you play on a hot summer’s
lunch hour, waiting for the nights to
draw in and the first sign of frost. LXF
LINUX FORMAT VERDICT
FEATURES 4/10
PERFORMANCE 6/10
EASE OF USE 6/10
VALUE FOR MONEY 7/10
Fun but frustrating. This could have been
so much better.
RATING 6/10
LXF70 SEPTEMBER 2005 27
8/7/05 3:03:38 pm
 REVIEWS Astaro Security Linux
FIREWALL
Astaro Security Linux
David Coulson looks at a Linux firewall you don’t have to be a command-line guru to use.
KEY STUFF
■ VERSION 6.0
■ DEVELOPER Astaro AG
■ WEB www.astaro.com
■ PRICE From $290 for ten users
Linux makes the perfect
firewall platform: it has a
small resource footprint and
a slew of features available rivalling
any high-end alternative. Unfortunately,
its inability to clear the user interface
barrier has hampered Linux as an
enterprise network device: it seems
that many people holding the IT purse
strings just don’t like the command
line. Recognising this, projects such as
SmoothWall and ClarkConnect have
succeeded in providing firewalls with a
“INSTALLING ASTARO IS AS EASY AS
A FEW CLICKS OF A MOUSE BUTTON.”
graphical interface that even beginners
can use. Here we’re taking a look at
another user-friendly product, Astaro
Security Linux, which is marketed as
the ultimate in Linux-based firewalls.
Installing Astaro is as easy as a few
clicks of a mouse button – all you
have to do is acknowledge the licence
agreement and set an IP address. Disk
partitioning is done automatically, but
there’s no option to perform an
installation on to a software RAID
Lots of feedback is available through Astaro. This executive report is
extremely comprehensive.
28 LXF70 SEPTEMBER 2005
LXF70.rev_ast 28
A fairly simple web interface keeps the guts of Astaro away from the end
user, but SSH is available for those who crave the command line.
filesystem. Astaro supports numerous
PATA, SATA and SCSI controllers
which will let you use RAID in a
roundabout sort of way, but it would
be nice to use the built-in RAID
capabilities of Linux.
Installation from CD takes five
minutes, and most of this is taken up
by reading the data from the disc.
Once installed, the system will reboot
and Astaro will boot up, allowing for
administration via a web interface.
www.linuxformat.co.uk
As web interfaces go, Astaro’s is
neat and reasonably user-friendly.
Each section of the Astaro interface,
from packet filter and network
interface configuration to intrusion
detection software (IDS), has a
consistent look and feel, making it
easy for anyone to find their way
around. At 1024x768, though, things
do get a little cramped.
Click to configure
Configuring Astaro is easy – once the
interfaces are set up, everything else
just falls into place. With support for
both bridging and VLANs, Astaro fits
in to almost any network environment
and can even be dropped in
transparently on a network to perform
packet filtering or intrusion detection
without having to reconfigure any
existing infrastructure. This gives
Astaro a huge advantage over devices
such as Cisco PIX, which is a
nightmare to get running smoothly.
Astaro supports IPSec and PPTP
virtual private networks, allowing for
site-to-site VPNs across a variety of
platforms and remote access for users
who need to use the internal network
securely. A variant of the NCP IPSec
client is available for use with Astaro,
allowing Windows systems to connect
quickly to our test firewall. Enabling
IPSec on Astaro and configuring a
client can be done in under half an
hour, so you don’t have to compromise
security for ease of use.
Reportage
Astaro’s reports are one of its
strongest features. There are plenty of
graphs available via the web interface;
and it also provides an executive
report, which rather than being an
overview is simply all reports and
graphs combined into one document.
Unfortunately, Astaro uses a basic
log browser for reviewing IDS incidents
and packet filter violations – this really
lets the reporting side down. If only
they had applied the same standards
of user-friendliness that they did to
the rest of the UI! All the standard
Linux logs are accessible via the web
interface, but unless you’re familiar
with Linux or are working directly with
Astaro to debug something, they are
unlikely to be much use.
Astaro is a great choice for a small
or medium-sized business – with
systems capable of gigabit speeds,
there’s an appliance available for
almost any network environment
We’ve certainly never managed to
build a system Astaro couldn’t use the
NICs on, even those with low-cost
motherboards with strange chipsets.
Astaro doesn’t do anything you
can’t do with distros such as Debian or
Fedora. However, the easy installation
and management Astaro offers could
tempt IT departments, and a 30-day
free trial available online means there’s
no excuse for not trying it out. LXF
LINUX FORMAT VERDICT
FEATURES 9/10
PERFORMANCE 8/10
EASE OF USE 7/10
VALUE FOR MONEY 8/10
A capable firewall and a great choice for
almost any user.
RATING 8/10
8/7/05 3:02:33 pm
 REVIEWS ICC 9.0
HIGH-PERFORMANCE COMPILER
Intel C/C++ Compiler 9.0
Is Intel still producing the fastest compilers for x86 Linux? Biagio Lucini is very keen to find out.
KEY STUFF
A C/C++ compiler for x86, EM64T
and AMD64 architectures that
produces optimised code. Also
consider: GCC/G++ and the Portland
Group’s C/C++ compiler.
■ DEVELOPER Intel
■ SUPPLIER Polyhedron Software
www.polyhedron.com
■ WEB www.intel.com/software/
products/compilers/clin
■ PRICE £232
or free for non-commercial use
As faithful readers may
know, we at Linux Format
take efficiency and
performance very seriously. We don’t
care about received wisdom or
common opinion: we like to test things
for ourselves. So when we heard that
a new release of the Intel C/C++
Compiler (ICC) had hit the shelves, we
couldn’t resist the chance to put its
performance to the test.
“FOR THE FIRST TIME, INTEL HAS
ADMITTED THAT ITS COMPILERS PRODUCE
GOOD CODE ON NON-INTEL SYSTEMS.”
SciMark2 benchmark results on the Pentium 4 system
running PCLinuxOS Preview 9.
1,000 ∆ GCC 4.0.1
∆ ICC 8.1
∆ ICC 9.0
800
600
MFLOPS
400
200
0 0
Composite FFT SOR
score
COMPUTATIONAL KERNELS
30 LXF70 SEPTEMBER 2005
LXF70.rev_icc 30
It was also a chance to test Intel’s
reputation as the best producer of
compilers around. Intel made an
unexpected impact in the field a few
years ago, and since then ICC has
been known as the fastest compiler
for Linux, but continuing development
among its competitors (particularly
GCC) has made it a fast-evolving
market and we were keen to see if this
new release keeps ICC ahead.
Let’s get going
ICC 9.0 weighs in at around 200MB of
tarballs, which contain the various
components of the compiler for all of
its supported architectures – IA32,
EM64T and IA64. Installation is
performed by executing install.sh. In
previous versions, this script only
worked on RPM-based systems, so it’s
a pleasant surprise to learn that this
time it no longer refuses to execute if
run by an unprivileged user or on a
non-RPM system. Although today’s
MC Sparse LU
matmult
www.linuxformat.co.uk
WHAT’S NEW IN 9.0?
■ Better support for 64-bit code.
■ Improved diagnostics and warnings.
■ Re-organised optimisation options, with
more of the aggressive optimisations
appearing at the -O2 level.
■ New compiler options for more finely-
grained control on the compiled code.
■ Software-based speculative
pre-computation, mainly useful for
threaded code.
■ Introduction of the GCC_EXEC_PREFIX
environment variable, for better
interoperability with the host system.
point-and-click user might not expect file. Once you’ve done that, it is just
this command-line installation, its matter of typing ‘accept’ on each
simplicity rivals the best graphical licence screen, and inserting a few
installers we’ve seen. paths. Installation takes about five
ICC is commercial software, and minutes, depending on how fast you
very early in the installation you’ll be read the licence agreements. What do
prompted to enter either a serial you mean, you don’t read them?
number or the path to a valid licence Intel is fully committed to providing
not just a compiler but a complete
development environment. Proof of
this is 9.0’s vastly improved integration
with the Eclipse IDE, which adds to the
already impressive number of
development tools (debugger, profiler,
test manager and so on) distributed
with the compiler. A custom version of
SciMark2 benchmark results on the AMD Athlon XP system
running Debian Woody.
1,000
∆ GCC 4.0.1
∆ ICC 8.1
∆ ICC 9.0
800
600
MFLOPS
400
200
Composite FFT SOR MC Sparse LU
score matmult
COMPUTATIONAL KERNELS
11/7/05 1:43:12 pm
 REVIEWS ICC 9.0

1,000
∆ GCG 4.0.1 32-bit SciMark2 benchmark results on the Opteron system
∆ GCC 4.0.1 64-bit running Red Hat Enterprise Linux AS 3.
∆ EKOPath 2.1 64-bit
∆ ICC 8.1 32-bit
∆ ICC 9.0 32-bit
∆ ICC 9.0 64-bit

800

600
MFLOPS

400

200

Composite score FFT SOR MC Sparse matmult LU

COMPUTATIONAL KERNELS

the whole Eclipse environment and a
Java virtual machine guaranteed to
work with it are optional extras.
The documentation is the best we
have seen from a compiler. It can get
very technical, but thanks to highly
visual tutorials doesn’t alienate
newcomers. These tutorials cover
everything from getting started with
Eclipse to the most obscure options on
the command line.
Rival architectures
For the first time, Intel has admitted
what was obvious to everyone who
has worked with its compilers: they
produce very good code on rival
AMD-branded 32 and 64-bit
processors too. In light of this, we
chose to test the compiler not just on
a Pentium but two other systems:
■ An Intel Pentium 4 with 768MB
RAM running a current Linux
KEY TO KERNELS
Mflops = Floating point operations per
second, in millions
FFT = Fast Fourier Transformation
SOR = Jacobi successive over-relaxation
MC = Monte Carlo integration
Sparse matmult = Sparse matrix vector
multiplications
LU = LU factorisation
Details at http://math.nist.gov/scimark2
distribution – the unstable Preview 9
version of PCLinuxOS.
■ An AMD Athlon XP 2800+ with
896MB RAM running Debian Woody.
■ A two-way AMD Opteron 244 with
4GB RAM running Red Hat Enterprise
Linux AS 3.
As an indication of the speed of
the generated code, we took the
SciMark2 benchmark (http://math.
nist.gov/scimark2/index.html) and
compared ICC 9.0 with its previous
incarnation (ICC 8.1) and GCC 4.0.1.
On the Opteron system we also
performed tests for both 64- and 32-
bit code and included the EKOPath 2.1
compiler in the comparison.
Serious analysis of these results is
beyond the scope of this review. But
some patterns clearly emerged:
■ On the Intel-branded processor, ICC
produces the fastest code, while on
AMD-based systems other compilers
take the top spot.
■ The 64-bit version of the compiler
is slower than the 32-bit one.
■ On the Opteron, ICC 9.0 produces
worse code than its predecessor, while
on x86 systems it produces code that
is, at most, only marginally (around
2%) faster than ICC 8.1.
ICC 9.0 provides a complete
development environment for Linux,
benefits from a simplified and elegant
installation procedure regardless of
your package manager preferences
and comes with extremely good
documentation. As far as speed is
concerned, there hasn’t been much
progress made since the last release –
which itself wasn’t much better than
7.0. ICC still produces very good code,
but the competition is stiff. LXF
LINUX FORMAT VERDICT
FEATURES 9/10
PERFORMANCE 7/10
EASE OF USE 8/10
VALUE FOR MONEY 8/10
Still a good performer, but the focus has
shifted from pure speed to providing a
complete development environment.
RATING 8/10

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 31

LXF70.rev_icc 31 11/7/05 1:43:14 pm


Classic Shell Scripting
Paul Hudson is never letting this book out of his sight. Ever.
BUYER INFO
■ AUTHORS Arnold Robbins and
Nelson H F Beebe
■ PUBLISHER O’Reilly
■ ISBN 0-596-00595-4
■ PRICE £24.95
■ PAGES 534
The centrepiece of the LXF Towers
reference library, the Magna Carta and
Domesday Book all rolled into one on
our shelves, is Unix Power Tools by
O’Reilly. Not that we’d ever lend out
our copy, but if we could recommend
only one book to someone who
wanted to master Unix, it would be
UPT. At least that used to be so, but
since Classic Shell Scripting found its
way on to our desks we’re not so sure.
We had better commission another
alarmed glass case for the library,
because this is officially our new
favourite text.
Classic Shell Scripting sounds like
one of those tomes that encourage
narcolepsy in the bookstore, but that
just shows we’re getting prejudiced in
Python Cookbook, 2nd Edition
Nick Veitch slithers into the kitchen.
BUYER INFO
■ AUTHORS David Ascher, Alex
Martelli and Anna Ravenscroft
■ PUBLISHER O’Reilly
■ ISBN 0-596-00797-3
■ PRICE £35.50
■ PAGES 810
What else would you expect on
the cover of a book about Python
than what looks like a cross between
a mouse, a rabbit and a kangaroo?
This is the second edition of the
Python Cookbook, and now spans
more than 800 pages of code
snippets. The 20 chapters cover a
diverse range of topics including
simple file and text handling, building
extensions to the language and using
exotic third-party libraries.
Each tip is composed of a problem,
a solution and a discussion, which
might include alternative suggestions
or common ‘gotchas’. The nuggets of
wisdom are thoughtfully submitted by
LXF70.rev_book 33
our old age. Instead, it’s a jazzy,
practical and fascinating book that
totally took us by surprise. Unlike most
programming languages, shell scripts
make extensive use of the built-in
Unix commands – grep, tr, xargs, find,
sort and more must be mastered to
make competent scripts. As a result,
this book is really a hands-on guide to
becoming a Unix guru. We can’t tell
everyday Python coders, so the style
and approach of the advice can vary
somewhat between authors.
The recipes themselves will suit a
varied palate. They range from the
simple (‘Processing every word in a
file’) and the useful (‘Pretty printing
the contents of a database’) to the
www.linuxformat.co.uk
whether this was the original goal of
the authors, but we don’t care
because this is a masterpiece we
refuse to part with.
The book is divided into sections so
random as to make the Bible look
organised, but when you read through
it fits together like a jigsaw. So you
could go from fairly simple topics like
pipelines or text processing to a
specific (‘Accented characters in
Tkinter Widgets’) and the downright
extreme! ( ‘Automating the py2exe
compilation of scripts into Windows
executables’). It’s such a random
collection of topics that we can’t say
for sure if you will get much mileage
out of it; then again, you can be pretty
REVIEWS Books
walkthrough with awk (still the bane of
many users) then on to files, processes
and even portability issues, but it all
meshes together nicely. As you learn
new topics, you are presented with
practical scripts to drill home the
learning. To top this all off, there’s a
marvellous (if small) chapter near the
end on how to write secure scripts,
which doles out good advice such as
checking that the file you’re working
with isn’t actually a symbolic link to
something unexpected.
If we could change one thing, it
would be to convert some of the
footnotes to general text, but only
because we prefer books to flow from
start to finish. These are nitpicks, we
know, but this is truly a top book.
LINUX FORMAT VERDICT
Failing a fourth edition of Unix Power Tools,
this is as good as it gets.
RATING 10/10
confident that if you have a Python
problem, however obscure, it will be
touched on somewhere here.
The only real problem with the
book is how long it can take you to
find what you want. The index isn’t
exactly expansive, and you may start
thinking about some sort of wheat.
chaff.sort() to root out the more
useful passages.
Still, through sheer volume this
book should have something you will
find useful, even if some parts of it are
merely descriptions of well-
documented module functions.
Oh, and it’s a springhaas on the
cover, by the way.
LINUX FORMAT VERDICT
The scattergun approach to tips, but some
are sure to hit your target.
RATING 7/10
LXF70 SEPTEMBER 2005 33
11/7/05 1:18:30 pm
 REVIEWS Books
Don’t Click on the Blue E
At last, an O’Reilly book that Paul Hudson’s dad can understand…
BUYER INFO
■ AUTHOR Scott Granneman
■ PUBLISHER O’Reilly
■ ISBN 0-596-00939-9
■ PRICE £12.95
■ PAGES 267
When it comes to matters of God,
my dad is spot on – he is a church
minister, after all. But when it comes to
the internet… well, suffice to say that
he’s still using AOL. That’s not to say
he’s beyond hope: he may still
consider Microsoft Internet Explorer to
be “the internet”, but at least he has a
Gmail account now.
So, while working to solve the
inevitable spyware issues on his
machine, I happened to mention to
him that he would have none of these
problems if he used Firefox.
Unsurprisingly, this was an alien term
to him, Microsoft’s monopoly tactics
having brainwashed him several years
ago. But while I could try to explain
why IE is bad and why Firefox is good,
Linux Desktop Hacks
Graham Morrison attempts to make his system as unstable as possible by trying all of these 100 hacks.
BUYER INFO
■ AUTHORS Nicholas Petreley and
Jono Bacon
■ PUBLISHER O’Reilly
■ ISBN 0-596-00911-9
■ PRICE £17.50
■ PAGES 318
Jono Bacon certainly keeps himself
busy. Not only does he write for us
and numerous other publications, he’s
got a real job with open-source
consultancy OpenAdvantage and is an
integral member of the LUG Radio
team. He’s also one of O’Reilly’s latest
signings, and along with Nicholas
Petreley has written the latest addition
to the Hacks series.
‘Desktop’, in this sense, is used to
describe the machine of a competent
Linux user, not the warm and fuzzy
world of point-and-click everything,
which might not be what some buyers
are after. While we’re picking nits out
of the title, a ‘hack’ is usually
understood to be an unusual, clever
34 LXF70 SEPTEMBER 2005
LXF70.rev_book 34
there is no longer any need: this book
does that for me, and legions of other
‘geeks next door’ around the world.
Yes, this is a whole book on Firefox,
taking you from installation and
configuration to extensions and
security. It might seem like a hard task
to fill 250 pages talking about a web
browser but the laid back style adds to
the weight, as does the vast number
of screenshots designed to hand-hold
people through basic tasks. This is,
and concise solution to a tricky
problem. O’Reilly, on the other hand,
seems to believe it’s more of a step-
by-step overview with a little
background information thrown in for
good measure. That’s the case with
the other books in the Hacks series,
and Linux Desktop Hacks is no different.
What we get here are 100 tutorial-
style vignettes on such subjects as
setting up an old computer to use as
a firewall and adding drop shadows to
your desktop. Some entries are the
brief, smart ideas you may expect,
www.linuxformat.co.uk
after all, a book designed to cope with
newbies taking the first steps in open
source. Although there is some
discussion of alternative browsers, it’s
hidden away towards the end of the
book and uses Firefox as the yardstick
against which other browsers are
measured. This shows readers that
Firefox isn’t the only alternative, but
don’t expect to hear any bad words
about it – as far as Granneman is
concerned, Firefox is nigh-on perfect.
others are longer, more rambling
affairs that often break into general
discussions of the techniques involved.
For example, hack number 12 is how
to take a screenshot from the
command line, whereas hack 37 is a
detailed look at the KDE desktop
enhancement tool, SuperKaramba.
Consequently, the quality of the
hacks varies from the blindly obvious
to the genuinely interesting, and
there’s something for everyone with a
little Linux experience behind them.
It’s written to be one of those books
It’s this devotion to a single
solution that may actually prove the
biggest negative of the book. Rather
than saying, “IE is good, Firefox is
better”, the author takes great pains to
put down Microsoft and Internet
Explorer at every available opportunity.
This gives the book a childish attitude
in places that isn’t doing it any favours;
it just feels like FUD in reverse. On the
surface, the other big downside is that
the book demonstrates Firefox running
only on Windows, but we can forgive
that. Switching away from IE is the
beginning of the end of Microsoft’s
monopoly for many, and anything that
speeds that up gets our vote.
LINUX FORMAT VERDICT
Information-packed and accessible enough
for your family and friends, but it may
leave them with more questions than they
started with.
RATING 6/10
you pick up when you’ve got a spare
30 seconds, and we often found
ourselves flicking through the pages
for some instant gratification.
If there’s anything to criticise, it’s
that Linux Desktop Hacks feels more
like a disparate collection of howtos
than an insightful group of cutting-
edge, system-refining hacks. This
means that it may disappoint those
who expect new ideas, but it does
give the book broad appeal. In fact,
this probably works best for the
intrepid desktop user who’s itching to
start messing around with their system,
but doesn’t know where to start, and
this book will provide them with 100
starting points.
LINUX FORMAT VERDICT
An enjoyable read that’s easy to pick up in
a spare moment. Great to leave by the
side of your PC.
RATING 7/10
11/7/05 1:18:34 pm
 MySQL in a Nutshell
Nutshell books are terse, but always packed with
info. And that’s just the way Paul Hudson likes it…
BUYER INFO
■ AUTHOR Russell J T Dyer
■ PUBLISHER O’Reilly
■ ISBN 0-596-00789-2
■ PRICE £28.50
■ PAGES 321
Once you’ve spent weeks
mastering a system, it’s time to put
your learning guides aside and reach
for the relevant O’Reilly Nutshell book.
The series is well established, and in
fact this new edition seems eerily
overdue: surely MySQL in a Nutshell
was out years ago? Not so. This is the
first edition of this Nutshell guide, but
surprisingly it’s arranged sensibly and
devoid of the usual typographical
errors – they managed to get this one
right first time, thus depriving us of
comedy review fodder.
The contents are comprehensive
in their coverage of MySQL. At only
just over 300 pages, this depth of
detail is possible by keeping examples
to the point: you don’t get waffle, you
don’t get excess code blocks and you
certainly don’t get picture after
gratuitous picture to demonstrate a
single point. This is the traditional
Nutshell format and needs to come
with the traditional Nutshell disclaimer:
if you don’t already know something
of the topic the book covers, this isn’t
the place to learn unless you have a
lot of patience or experience in
something similar.
The largest section of the book is
that discussing SQL statements. Each
type of statement has a prototype
showing the various options you can
LXF70.rev_book 35
use with it, plus text describing what it
does, with sample output. These
statements aren’t grouped linear-style
to help you learn to use SQL, as that
would be un-Nutshellish. Instead, you
get them in an alphabetical list, with a
quick reference at the front that
groups them by type. Elsewhere, many
of the titles simply carry the name of
the function they cover, again justified
on the grounds of alphabetical sorting.
Given that O’Reilly books usually
come with high-quality indexes to find
such function names, it might have
been better to throw alphabetising to
the wind and provide descriptive titles,
but it’s horses for courses. We
certainly feel that, in a reference book
at least, making things easy to find is
pretty important.
Towards the back of the book is an
excellent reference guide to
programming with MySQL, covering
the Perl, PHP and C APIs. Here the
code samples come thick and fast and
we can’t get enough of it – each
function gets an outing in the
appropriate language, always neatly
written so that you can copy and paste
the code in to your own work without
too much fuss. We’ve waited too long
for a MySQL Nutshell book, and this
book satisfies completely. LXF
LINUX FORMAT VERDICT
This is an average Nutshell book, which
places it well ahead of the mass of
competitors. Not for newbies.
RATING 8/10
11/7/05 1:18:38 pm
 ROUNDUP Text editors
Roundup
Every month we compare tons of software, so you don’t have to!
TEXT EDITORS
Tom Wilkinson tests the widely-held theory on LXF
that you can’t find a decent editor these days…
At some point in your Linux-
using life you will probably
need a text editor – whether
to write code or simply to edit an
application’s config file. There are
countless editors on the market, all
behaving differently and all with
different strengths and weaknesses.
On test this month we have eight
of the most popular text editors.
Between them, they offer both text-
based and graphical interfaces (the
latter are known as graphical editors),
and a variety of features, some of the
most valuable being spellchecking,
syntax highlighting and automatic
code indentation.
Text editors have a different
purpose from word processors,
although the two often overlap in
terms of tools and functionality. While
a word processor is designed for the
input and formatting of text
documents, a text editor will allow any
form of text file to be edited, and is
36 LXF70 SEPTEMBER 2005
LXF70.round 36
instead focused on the manipulation
of the text displayed. Formatting and
output display is of no concern, and is
left to the web browser, LaTeX
renderer or some other app that turns
the raw text into a finished document.
Many editors have features and
plugins that make them especially
useful for programmers. Syntax
highlighting and automatic indentation
are among the most important tools in
the programmer’s editor. These allow
the user to see at a glance whether
the code they’re entering is broadly
valid, and that all their code blocks are
self-contained with no glaring errors.
With syntax highlighting, each
keyword is highlighted are variable
names and data are in a different
colour. There may also be distinct
colours for different variable types or
function calls. The automatic
indentation makes it easier to see
each code block, as sub-blocks are
indented further, aiding readability. Of
www.linuxformat.co.uk
OUR SELECTION
AT A GLANCE
CoolEdit ...................................39
gedit .......................................... 40
GNU Emacs ...........................39
Kate ............................................37
nano ...........................................37
NEdit ..........................................38
Minimum Profit................. 40
Vim ..............................................38
course, this is only helpful if you agree will run in a text-only environment is
with the program on what a sensible extremely useful.
indentation policy is – there are Finally, we’ll look at how easy an
almost as many styles of indentation as editor is to learn and to use. A very
there are programmers. A good editor basic editor is easy to learn but hard
will let you specify your own style. to use for anything but the basics; a
Another important consideration is more advanced editor may make it
whether the editor can run in a text easy to perform complex changes but
console. In the event of serious system alienate new users with a complex or
problems a graphical environment unintuitive interface – and if you can’t
won’t be available – so an editor that use it, it doesn’t matter how good it is.
EDITING WITHOUT AN EDITOR
There are ways to edit text files without using a text editor, of course. Any
command-line tool that allows the manipulation of a file can be used as an editor,
including cat, which is usually used for outputting the contents of a file into another
program’s expected input, and dd, which is more often used in the creation and
copying of disk images. The tools sed and grep can be used in combination with
these to perform any function you might wish for. While you’re not likely to use this
approach very often, it could prove invaluable in the event of a binary-destroying
disk problem.
If you’ve really got a lot of time in your hands, you could write your own tool
suited to the task using whatever programming language you want. It would
probably be quicker to use something else, though.
11/7/05 11:02:28 am

nano
A command-line editor included with many distros.
■ VERSION 1.2.4 ■ WEB www.nano-editor.org ■ PRICE Free under GPL
Nano, released by the Free
Software Foundation, is intended as a
free (as in speech) replacement for
the popular Pico editor, which was
developed by the University of
Washington. Its interface will be
familiar to anyone who has used the
text-mode email program Pine, on
which Pico is based. It was released
because although the source code for
Pico is freely available, it isn’t classified
as an open-source program – anyone
building an application on Pico must
first ask the University of Washington
for permission. This is of course less
free than the GPL, and pressure from
distributors such as Debian led to
nano’s creation.
Nano is one of the most basic text
editors available – it has no support
for features that many programmers
find useful, such as bracket matching
or syntax highlighting. As an editor for
plain text, though, it’s well up to the
Kate
Editor with project management functionality.
■ VERSION 3.3.2 ■ WEB http://kate.kde.org ■ PRICE Free under GPL/LGPL
Kate is one of two editors that are
integrated into the KDE desktop. The
other, KWrite, provides a basic text
editing option. Kate is a much more
fully featured editor, aimed more at
programmers than general users, and
supports plugins for extending its
functionality even further.
As it’s part of KDE, Kate is a
graphical-only editor. This allows Kate’s
developers to concentrate on its
strengths and provide a solid editor
with enough functionality and
extensibility to keep power users
happy, while at the same time being
approachable enough not to frighten
new users.
In common with most of the
programmer’s editors available, Kate
allows the use of regular expressions
when searching and offers the usual
function pair of syntax highlighting and
auto indentation. Unusually, Kate also
supports code folding. This enables it
LXF70.round 37
task. The basic keystrokes required for
the most common functions are
explained at the bottom of the screen,
and it’s quite straightforward to edit a
configuration file without getting lost in
modes, or millions of menu options.
This program is aimed at a specific
market and fulfils its aims well – new
users who aren’t yet ready to get to
grips with Vi or Emacs can still edit
files on a text console.
Although it lacks most of the other
features of a text editor, nano includes
basic spellchecking functionality.
Unfortunately missing, though, is the
ability to add words to the dictionary.
This limits the usefulness of the
spellchecker, but a limited spellchecker
is better than none at all.
Basic functions are accessed by
holding down Ctrl and pressing
another key. This is about the most
straightforward way there is of using
the different options in a text-only
to perform tasks such as collapsing
functions within a program so that only
the first and last lines are visible. The
code can be restored to its full size
when required. It’s an innovative and
useful approach to coding, as it allows
the programmer to look at an overview
of the code they’re working on and
zoom in on any particular part that
may need attention.
As well as the standard
programmer’s stuff, Kate provides
some functionality for project
management. A project can be
defined as a collection of files, and
opening the project will open each of
the files in its own tab. This allows for
the editing of multiple dependent files
at once – for example, a header file,
C source code and the relevant
makefile. When you’re working with
larger projects, such as applications,
this aspect of the Kate becomes
extremely useful.
www.linuxformat.co.uk
ROUNDUP Text editors
Basic help is
provided in
the form of
keystroke
hints at the
bottom of
the screen.
editor, though it does limit what can
be done. We suspect this won’t be a finding whatever introduced the error
problem for most people, though – as can be time-consuming. However, it is
users expand their abilities they will possible to remove this ‘feature’ with a
move on to something more powerful. simple command-line argument.
Only one thing struck us as
particularly irritating about this editor: LINUX FORMAT VERDICT
line breaks are inserted automatically
Easy to use, but with little advanced
by default. This isn’t necessarily functionality nano has limited appeal.
desired functionality – imagine writing
a long block of code that didn’t RATING 5/10
compile because the editor had added
a line break in the wrong place – and
Kate provides
useful tools such
as project
management and
collapsible code
blocks.
Bookmarks can be set within the
code, enabling the user to jump to a browser. This is most useful when
given file location with a quick menu working with new APIs – highlight the
selection. Another useful feature function you want to know more about,
involves adding a terminal window into and you’ll likely have manuals,
one of the editing tabs, rather than a discussions and hints at your fingertips.
text file. This means that it’s quicker to
switch out to the shell (for example to
LINUX FORMAT VERDICT
try to compile the program) than it
might be otherwise. Project management features build on
Kate’s solid base. Overall it’s a useful tool
One of Kate’s more innovative but won’t run in a text console.
features allows you to highlight
anything in your text and perform a RATING 7/10
Google search for it, displaying the
results in Konqueror, KDE’s web
LXF70 SEPTEMBER 2005 37
11/7/05 11:02:35 am
 ROUNDUP Text editors
NEdit
Motif-based editor with plenty going for it.
■ VERSION 5.5 ■ WEB www.nedit.org ■ PRICE Free under GPL
Begun in 1991, NEdit is one of the
older graphical editors on test this
month. It predates the dominance of
the KDE and Gnome desktops by
several years, so was instead built with
the multi-platform toolkit Motif. Until
recently, Motif was a closed-source
product, which could cause problems
with obtaining and building it (an open-
source clone, LessTif, was available, but
it wasn’t always sufficiently compatible).
The release of Open Motif in 2000 has
meant that obtaining and building
NEdit is much easier.
Although it’s been around for a
while, the NEdit development team
continue to add innovations and keep
their editor up to date. Recent
releases have seen the introduction of
tabbed editing, which allows you to
have more than one file open at once,
and UTF support, meaning that any
character in any language can be
entered into the text file.
Vim
Mode-based but powerful and near ubiquitous.
■ VERSION 6.3 ■ WEB www.vim.org ■ PRICE Free under Vim License
Vim is one of the most maligned
text editors available – but quite
probably the most popular too.
It has a reputation for being
impossible for new users to get to
grips with – and this is probably
deserved, because unlike every other
editor on test here it uses what is
known as a modal interface, which is a
throwback to early text editors that ran
on very basic text-only consoles that
lacked the ability to scroll in more
than one direction. Vim’s interface is
based on Vi (indeed, the name is
short for Vi improved), which in turn
was based on an editor called ed. If
you’ve tried Vim and found it
incomprehensible we’d urge to give ed
a go – it’s still installed in most major
distributions and will show you what
things were like 20 years ago.
Vim’s two modes are often jokingly
referred to as ‘The one that beeps,
and the one that doesn’t’. The latter of
38 LXF70 SEPTEMBER 2005
LXF70.round 38
NEdit comes from the top drawer –
it’s extremely easy to use even for a
newcomer, and has a large feature set
for more experienced users, including
syntax highlighting in nearly 30
languages, auto-indentation and a
number of options for text wrapping.
One particularly useful option shows
line numbers, which makes it much
easier to see where a spurious
carriage return has been added to the
file, and aids greatly in debugging.
The program also makes it
incredibly easy to run any given shell
command on the text without saving it
out. Some predefined menu options
for this include spellchecking using
ispell, word counts using wc, and make,
which enables the program you’re
editing to be rebuilt and tested without
leaving the editor environment – a
helpful function when building
programs in compiled languages such
as C or Java.
these is known as insertion mode, and
is used to enter text into the file. This
is quite straightforward, and supports
the usual features one would expect –
auto-indentation, syntax highlighting,
bracket matching and so on. Any or all
of these, plus a whole bucket-load of
other functions, can be enabled or
disabled in the mode that doesn’t
beep, the command mode.
The command mode can be
bewildering for new users – it’s often
unclear what needs to be typed to
achieve the outcome you want, despite
the excellent vimtutor.
Press I (capital i) to enter text and
Esc to return to command mode,
where the really interesting stuff takes
place. Commands range from the
straightforward (:q to quit, for example)
to the complicated – you can tell it to
emulate Vi for example, or to disable
or enable automatic indentation or
syntax highlighting.
www.linuxformat.co.uk
NEdit provides
a clean but
functional
interface, and
is easy to get
started with.
Searching, as you’d expect from an
editor aimed at a generic ‘everyuser’, to do, and does it well. Sadly, too few
supports but does not require regular people will come across this excellent
expressions – this is probably the best editor, simply because of its reliance
compromise between new and more on a toolkit that is no longer in fashion.
experienced users. We also like
position marking. This allows you to LINUX FORMAT VERDICT
mark positions in the text that you can
jump back to with a couple of An excellent editor that fuses
sophistication with ease of use. Its limited
keystrokes, and is invaluable if you’re availability on the desktop is a weakness.
switching between functions and
declarations earlier in the file. RATING 8/10
Despite its slightly old-fashioned
looks, NEdit does the job it’s intended
Vim’s modal interface can be
hard to get to grips with at first.
Searching is done by means of be done in Vim, while lesser editors
regular expressions, which can then might have to resort to external tools
be expanded to replace text with far such as perl or sed.
more precision and variation than a
standard search and replace. An
LINUX FORMAT VERDICT
example would be replacing all
instances of the word ‘Linux’ that An extremely powerful but complex
application that is included in most distros.
precede the word ‘Format’ with the
abbreviation ‘PC’. Even more powerful RATING 9/10
changes than this are possible – if a
regular expression can match it, it can
11/7/05 11:02:41 am

GNU Emacs
Allows extensions to be written in a variant of Lisp.
■ VERSION 21.2.1 ■ WEB www.gnu.org/software/emacs
■ PRICE Free under GPL
Loved and loathed in equal
measure, Emacs is probably more
powerful than any other text editor on
test here. Unlike its chief rival Vim,
Emacs has only one input mode.
Additional functionality is achieved by
pressing key combinations. Indeed,
many applications strive to provide
Emacs-compatible key bindings, so
you’ll probably feel at home with
this text editor even if you’ve never
used it before.
Emacs does have some
disadvantages – chief among them is
that, because of its extensibility, it
takes up a large amount of memory –
at least in relation to other text editors.
Extensions to the program are written
in Lisp, a language usually used to
write artificial intelligence routines
within larger programs, and the Lisp
interpreter takes up much of the
space used by the program. This isn’t
CoolEdit
Python-extensible graphical editor with debugger.
■ VERSION 3.17 ■ WEB ftp://ftp.ibiblio.org/pub/Linux/apps/editors/X/cooledit
■ PRICE Free under GPL
CoolEdit has a very different feel
from the other editors on test this
month, despite being obviously a very
similar program. This is probably
because, rather than use an existing
windowing toolkit, the authors have
chosen to implement their own. The
program was released in 1996, so it
predates the ubiquity of the KDE and
Gnome toolkits – viewed in this light,
the decision to use bespoke toolkits
makes a little more sense.
CoolEdit provides all the
functionality needed in a text editor –
including our benchmarks of syntax
highlighting, automatic indentation and
spellchecking. But you can expand on
the program’s abilities through scripting.
These features are more accessible
than they are in Emacs, though – the
language of choice here is Python
rather than Lisp, so getting to grips
with writing your own extensions isn’t
LXF70.round 39
a problem with a modern PC, but
older machines might struggle at
times. It’s for this reason that Emacs is
sometimes said to stand for Eighty
Megabytes And Continues Swapping
(actually it stands for Editor Macros).
Of course, this extensibility is the
reason why the editor is so popular.
Emacs can be used as a graphical
web browser and an email program,
and there are countless games
available, including Tetris, a version of
Eliza (an eighties-style text adventure
game) and many more. It’s also
possible to use it for calendaring and
managing appointments.
Some people use Emacs almost
exclusively as a whole operating
environment, so they’ve invented the
alternative name Emacs Makes A
Computer Simple.
As a text editor, Emacs rarely falls
short. Its primary function – despite all
as time consuming. Other scripting
languages are supported, but these
don’t integrate as comfortably with
the editor.
One feature that is unique to
CoolEdit among the editors in our
Roundup is an in-line debugger. While
some of the other editors might push
the boat out and include an external
copy of GDB, CoolEdit has a complete
debugger built in – press a key and
your program will be compiled and run
through the debugger. This is
invaluable for finding code faults
quickly and easily, and gives you an
efficient way to fix a program and test
it again. It’s also a testament to
CoolEdit’s expandability – while Emacs
has numerous plugins and feature
expansions, few are as useful as this.
Curiously, CoolEdit features an
option to send an email, which uses
the current text as a message body.
www.linuxformat.co.uk
ROUNDUP Text editors
Highly
extensible,
Emacs is less
a text editor,
more a way
of life.
the extensions that have been bolted
on later – is most definitely software program, though, you’ll find you’re
development, and in this it excels, using these less – as everything can
including menu options for compiling be invoked as a keystroke, there’s no
as well as invoking a debugger on a need to take your hands off the
running program. It can interface with keyboard even for a moment.
a CVS server for the more serious ■ See Emacs in action on page 88.
developers, and it can give you options
for viewing the differences between
LINUX FORMAT VERDICT
files and applying patches.
Like Vim, Emacs can be run in An excellent and extensible programmer’s
editor, but can be problematic if memory is
either graphical or text-only mode. at a premium.
This allows the use of a familiar editor
in the event of an X failure, but of RATING 8/10
course without the on-screen menus
and tool bars. After a while using the
CoolEdit’s
novel looks and behaviour may
not suit everyone, but it’s capable and innovative.
It’s hard to see why this might have present. Happily, this was easily fixed,
been included, as it’s impossible to and installing from a binary package
receive mail with the application. we’d have missed this problem entirely,
Perhaps this is part of the plan for a so this is only a minor issue.
future version, but we’re not why the
program’s authors would try to LINUX FORMAT VERDICT
compete with Emacs, which already
Looks ugly, but if you want a built-in
has a large and established user base. debugger it might be just what you need.
We also had problems building
CoolEdit – it requires the X libraries, RATING 6/10
but the configuration script neglected
to throw an error if they weren’t
LXF70 SEPTEMBER 2005 39
11/7/05 11:02:44 am
 ROUNDUP Text editors
Minimum Profit
Silly name but a serious, functional editor.
■ VERSION 3.3.10 ■ WEB www.triptico.com/software/mp.html
■ PRICE Free under GPL
Minimum Profit is another editor
that can be run either in an X
graphical environment or from the
console, allowing users to choose
which version they’re happier with.
Although parts of Minimum Profit
date back to 1991 it’s unlikely that
many users will have come across it,
mostly because it’s a much smaller
project than most text editors, being
maintained by just one person.
This is reflected in its functionality,
which is somewhat uneven in its
implementation – while it supports
multiple tabs so that more than one
file can be edited at once, there’s no
simple undo function. Syntax
highlighting for a large number of
languages is supported, but again
there’s no default support for regular
expressions, something lots of
programmers want to use. Regular
expressions can, however, be enabled
as an option.
This variation is probably due to
Minimum Profit being the product of a
single developer (Angel Ortega) – it
only provides the functionality that he
thought to put in. Still, Minimum Profit
is much more accessible than Emacs
or Vim, and more functional than
basic editors such as nano.
One of the most interesting
functions is the ability to get Minimum
Profit to explain any keyword within
the file you’re editing – just position
A menu-driven
text-mode
interface makes
Minimum Profit
easy to get
started with.
40 LXF70 SEPTEMBER 2005
LXF70.round 40
gedit
Gnome’s text editor leaves a lot to be desired...
■ VERSION 2.8.3 ■ WEB www.gnome.org/projects/gedit
■ PRICE Free under GPL
the cursor over the word you need
explaining, and the program will
search the man, perldoc or ri
(shorthand for Ruby Information)
pages to find you an explanation.
Another feature that’s emphasised
is the program’s ability to encrypt the
text file, protecting it with a password.
While it’s not a new feature (Vim has
supported it for some time), most
editors don’t make encryption this
obvious or easy to use.
Templates are also supported, and
these can be incredibly useful in
preparing documents, web pages and
macros. If there’s a task you use often,
defining a macro for it will save a lot
of time – you could run a macro to
insert header and footer information
to a number of files, or to search all
open files for a given set of text to
replace it. Other editors support this,
but it’s much easier to use here.
As an intermediate editor,
Minimum Profit succeeds where many Syntax highlighting appears to be the pinnacle of gedit’s abilities.
others have failed. We hope to see
more of it in future. Gedit is the official editor of the some of the most common (including
Gnome environment, so it is probably shell scripting). The syntax highlight
LINUX FORMAT VERDICT already installed on any computer method also makes a distinction
Fitting nicely in the gap between a basic which already uses (or has installed) between compiled and interpreted
notepad and an advanced developer’s the popular desktop. Unlike KDE, languages, for reasons we couldn’t
editor, Minimum Profit offers some of the
Gnome comes with a single text editor work out – all that seems to be gained
best features of both.
rather than two aimed at different is further confusion for the user.
RATING 7/10 tasks. As a result, gedit falls There’s no scope for extensibility,
somewhere between a programmer’s either by allowing plugins to be written
editor and a basic notepad application, by or allowing the user to run shell
and it doesn’t really do either job commands against the open file.
particularly well. There’s a Run Shell Command option,
While gedit offers the usual but it’s not clear how to make it run
programming functionality of syntax on what’s currently being edited. The
highlighting, indentation and so on, spellchecker is also deficient
none of this functionality is evident (apostrophes are important, folks!).
the first time the program is started. As a basic notepad editor without
Digging around in the options and the programming functions, gedit
settings will eventually reveal the would excel. Unfortunately, with these
program’s features, but we’d rather added so haphazardly it becomes
they were present from the first time confusing for the new user and
it’s started – an empty menu is even frustrating for the programmer, so
more confusing to a new user than there’s little to recommend it to either.
one full of unfamiliar entries.
Even when you do find them, the LINUX FORMAT VERDICT
functionality here is too basic to Gedit offers little to entice either the novice
provide a useful editing environment – or experienced users.
there’s no regular expression matching
in the search function; and the syntax RATING 3 /10
highlighting supports only a small
number of languages, missing out
www.linuxformat.co.uk
11/7/05 11:02:48 am
 ROUNDUP Text editors

TEXT EDITORS OVER TO YOU

THE VERDICT We couldn’t let

to stoke up an
flame war pass
you’re unmove
your opinion of
editors on the
this opportunit
Emacs vs Vim
by. Or perhaps
d by either. Post
the Roundup
LXF
y

Even the most basic
editors covered in this
roundup are good enough to
make a few quick changes to a config
file. It’s in the provision of useful and
extensible behaviours that they differ.
One of the big debates, which has
been going on for decades, is whether
Vi or Emacs provides the
better text-editing environment. Each
has its champions, and given that Vi
and Emacs users want different things
out of their editors, the argument isn’t
likely to be settled any decade soon.
Although GNU Emacs and Vim
have made the transition from text-
only to graphical editors and retained
VIM
their underlying structure and
interface, newer users are more likely
to come across graphical editors built
into their environment, such as Kate.
These editors are being used by an
increasing number of new users who
would have used one of the older
editors in the past – at least until they
need a more powerful program.
Kate came out well in our
Roundup, as well as another graphics-
only editor, NEdit. They both provide a
more than adequate environment for
www.linuxform forum at
at.co.uk.
which is the only
truly poor program we
looked at. Even this has potential,
though – if the developers could focus
on one group of users it would help
gedit’s cause immeasurably.
For newer users needing a
console-based editor with more power
than nano, Minimum Profit provides an
easy-to-use interface and a gentle
learning curve, allowing the novice to

9/10 software development. Kate is likely to

be installed on many systems by
default and it has better project
support, making for easier
management of large programs.
But NEdit wins through on
simplicity – it’s had several years to
perfect an environment that makes
complex and simple changes
straightforward, and when it comes to
editing single files, it’s harder to find a
graphical editor that will do any better.
The Gnome project would do well
to learn from this, and if NEdit were
rebuilt against GTK it would lead to a
Vim is an incredibly powerful editing tool, but is not to everyone’s taste. far more useful editor than gedit,
get stuck right in. It isn’t (yet) a very
popular editor, so it isn’t installed on
many systems by default. Nano or its
predecessor Pico, however, are
available almost everywhere.
Overall, though, the best editor we
found is Vim. While it’s hard to learn,
it’s extremely powerful and has a light
memory footprint compared with
Emacs. It’s also about as widespread
as they come – and small enough to
fit on a rescue disk.
If you’re serious about learning the
internals of Linux, give Vim a go –
you’ll soon learn enough to survive in
even the most dire situations. LXF

TABLE OF FEATURES
Name Toolkit Text mode Syntax highlighting Script language Debugger Regex support
CoolEdit Home-made ✔ Python ✔ ✔
gedit GTK ✔ N/A
GNU Emacs GTK ✔ ✔ Lisp External (GBD) ✔
Kate Qt ✔ Plugins ✔

nano None ✔ N/A

NEdit Motif ✔ Plugins External (GBD) ✔
Minimum Profit GTK ✔ ✔ N/A ✔
Vim GTK ✔ ✔ Perl/Python/Ruby ✔

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 41

LXF70.round 41 11/7/05 11:02:52 am

 HOTPICKS

HotPicks The best new open source software on the planet!

FIREWALL CONFIGURER Firewall Builder works primarily

Mike Saunders

Firewall Builder
with the concept of objects shared
When not hunting
HotPicks, Mike works across firewalls – networks, hosts, TCP
for Willy Wonka. ports and so forth. These are added to
your firewall via a drop-down menu. To
■ VERSION 2.07 ■ WEB www.fwbuilder.org create a new host entry, for example,
This is the place where we get
to profile some of the hottest you work through a dialog that offers
software around. to determine the interfaces
Each month we trawl automatically via SNMP or through
through the hundreds of open manual entry. Helpfully, most of the
source projects that are entries can have textual comments
released or updated, and attached, which is a particular bonus if
select the newest, most another admin needs to fathom out
inventive and best for your what you’ve been doing at some point.
perusal. Most of the HotPicks
Firewall Builder stores its projects in
are available for you to try out
on our coverdiscs, but we've XML format, and generates bash
provided web links if you want scripts for use on the target system,
to make sure you have the ready to be dropped into the system’s
very latest version. startup scripts. These bash scripts are
If you know some open furnished with appropriate variable
source software that you’d like and function names, so manual editing,
us to cover, email me at mike. if necessary, won’t be too much of a
saunders@futurenet.co.uk. chore. The compilation and script
generation process gives you adequate
feedback, and it performs fairly well,
HOTPICKS churning out highly complex projects
AT A GLANCE in a matter of seconds.
Anyterm ................................ 46 Firewall Builder’s interface is busy, but not overly cluttered.

Critical Mass ....................... 45 SSH flexibility

Drivel ........................................47
Firewall Builder ............... 42
GRAMPS ................................47
KsirK......................................... 45
KXDocker ............................. 46
Lukutoukka......................... 43
There are two ways to configure a
firewall in Linux: via the command line,
or with a graphical setup tool. Most of
the major distros now include some
form of firewall graphical configurator,
but often a more advanced, dedicated
solutions such as Firewall Builder can
Numerous binary packages are
available, but if you choose to compile
from source you won’t have any
problems – only Qt is required, and
the project’s own libfwbuilder has to be
compiled before starting.
Another nifty feature is the option to
upload your resulting script on to a
different system via SSH. This means
you can run the configuration
front-end on a separate machine.
Special mention should also go to
Firewall Builder’s supremely good

QPaMaT ................................. 43
do a better job.
The project’s website is jam-
Vazaar ..................................... 44 packed with extensive documentation
and information, explaining the app’s
LOOK OUT FOR THE underlying workings in great detail.
HOTPICKS AWARD This level of professionalism in the
Everything covered in our HotPicks website and support is always a bonus
section is unmissable, but each
month we single out one project for
– it also keeps middle-manager types
outstanding feeling confident.
brilliance. Only As it’s built on Qt, Firewall Builder’s
the very best
is chosen! front-end is available for Linux,
FreeBSD, Mac OS X and Windows.
The current releases support a decent
range of firewalling systems: iptables
on Linux; ipfilter on Free/OpenBSD
and Solaris; ipfw on FreeBSD and OS
X; and lastly pf on OpenBSD.
Welcoming layout
Once you’ve launched it, Firewall
Builder offers you the choice of
opening a file or starting a new project
with a short wizard-like process.
So, how easy is it to use? The main
window where configuration takes
place is clean and attractive.
Tooltipped icons and keybinding-happy
menu entries abound, and the panels
can be easily resized to make good
use of the screen. A handy tree
structure in the top-left assists in
navigation while avoiding too much
clutter, and in general, anyone with
elementary experience configuring
firewalls shouldn’t be too challenged.
documentation: the 116-page user
guide is peppered with tips and
screenshots, and is very well written.
Firewall Builder is remarkably
accomplished, boasting an excellent
interface, stacks of flexibility and
fantastic documentation. The ability to
generate firewall scripts for OSs other
than Linux is a great bonus, and while
it’s not as welcoming as some of the
simplistic tools that often come
bundled with distros, the added
versatility makes it worth the effort. If
you’re looking for a firewall config tool
that’s loaded with options, it’s well
worth seeing if it matches your specific
requirements – chances are, it will.

42 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.hot 42 11/7/05 1:17:16 pm


PASSWORD MANAGER
QPaMaT
■ VERSION 0.4.3 ■ WEB http://qpamat.berlios.de
Where’s the best place to store
passwords? Ideally, we’d all keep them
in our head and never write them
down, but that’s not so easy when
you’ve got a whole stack to remember
(email, websites, SSH logins…) And
then there’s the bothersome problem
of password length – it’s a doddle to
keep track of ultra-short ones, but
they are hardly the most secure.
Password managers like the awkwardly
named QPaMaT are the answer,
providing a single store for all your
passwords with added safety on top.
QPaMaT is the Qt Password
Management Tool, and you’ll need Qt
along with OpenSSL in order to build
from source. This should be no
problem on most distros, as long as
you have the Qt development
packages installed. If compiling from
source fails for some reason, you can
try the RPM binary.
TEXT READER
Lukutoukka
■ VERSION 0.2 ■ WEB www.helsinki.fi/~pakaste/lukutoukka
From snappy news headlines to
long emails, short blogs to PDF
documents, there isn’t much text that
can’t be displayed on our monitors,
but many people are still getting
used to this new way of on-screen
reading. Lukutoukka is an intriguing
little app that gives you an alternative
to going through a document one
line at a time – instead, it displays
one word at a time in boxes that fly
past your eyes at great pace. It can
be an effective (although rather
strange) way to read.
Compiling Lukutoukka involves a
few somersaults: you’ll need to have
Guile and Guile-Gnome installed
beforehand, along with their
dependencies (G-Wrap and co).
Annoyingly, this means that some
components may have to be
compiled manually as not many
distros have them all neatly
packaged up. And the Guile wrappers
take an absolute age to compile. It’s
a shame that such a vast tree of
LXF70.hot 43
HOTPICKS
When first started, QPaMaT
prompts you for a global password:
this is the crucial string that keeps
all the individual passwords secure,
so you have to select something
difficult. Indeed, the program
won’t proceed unless you enter
something übercomplex with
non-alphanumeric characters.
QPaMaT’s main window is
easy to navigate, containing a
tree list of categories for sorting
your passwords – shopping sites, QPaMaT’s
level of configuration
server logins and so forth. In the right-
for password strengths
hand pane you enter usernames, is top-notch.
passwords and URLs, along with
additional information such as answers
to secret questions. blob appears and you can generate QPaMaT’s simplicity, coupled with
Your entered data is encrypted new passwords to a specific strength. the checking and suggestion features,
form and squeezed into an XML file. You can set the estimated days it will makes it a pleasure to use. If you’re
Brilliantly, QPaMaT colour-codes take to crack an entry, and various juggling passwords, always forgetting
entries according to their strength, so options for auto-filling of entries can vital logins and obscuring your monitor
if one of your passwords is weak, a red be tweaked. with Post-It notes, it’s a superb solution.
dependencies is required for such a
small app, but that’s the downside to
using brand-new apps on a cutting-
edge operating system like Linux.
Lukutoukka’s main window is a
spartan affair, designed with
cleanliness in mind – you don’t want
any distractions as the text is
whizzing past.
After you open a text file and hit
the Play button, Lukutoukka flashes
up the text one word at a time, as
quickly or slowly as you want. Smartly,
it adds a slightly longer pause
between words for full stops and
commas, which works very well.
A speed slider lets you step
backwards and forwards through a
document, and there’s a spinner
widget for navigating to specific Wait, what was the last word? Thankfully there’s a rewind option...
points in the text.
And that’s basically all there is to of other prefs – there’s no need for fascinating to try nonetheless. Just
it. The font can be changed to ’em. Whether or not Lukutoukka turns don’t try to read a source-code file
something larger (useful if you’re out to be a gimmick or a useful through it – you might go insane.
sitting back from the monitor and alternative to normal reading is That, or become the best
reading), but there’s little in the way entirely down to the user; it’s programmer in the universe.
www.linuxformat.co.uk LXF70 SEPTEMBER 2005 43
11/7/05 1:17:21 pm
 HOTPICKS
ITEM ORGANISER
Vazaar
■ VERSION 0.19 ■ WEB www.kaskaras.net/vazaar
Trivia question: in which HotPicks was that F1GP shot last included?
Where would we be without
material possessions? Despite all the
talk of the paperless office, most of us
still have a heap of objects on our
desks that get lost, broken or
borrowed indefinitely. So, unless you're
fantastically organised and have
everything neatly pigeonholed in the
appropriate places, things get a little
messy.
This extends to computers too.
Even if the paperless office were a
reality, we have no reason to believe
that things would be any better
electronically. On our personal systems
we tend to have music files, emails,
mid-development projects and a
bajillion other things strewn about that
can only just be organised with a
traditional directory structure. Gnome
Vazaar strives to make cataloguing
and organising these items on your
drive altogether more elegant, using a
single repository split into categories.
Even though it’s still in early
development, Vazaar is moving along
well and is already good enough for
regular use.
To compile Vazaar from source
(the only option at the moment, as no
binary packages are available), you’ll
need to have the Gnome development
44 LXF70 SEPTEMBER 2005
LXF70.hot 44
packages installed. On most distros
these follow the naming convention of
gtk-devel, gnome-devel and so on, but
the configuration script should let you
know what’s missing (gnet is one
dependency that a lot of systems
don’t have installed by default).
Once you have all the building
blocks, running make and make
install (as root) will drop the files into
place, and you can start it up by
running vazaar from a shell prompt.
Pain-free panes
Vazaar’s main window adopts the
current Gnome icon set, fitting in
perfectly with the rest of the desktop.
A list of categories sits down the left-
hand side, while on the right two
panes contain the list of items and
their actual contents. The toolbar is
responsible for most of Vazaar’s
current functionality; some of the
menu items have yet to be coded in to
it, but there’s not a lot in the menu
anyway so it’s no great loss. It’s all
perfectly simple and easy-going.
By default, Vazaar is supplied with
four general-purpose categories –
Favourites, Generic, Important and
Wastebasket – but new ones can be
added in a snap. Each category can be
www.linuxformat.co.uk
assigned descriptive text and an icon,
making it easier to locate something
when the list of items has started to
build up. In another neat cosmetic
touch, the category and item lists have
alternating shaded backgrounds, so
looking through a long list of stuff
doesn’t hurt your eyes. Categories can
even be nested for especially
complicated collections.
The category options box offers a plentiful range of colourful icons.
The items can contain nigh-on any
type of file, and can be placed into
multiple categories, so you could have
something related to ‘Work’ that is
also designated ‘Important’. Vazaar
attempts to display relevant
information about the file in the
bottom pane, even providing a Launch
App button for many, so text files, for
instance, will fire up gEdit. The info
panel also displays some file
information such as date of creation
and last modification.
What’s up, docs?
There’s little in the way of configuration
options as yet; nor is there any
documentation at this stage, aside
from the few scraps of information on
the project’s website. It’s still early
days for the project and there is a lot
to be done.
Vazaar is a breeze to pick up and
use, but we’d like to see more desktop
integration. Drag-and-drop support for
adding items from your desktop is
planned for a future release, so open-
source fingers are crossed. Crucially,
the developer has already put care
and attention into crafting a welcoming
interface – this will pay off as the
program matures and finds use in
more ambitious collections.
By the time this issue hits the
shelves, another release of Vazaar
may be doing the rounds with a few of
the missing features coded in and a bit
more refinement. But regardless, it’s
already a speedy and attractive little
tool for managing files in a more
elegant way. So if your hard drive is
falling to pieces with the jumble of
documents and images scattered
everywhere, give it a go.
11/7/05 1:17:43 pm

STRATEGY GAME
KsirK
■ VERSION 1.2 ■ WEB https://gna.org/projects/ksirk
A diverting battle for global
domination, the board game Risk has
many fans – everybody wants to rule
the world, it seems. It offers a good
blend of strategy elements, and a
cathartic all-out attack. Experienced
players attend Risk Clubs (military
attire optional), and it’s no surprise
that computer-based clones have
been doing the rounds for years. One
such is KsirK. As the K profusion
suggests, it’s an implementation for
KDE, so you’ll need the KDE
development packages installed to
build the game from source.
KsirK is a tad rough round the
edges when it comes to presentation;
the tiny controls don’t give much
away (save for the tooltips) and the
in-game icons are rather minuscule.
Animated battles and rudimentary
sound effects liven things up a little –
SHOOT-’EM-UP GAME
Critical Mass
■ VERSION 0.9.11 ■ WEB http://criticalmass.sourceforge.net
“Your world has been overrun by
an aggressive army of space critters,”
the game’s blurb notifies us. Well, it
beats the current situation of the
world being overrun by lawyers and
marketing types, but clearly
something has to be done. And
It lacks visual finesse but Critical Mass has plenty of retro charm.
LXF70.hot 45
all the goings-on are detailed along
the bottom in the status bar. Fancy
visuals aren’t of supreme importance
in a strategy romp, but if you’re
running in any resolution lower than
1024x768 your window manager will
have to behave well, as the play area
takes up acres of space.
Two to six megalomaniacs can play,
with the option of using computer-
driven opponents if necessary. KsirK’s
author describes it as a “basic AI”, so it
may be something of a pushover for
more experienced Riskers and strategy
fans. Using the toolbars you can
organise your army, invade territories
and control various other aspects of
the play without any big hassles. World
domination has never been simpler.
KsirK is playable as it stands right
now, but in the pipeline are better CPU
opponents, networked multiplayer
guess what? You’re just the person to
do it. Having been put in control of a
tiny spacecraft, you must eliminate the
invading forces and pick up resource
containers along the way. Yes, it’s the
usual blasting fare, but it’s all rather
compelling nevertheless.
www.linuxformat.co.uk
The KDE–Gnome flame war soon engulfed the planet.
mode (which would be a major boost)
and the option to select different
maps. It’s worth checking out the
development versions to see these
features implemented. But even at the
Building from source should be
hassle-free – it requires only SDL and
some add-on libraries, which are
available for almost every distro.
Critical Mass launches straight into
thumping techno music, a parallax
scrolling starfield background and a
menu for tweaking various settings
(full-screen mode, resolution, sound
and the like) or jumping straight into
the gameplay. On the whole, the
presentation is pretty decent.
Occasionally the enemy fire gets lost
in the background, but thankfully that
can be disabled.
What appear to be carefully drawn
little sprites turn out to be polygon
models when you zoom in, so the
game performs well even on less
sprightly boxes. The gameplay itself is
pretty generic: shift your craft around
the screen, obliterating everything in
sight and making sure you’re not
blasted to smithereens by enemy fire.
While the soundtrack begins to grate
after a while, it’s fairly appropriate for
the quickfire action, keeping your
HOTPICKS
HOT GA
ME
S
current release, KsirK is an enjoyable
little game. If you fancy something
more cerebral than Critical Mass
(covered below), try it out and
release your military urges.
adrenaline pumping when the fate of
the world’s at stake. It’s the frantic
pace of the action that gives Critical
Mass a lift; there’s very little let-up in
the intensity and you’re constantly on
your toes. We challenge anyone to
last more than one minute on the
Insane skill level, which is aptly
named – it’s quite comically difficult.
Scrolling shooters have a real
retro feel today, but there’s still plenty
of mileage in the gameplay and
they’re a good choice when you
fancy a quick blast of something
diverting. Critical Mass ticks these
boxes – it’s loud, lively and
challenging, even if it’s not the most
imaginative game ever conceived.
Firewall Builder is also an ace pilot.
LXF70 SEPTEMBER 2005 45
11/7/05 1:17:45 pm
 HOTPICKS
DOCK FOR KDE
KXDocker
■ VERSION 0.33 ■ WEB www.xiaprojects.com/www/prodotti/kxdocker/main.php
KXDocker is a workalike of the
‘dock’ panel of Mac OS X, with
snazzy extras sprinkled on top. Its
default appearance consists of a
rounded grey panel, into which a
handful of Mac-esque icons of
applications and folders have been
dropped. As you run your mouse
over these launcher icons they scale
up smoothly, which is a joy to
behold – even on older systems, the
effects don’t stutter. Each launcher
has an alternative icon that appears
when a program is launched, an
event also marked by amusing
bouncing animations.
To alter settings, KXDocker
includes a slightly clumsy config
dialog. It has most of the options
you’d expect, but they’re not
organised elegantly and many are
lacking proper descriptions – not a
major issue for a piece of software
that’s still only at version 0.33. You
can manage the launchers, alter the
theme, tweak effects and enable
additional plugins. These include a
graphical pager, analogue clock and
battery monitor – most of the add-
ons you’d expect to find in other
desktop panels, and hopefully
there’s more to come.
KXDocker’s icons resize very smoothly as the mouse runs over. Nice.
46 LXF70 SEPTEMBER 2005
LXF70.hot 46
Helpfully, binary packages are
available for most popular distros,
but compiling KXDocker from source
shouldn’t pose any problems if you
have recent versions of KDE and Qt
installed. Just do ./configure, make
and make install (the last step as
root) and then run kxdocker from
a Konsole. The additional resources
package bundles together a wealth
of themes, icons and plugins, so
that’s worth installing too.
Our only gripe with KXDocker is
the startup speed: it’s not alarmingly
slow but it’d be nice to have it
running a bit faster. Otherwise, it’s a
fine desktop addition, bringing fancy
cosmetic touches and a cheerful
OS X feel to KDE. It’s also a great
way to wow potential newcomers
with a flashy desktop, instead of the
relatively dull normal panels.
Linux desktops receive a fair bit
of criticism for aping other operating
systems, and many people argue
that we need more innovation. But
plucking ideas from established
desktops can work well – it gives
users a familiar environment – and
we see no harm in valuable features
like this one recreated on Linux.
Well worth a look.
www.linuxformat.co.uk
An Anyterm session logged into a laptop far, far away...
WEB-BASED TERMINAL
Anyterm
■ VERSION 1.00 ■ WEB http://anyterm.org
One of the worst nightmares on the web page, an XMLHTTP
that server administrators face is being channel to the web server and
unable to log on to a computer. It’s Anyterm’s Apache module, which
doubly frustrating when it’s not the passes on requests to an emulated
server at fault, but the lack of tools – terminal. You can even run several
a restricted installation on a desktop terminals at once which, with the
box, where you can’t install SSH tools addition of tabbed browsing, is very
such as PuTTY. In such a scenario, effective. Naturally, there are various
Anyterm could be a lifesaver: it security issues involved with such a
provides a terminal embedded in a design, but Anyterm supports Secure
web page, acting like a regular shell Sockets Layer encryption to allay
on the server. some of your safety fears.
Unlike other efforts, which involve It’s all delightfully compact and
Java applets or are limited to single easy to set up.. There’s no lengthy
commands, Anyterm is built with just configuration (though you can alter
JavaScript and some behind-the- the terminal’s dimensions) and the
scenes trickery. Installation is fairly end result is much lighter than a Java
straightforward: you’ll need to compile equivalent would be. Best of all, it’ll
the ROTE terminal emulation library, work just about anywhere – and the
then build Anyterm’s own Apache author plans WAP support in the
module (only Apache 2 is supported) development version!
before dropping the files into place. A
quick hack of the httpd.conf and
you’re ready to go – just point the
browser at the main HTML file.
Anyterm presents an unfurnished
page containing a terminal emulation
window, which for the most part
behaves like any standard terminal. It’s
not the fastest thing in the world, but
colours are displayed and most
ncurses-based apps run as expected
so it handles the vast majority of tasks
with ease. Indeed, the author’s own
demo site shows a playable Tetris
variant in action. In most cases it’s
likely to be used for admin chores –
nothing to fault here.
All this is achieved using JavaScript
11/7/05 1:17:48 pm

HotPicks REVISITED
WEB LOG CLIENT
Drivel
■ VERSION 2.0 BETA 1 ■ WEB www.dropline.net/drivel
Interface-wise, Drivel hasn’t changed a great deal since 1.0.
Designed as a dedicated blogging
app that’s separate from the typical
web interfaces, Drivel impressed us
back in LXF57 with its polished design
and rock-solid stability. You’ll need GTK,
GENEALOGY SOFTWARE
GRAMPS
■ VERSION 2.03 ■ WEB www.gramps-project.org
Back in LXF56, GRAMPS (the
Genealogical Research and Analysis
Management Programming System),
snapped up our Hottest Pick award
thanks to its admirable progress and
attention to detail. As one of the very
few genealogy-related tools for Linux,
GRAMPS really stands out, and has
made superb progress since we took a
look at version 1.0. You’ll need Python
2.3 and Gnome 2.8 installed to run
GRAMPS, along with the PyGTK and
Gnome–Python bindings (supplied on
our coverdisc).
GRAMPS’s interface hasn’t changed
drastically since the last major release
– the layout has been tweaked, but
otherwise it retains the attractive icons
and professional look. Managing your
data on individuals and their
relationships, pictures and any further
details is straightforward; and it’s great
to be able to generate reports and
graphs in various formats (such as
HTML or OpenOffice.org Writer files).
LXF70.hot 47
Gnome and curl to build from source;
the RPM on our coverdisc should work
for several distros.
Drivel’s initial window prompts you
for a username and password before
The supplied
example file
is a good way
to understand
GRAMPS’s
inner
workings.
A boatload of updates and new
features has arrived with 2.0. These
include: import and export support
for the GeneWeb format; a snazzy
export wizard for transferring data
elsewhere; find-as-you-type in the
list views; an undo facility; report
generation via command-line options
(useful for scripting); a bunch of new
database back-ends including
Berkeley DB; and a plethora of
tweaks and bugfixes. With three
minor revisions since the 2.0 release,
www.linuxformat.co.uk
opening up the entry editing box for
jotting down your blog. Drivel is
geared towards regular bloggers
and, as such, includes a few default
settings for posts, and options for
sitting quietly in the taskbar you can
also add coloured HTML tags .
By far the greatest new addition
since 1.0 is the widened support for
blogging systems and protocols.
Whereas the early release only
worked with LiveJournal, 2.0 brings
about Movable Type, Blogger, Atom,
and Advogato support. Syntax
highlighting and spellchecking are
two other notable improvements,
while the client can now ping the
popular blogging site
www.technorati.com after you’ve
made an update. Thankfully, the
new goodies haven’t affected
stability – it’s still very solid.
Drivel’s 2.0 release offers
support for even more blog systems,
which could bring it many more
users, and if you keep a journal with
any of the above systems we highly
recommend it. It’s clean, fast, simple
and fits in well with Gnome (or
indeed Xfce).
loads of glitches have been cleaned
up, and it proved to be robust and
speedy in our testing.
GRAMPS may be more of a
specialised app than, say, a word
processor, but it still deserves a lot
more attention than it’s getting at
the moment. It’s a triumphant
demonstration that open-source
teams can produce professional and
polished apps for the long tail of the
software market. Genealogists, give
it a spin!
HOTPICKS
ALSO RELEASED
New and updated software that
also deserves a look...
■ StressOut 0.1.0
MySQL benchmarking
www.anerva.com/stressout
■ Jackbeat 0.5.1 – Sequencer (see
LXF69) www.xung.org/jackbeat
■ Code Browser 2.0
Programmer’s editor
http://code-browser.sourceforge.net
■ Fotoalbum 0.10 – Image
management http://kisocd.
sourceforge.net/page11021946.htm
■ Knight’s Mail 0.1.1 – Email client in
Mozilla XUL www.knightsmail.com
■ Outgun1.0.2 – Multiplayer action
game http://koti.mbnet.fi/outgun
■ gslapt 0.3.1
Slackware package management
http://software.jaos.org/#gslapt
■ CheckGmail 0.9.5
Gmail notified in your systray
http://checkgmail.sourceforge.net
■ JOE 3.3 – Powerful text editor
http://sourceforge.net/projects/
joe-editor
■ OpenRelease 0.1.2 – Software release
assistant http://open-release.sf.net
■ Kat 0.5.2 – Desktop search (see
LXF68) http://kat.sourceforge.net
■ MoreAmp 0.1.12
Battery-saving MP3 player
http://sf.net/projects/moreamp
■ ttylinux 4.7 – Ultra-tiny distro
www.minimalinux.org/ttylinux
■ Sysprof 0.9 – System profiler
www.daimi.au.dk/~sandmann/sysprof
■ GGZ Gaming Zone 0.0.11
Networked board games
www.ggzgamingzone.org
■ MyServer 0.8.1 – User-friendly web
server www.myserverproject.net
LXF70 SEPTEMBER 2005 47
11/7/05 1:17:56 pm
 WHAT ON EARTH PDTP
What on Earth is...
PDTP?
>> What does a Post-Diploma
Training Programme have to do
with free software?
Nothing – this is an all-new four-letter acronym that
has very little to do with diplomas. Instead, it’s the
Peer Distributed Transfer Protocol: an advanced file
transfer system designed to distribute bandwidth
needs across many clients.
Call me stupid, but that sounds
>> like BitTorrent.
Indeed it does, and to be fair, PDTP is very close to
BitTorrent in its goal and design. However, there are
enough differences between the two to warrant the
developers starting from scratch. That said, if you
understand how BitTorrent works you’re already a
long way to understanding how PDTP works.
So it’s an improvement on
>> BitTorrent?
Some Java developers will tell you that their
language is an improvement on C++ because Sun
corrected all the design errors in C++. On the
flipside, lots of C++ developers will tell you that Java
has many brainfarts all of its own, proving that it’s
awfully hard to improve on something without
weakening it in other ways.
So you’re saying that PDTP
>> isn’t perfect?
Well, a key sign that the developers might not be
firing on all cylinders is their claim that PDTP must
be better than BitTorrent because BT is written in
Python and is therefore slow. That’s true, but unless
you’re running on a 486 you probably won’t notice
the difference. Python is very portable, which is a
huge advantage over C – the language in which
PDTP is written.
Surely the choice of language is
>> pretty minor in the grand scheme
of things?
Well, we can only hope. But there are many other
key differences between PDTP and BT.
48 LXF70 SEPTEMBER 2005
LXF70.woe 48
>> Such as…?
Nothing annoys Paul Hudson
more than choosing a mirror;
but will PDTP manage to end
the rule of BitTorrent?
Well, let’s first start by reviewing how BT
works. A hypothetical file, fc4dvd.iso, is on a server
somewhere. This being 4GB in size and very popular,
the poor server will soon find its bandwidth
overwhelmed, and most potential downloaders will
be turned away empty-handed. BT solves this
problem by allowing the creation of a tracker: a
server that monitors people trying to download the
file, and gets them to share it among themselves.
So that tracker hosts the file
>> instead of the server?
No – the original file stays in the same place. The
downloader goes to the tracker rather than directly
to the file. If they are the first person to request it,
they will start downloading it direct from the original
server. Otherwise, the tracker directs them to the
original server and also to the first client – the first
client will serve what it has so far to the second
client. As more clients come on board, the transfer
speed picks up.
How does the tracker know who
>> has what parts of the file?
When a client first connects, the tracker sends it
information over HTTP about other clients
downloading the file so that it can start downloading
from them. BT is quite smart in this respect, with
clients being programmed to download the least-
popular chunks of a file first so that overall
distribution of the file is kept equal. As clients receive
sections of the file, they report back to the server so
that other clients can use them as a server for the
part they’ve just downloaded.
OK, thanks for the refresher. Apart
>> from Python, what’s so
terribly wrong with BT that we
need PDTP?
8/7/05 5:16:51 pm

BT is designed to allow scalable downloads – to
distribute a file to thousands of users simultaneously
rather than just a few. It turns out that each chunk of
a file is just 256KB in size, which takes under ten
seconds to download on a broadband internet
connection. So, even if only 1,000 people are trying
to download our DVD ISO file we still get 1,000
HTTP requests going over the network every ten
seconds as the clients update the tracker with their
download status. A hundred requests a second is a
lot for most servers, and a huge bandwidth waste in
itself; but what if we 10,000 people are downloading
a file, or 100,000 people? In these scenarios BT falls
over spectacularly – instead of the original file server
getting swamped, the tracker gets swamped.
How does PDTP get around this?
>> Well, in the same way that BT distributes file
downloading among the clients, PDTP also allows
you to distribute the job of the tracker across
multiple machines. These trackers communicate to
clients, but also with each other to maximise
bandwidth usage.
LXF70.woe 49
WHAT ON EARTH PDTP
Ah. So are PDTP clients different
ILLUSTRATION: CHRIS WINN
>> from BT clients?
Apart from being written in C rather than Python,
they are largely the same – the real magic comes in
the protocol.
Go on…
>> Well, unlike the very basic Torrent files
associated with BT, PDTP uses an XML-based
format to allow metadata to be attached to files.
Want to know what codec a video file uses before
you download it? How about grabbing the name of
an artist and the album while you download music?
But downloading music is piracy!
>> That depends on where you live, and what
you’re downloading. But the use of XML is very cool…
Yes, but XML hardly makes things
>> easier – the nice thing about
standards is that there are so many to
choose from, right?
That’s true, but the developers are working towards
using the Dublin Core Metadata element set, so you
should be able to interface with the metadata from
many XML-enabled applications. It should make it
much easier to search networks for files you’re
interested in…
I didn’t think there were that
>> many BT downloads around…
Well, no, there aren’t. But that’s largely as a result of
the overhead of setting up a BT download – it’s
really designed for individual files. On the other hand,
PDTP is designed to replace FTP and HTTP for
popular downloads, and so has been designed to
handle multiple nested directories of files, like an
FTP server. One of the promised features is the
ability to run a simple script that indexes a directory
and makes it available online automatically.
That sounds a lot easier, but surely
>> if one file server is expected to
provide many files it will just get
overloaded that much faster?
In a very basic PDTP configuration, that’s quite true.
In fact, with multiple tracker servers it becomes
possible to serve many more clients, resulting in the
file server getting overloaded much faster than it
would with BT. However, PDTP lets you spread the
load, allowing you to set up a PDTP piece proxy
server that spreads the load of the PDTP file server.
Wait, your terminology is
>> confusing me – there’s a server, a
file server, a proxy server… they all
sound the same!
Although a lot of the technology is the same,
PDTP uses different terminology from BT,
which causes some initial confusion. You
already know what the client is – that’s us,
downloading files. A PDTP server is what we >>
would normally called a tracker – it tracks
LXF70 SEPTEMBER 2005 49
8/7/05 5:16:54 pm
 WHAT ON EARTH PDTP

>> which clients are around, which parts of the files

they have, and also makes sure that clients upload a
fair amount of data compared with what they
download. A PDTP hub hosts the files being served
to the network, and is the ultimate root of the
network – which is why it’s called the hub!
The hub serves up the XML description files so
that everyone knows what files are available and
their checksums. There’s also the piece proxy, which
spreads the load of the hub, and a standard proxy,
which is designed to allow hosts behind firewalls to
take part in the network.
>> Hold up there – you didn’t
mention checksums before.
Only because it’s a common-sense precaution.
Transferring a big file over the internet is very hard to
do, but transferring several thousand small files that
combine to make the big file is even harder. BT gets
around this by using the Secure Hash Algorithm
(SHA1) to generate a checksum of each file part,
which is used to ensure that the part arrived safely.
PDTP takes this a step further by using the Digital
Signature Standard (DSS) to cryptographically
protect and checksum each file part.
>> That sounds quite CPU-intensive –
I thought PDTP was fast?
It is fast, but it doesn’t achieve that by compromising
on security. For a start, PDTP ditches HTTP in place
of its own new protocol. HTTP is a very easy,
connectionless protocol that was designed for
transferring textual information. BT uses it because
most developers understand how HTTP works and it
already has a lot of support in the industry. However,
HTTP is much too complicated when all you want to
do is let a server know you have successfully
downloaded a file chunk and want the next one.
PDTP uses a binary protocol that can be validated
easily to make sure each request and response is
valid, and is much faster for computers to use than
HTTP. Keep in mind that HTTP is text-based, which
means that it needs to be parsed and understood by
the computer before it can be acted on; binary
protocols get around this.

But what about the security issues

>> associated with binary protocols?
The ASN.1 protocol recently had two
big security holes!
That’s true, and that’s why PDTP doesn’t use ASN.1.
Not only does the complexity of ASN.1 breed
security issues, but it also slows things down – PDTP
is about performance, remember! Having a binary
protocol does make development harder, but the
PDTP development team are working around that by
creating a client library.

Something accessible that

>> everyone can use…
…without having to learn the protocol themselves –
precisely. The client library will have all the binary
protocol data handling and validation, but expose to
developers a wrapper that means they shouldn’t
need to get their hands dirty with the hard stuff. It
also means that developers can go ahead and use
Python or their language of choice, but the core of
the system will remain in lightning-fast C.

I’m seeing lots of “will”, “should”,

>> and “promised”; nothing concrete
at all. What’s going on?
PDTP is largely still on the drawing board, where it
has been for some time. You see, BT filled an
immediate need: home users were increasingly using
broadband, which moved the bottleneck from users
to servers – they just couldn’t keep up with people.
BT made it possible for lots of users to download
popular files; the design was very simple. Of course,
it has evolved from that point to add more features,
but this bottom-up approach to development has
meant that incremental improvements have been
added to something that’s already in use. The PDTP
developers are trying to put all the features they
want into the initial release, which is taking a very
long time. As Linus Torvalds once said: “Nobody
should start to undertake a large project. You start
with a small, trivial project, and you should never
expect it to get large. If you do, you’ll just overdesign.”

50 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.woe 50 8/7/05 5:16:57 pm


So PDTP has overdesigned a little?
>> Well, it’s certainly not starting small, that’s for
sure. The team had planned to release the first
milestone version at DefCon 12 last summer, but
they missed that deadline. Large parts of the project
are still in pre-alpha stage – the code is there, it’s
usable, but it’s not suitable for public testing. Some
parts are already in alpha testing, but the whole
project needs to be complete for any one part to
really be usable. Top down, you see. Some things
have been left out so far, such as support for IPv6.
Support for standard IPv4 is already there, and it’s
not hard to add support for IPv6, but it is hard to
add support for IPv6 hosts that want to talk to IPv4
hosts – hence the lack of code. This may be added
in the initial release, but it will probably be delayed.
Hey, this is supposed to be “What
>> on Earth is PDTP?”, not “What on
Earth might PDTP be one day?”!
“WITH PDTP YOU DON’T HAVE TO
CHOOSE A MIRROR, WHICH SHOULD
SAVE MONEY AND SPEED THINGS UP.”
Yes, but we do try to use this space to highlight new
and interesting technologies. Otherwise, how can you
look cool in front of your friends?
Fair enough. So are they looking
>> for developers to help out?
Is a bear Catholic? Anyone with an interest in C or
network programming is encouraged to help out –
even if it’s just running the code and seeing what
breaks. It’s under the OSI-approved Apache License,
and you should be prepared to contribute your work
back under the same licence.
So is the actual specification
>> available to work to?
There are drafts at http://ptdp.org/protocol.php,
but at the time of writing they are all under
production and remain incomplete. For example, the
developers are still waiting for a permanent port
number to be assigned to them from the Internet
Assigned Numbers Authority (IANA). However, aside
from that, the rest of the protocol specification is
available for developers to read and implement if
they can’t wait for the client library.
Are the developers producing the
>> client library alone, leaving third
parties to do the rest?
Not at all! The server components – the hub, the
server, the proxy, and the piece proxy – are all being
actively developed by the core PDTP development
team. The project is called Squall, and you can
download it from the SourceForge site at http://
squall.sourceforge.net. They are also working on
a client program that uses the library, called Skyfire.
LXF70.woe 51
This is written using Qt for cross-platform support –
Linux is one of the key target platforms, naturally,
but Windows, BSD, Mac OS X and Solaris are also
on the list.
I think I’ve grasped all the
>> technical bits now. Practically
speaking, how will PDTP make a
difference in a world of BT?
For most people, the key difference will be PDTP’s
directory-oriented nature. Sites like SourceForge,
MySQL.com and PHP.net all rely on an outdated
system of mirrors that requires you to go to a
website, choose your file, go to a different website
depending on your location in the world, then finally
get the file you were after. Not only is this a waste of
time, it’s a waste of resources: while one server is
groaning under the weight of a thousand users, it’s
quite possible that three others are crunching
SETI@Home blocks in their idle time. PDTP
eliminates the need for human intervention. Rather
than choosing a mirror, each user is served by every
mirror along with other users. Not only does this
eliminate the need for distributors to spend so much
time mirroring data, it should also save money and
make things faster.
It sounds like they’ve really got
>> it under control. Is this the end
for BitTorrent?
Well, no. In fact, BT is moving the goalposts
somewhat by altering its infrastructure to work
around many of the problems PDTP is designed to
fix. In May of this year, Bram Cohen (the creator of
BitTorrent) uploaded a new beta of BT that added
support for trackerless torrents, effectively
decentralising the BT system. This was the biggest
scaling issue in BT, and the prime reason for the
creation of PDTP – the slow development of PDTP
has caught up with it. And there are other BT
improvements on their way, such as the ability to
search the new trackerless torrents for specific
content direct from Firefox. This is a busy time for BT,
but the important thing is that each improvement is
incremental: users can enjoy all the benefits of the
existing system without having to wait for the
updates, unlike with PDTP.
Is PDTP the only thing trying to
>> edge out BT?
In the spirit of choice, there are a number of other
distributed download systems around, including
GNUnet, Rodi, MUTE, and others. Rodi is particularly
interesting, as it is able to offer some degree of
anonymity by using other clients on the network to
www.linuxformat.co.uk
WHAT ON EARTH PDTP
stand between yourself and the server. So, instead of
client A downloading a file directly from server B,
client A asks client C to download the file from B,
and it passes the file along – B need never know A’s
IP address. Of course, even this isn’t secure: all it
would take is the RIAA (or some other litigious
group) to create a number of C clients and watch all
the traffic.
Wait – BitTorrent isn’t anonymous?
>> No, and neither is PDTP. Can you hear the
sound of sirens approaching?
I’m innocent!
>> Tell that to the judge! BT and PDTP aren’t
designed to be anonymous, and it’s unlikely they
ever will be. To be fair, there’s little need: unless
you’re trying to download Revenge of the Sith, you
don’t really need anonymity. Downloading illegal
software and movies would be made easier, yes, but
that’s the last thing that the BT and PDTP
developers want. They already have a hard enough
time avoiding lawsuits as it is, without adding
features that will almost inevitably be used for
copyright theft.
But what about freedom of
>> speech? What about my rights?
I’m sure there’s something in the
Universal Declaration of Human Rights
about this!
I think you’re on your own there. If you want to
download legal stuff, great, just do it – no one cares
who downloads Fedora or Mandriva, and no one is
spying on you. You’re just not that interesting, sorry.
But if you want to avoid paying people for their work
and download copyrighted things for free, think
again: PDTP doesn’t come to the rescue.
So where do I go from here if the
>> software isn’t available just yet?
You can download one of the nightly source code
builds straight from the PDTP repositories. The client
library is available from http://svn.pdtp.org/
snapshots/libpdtp-snapshot.tgz and the server
components available from http://svn.pdtp.org/
snapshots/squall-snapshot.tgz. If you have
Subversion installed (see our tutorial series if not),
you can get your mitts on the very latest code with
the command svn co http://svn.pdtp.org/svn/
libpdtp or svn co http://svn.pdtp.org/svn/squall.
If you’re interested in helping out with the
development effort, you can contact the lead
developer directly at tarcieri@pdtp.org.
Any particular dependencies I
>> need to be aware of?
Well, remember that the software is still pre-alpha,
so the developers haven’t had time to clean the
code up much. One thing you definitely should have
installed is libcrypto, which adds the DSS support for
checksumming. For more information, go to the
ecxellent FAQ at http://pdtp.org. LXF
LXF70 SEPTEMBER 2005 51
8/7/05 5:17:01 pm
 DEBIAN AT A CROSSROADS

COVER
FEATURE Debian at a
crossroads
Sarge is out at last – a release that at one time seemed as likely as soaring swine –
but it’s delays and waning influence, not features and improvements, that people are talking
about. Richard Drummond assesses the road ahead for this freest of Linux distributions.

O
n June 6 this year you could
almost hear the collective sigh
of relief from a thousand
Debian developers around the
globe: the next version of the
Debian GNU/Linux distribution
– version 3.1, codenamed Sarge – had
at last been officially released after a
difficult three-year gestation period.
Some of us had wondered if we would
ever see it.
In software development you’re
seldom allowed much time for rest,
however. No sooner is one version out
the door than it’s time to think about
the next one. And the Debian Project
has some challenges to face as its
developers get to work on Etch, the
name for what will eventually become
the next stable version of Debian.
The primary challenge appears to
be an internal management one: how
does the Debian Project ensure that
users aren’t made to wait another
three years for Etch? That people
might be prepared to wait that long at
all says much for the project’s unique
standing in the open source community
and the affection users feel for it. But
with its many derivatives gaining
momentum and large user bases of >>
their own, Debian has little time to lose.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 53

LXF70.feat_debian 53 11/7/05 11:01:23 am

 DEBIAN AT A CROSSROADS
>>
What a difference three years
make. K3b, OpenOffice.org and
Firefox were all just twinkles in
developers’ eyes when Debian 3.0
(aka Woody) was released.
A TRULY COMMUNAL PROJECT
The Debian Project is
1993 born when Purdue
University undergraduate Ian Murdock
announces on the comp.os.linux.
development newsgroup the “imminent
completion” of his Debian Linux
distribution release. Murdock pens the
Debian Manifesto, setting out the goals of
the Debian Project.
The first recognisably
1995 modern version of
Debian, Debian 0.93 release 5, is made
available. It is the first to use Ian
Jackson’s dpkg tool to install and
maintain packages. Meanwhile, the first
port of Debian to a non-x86 architecture
is started for the Motorola 68k family by
Harmut Kopstein.
54 LXF70 SEPTEMBER 2005
LXF70.feat_debian 54
The Debian Project is an
association of developers who have
come together with the purpose of
creating a high-quality operating
system built entirely from free software.
The name Debian combines the first
names of its founder, Ian Murdock, and
his girlfriend (now wife) Deborah. To
date, all official Debian releases have
been built on the Linux kernel
surrounded by an essential core of
tools from the GNU Project, and so
have been rightfully called Debian
Ian Murdock disappears
1996 from Debian
development due to his university
workload and leadership of the Debian
Project passes to project member and
Pixar programmer Bruce Perens. Under
Perens’s stewardship, Debian 1.1 (Buzz)
appears in June, and 1.2 (Rex) in
December. Ian Jackson writes the first
Debian Policy Manual spelling out how
Debian software should be packaged.
A busy year. Debian goes
1997 into orbit aboard the
Space Shuttle Columbia, running a
computer controlling a hydroponics
experiment; Debian 1.3 (Bo) is released;
and Perens drafts the Debian Social
Contract and the Debian Free Software
Guidelines, codifying the Debian Project’s
commitment to free software.
www.linuxformat.co.uk
GNU/Linux, although they’re generally
just known as Debian for short.
In terms of user numbers and
mindshare, Debian is by best estimates
the second most popular Linux
distribution today, behind Red Hat. It’s
also the largest, supporting 15,000
software packages on 11 fundamental
hardware architectures and many
more sub-architectures, ranging from
PDAs to mainframes. But Debian is
very different from its competitors.
Ian Jackson becomes
1998 Debian project leader in
January. Later, in July, Debian 2.0
(Hamm) is released for i386 and m68k
architectures. This is the first Debian
release to be based on the glibc2
C library rather than libc5. Scott K Ellis
starts the APT, or Advanced Package Tool,
project, which will become the famous
front-end to dpkg for package
management. Debian developers ratify
version 1.0 of the Debian Constitution,
which specifies the organisational and
decision-making policy within the Project.
The first Debian
1999 derivatives are
announced, as Corel and then Stormix
choose to use Debian as a base for
commercial distributions. Debian
developers vote on a new logo and
First, Debian is a non-commercial
entity, funded by the charity Software
In The Public Interest, and, in the main,
Debian developers are all volunteers.
Second, Debian is serious about
only using free software. Its promise to
free software and the needs of its
users are set out in the Debian Social
Contract, which explicitly states,
‘Debian will remain 100% free’. The
Debian Free Software Guidelines
(DFSG) list the conditions under which
software is considered to be free and
thus eligible to be part of the main
Debian distribution.
Incidentally, the DFSG was used as
a basis for the ‘Open Source Definition’
used by the Open Source Initiative to
sell the concept of free software to big
business. Community guidelines and
statements like this date back to
Debian’s inception, 12 years ago.
A new kind of distro
The Debian Project was announced to
the world in August 1993. At that time
none of the big commercial Linux
vendors of today – Red Hat, SUSE,
Mandriva – even existed, and the
distributions that were available were
generally of poor quality. Debian
founder Ian Murdock was scathing of
the SLS distribution popular at the
time. In the Debian Manifesto he wrote:
“It is quite possibly the most bug-ridden
and badly maintained Linux distribution
available; unfortunately, it is also quite
possibly the most popular.”
With the confidence that only the
very young and very old have,
Bruce Perens: the man behind
those codenames.
choose the swirl design still in use today.
Wichert Akkerman is elected Debian
project leader. Debian 2.1 (Slink) is
released in March, supporting the Alpha
and Sparc architectures and becoming
the first release to ship with APT.
11/7/05 11:01:27 am
 DEBIAN AT A CROSSROADS

Murdock, then a student at Purdue
University, Indiana, believed he could
do better. Fortunately, his hubris didn’t
stop him from recognising where the
other distributors were going wrong.
Distros such as SLS were typically
compiled by a single person or small
group and simply didn’t have enough
manpower to produce, test and
support a high-quality distribution. To
do the job properly, Murdock realised,
he’d need help from volunteers.
His second big realisation was that
he could borrow the distributed
development model that Linus
Torvalds used so successfully to
develop the Linux kernel – a model
later christened the Bazaar by the free
software community’s anthropologist,
Eric Raymond. The key to developing
an operating system in a distributed,
decentralised manner was modularity,
and in Murdock’s scheme the unit of
modularity was the software package.
Each volunteer was given
responsibility to separately maintain
one, two or a handful of packages.
Strict guidelines as to how packages
should be constructed ensured that
when combined the packages would
create a seamless whole, a cohesive
operating system.
It’s easy to forget how revolutionary
this idea was, because now all
distributions are built this way, but
Debian pioneered it. It took time for
Debian to develop and evolve the
tools and practices that make it what
is today. The very first Debian releases
“IT’S EASY TO FORGET HOW
REVOLUTIONARY IAN MURDOCK’S IDEA
OF MODULAR PACKAGES WAS.”
didn’t use the dpkg suite of tools for
packaging software. That and the
policy to go with it were created by
Debian developer Ian Jackson during
1994. Also, while Debian has always
been committed to using free software
– Murdock’s Manifesto criticises other
distributors for “a tendency to not
mention that Linux is free” – it wasn’t
until Murdock passed the leadership
baton to Bruce Perens that this
commitment was carved in stone in
the Debian Social Contract.
No matter how Debian has
changed throughout the years,
however, its goal has always been to
create and support a high-quality, bug-
free Linux distribution. In Murdock’s
words, again from his Manifesto: “The
primary purpose of the Debian project
is to finally create a distribution that
lives up to the Linux name. Debian is
being carefully and conscientiously put
together and will be maintained and
supported with
similar care.”
>>

CODENAME: TOY STORY

Since the release of Debian 1.1, all
Debian versions have been blessed with
an alias based on a character from the
film Toy Story. Thus Debian 1.1 was
called Buzz after Buzz Lightyear, 2.2
was called Potato in memory of Mr
Potato Head, 3.0 was Woody, named
after the cowboy star of the film, and so
on. Sid, always the name for Debian’s
unstable development tree, is from Sid
Philips, the unstable kid from next door.
So what’s the connection between
Debian and Toy Story? Well, former
Debian developer and project leader
Bruce Perens worked as a systems
programmer at Pixar, the maker of Toy
Story, until 1999. Pixar even hosted the
Debian mailing lists in the early days.

Debian 2.2 (Potato) is
2000 released, the first to
officially support PowerPC and ARM
architectures. A study shows that Debian
2.2 contains 55 million lines of source
code that would have cost $1.9 billion if
developed using traditional proprietary
methods. The first unofficial ISOs for
Debian GNU/Hurd are made available. It
is the first Debian distribution to use a
kernel other than the Linux kernel.
export laws, and Debian begins to Debian 3.0
support big iron as IA-64 and S/390
ports are started.
2002 (Woody)
eventually appears in July, having been
delayed by problems with boot floppies.
Meanwhile, Bdale Garbee is elected
Debian project leader and HP ships blade
servers with Debian pre-installed.
Lindows Inc ships its Debian-based
Windows replacement, LindowsOS, later
to be available pre-installed on $199 PCs
from Wal-Mart.
Debian developers vote
2004 to amend the Social
Contract so that ‘free’ applies to
documentation and firmware, not just
software proper. This and a slew of
problems throughout the year hold up
Sarge’s advance, but progress is made as
the first release candidates of the new,
modular Debian-Installer become
available after four years of
development. A new easy-to-use desktop
Linux based on Debian, Ubuntu Linux,

2001 A bad year for Debian

derivatives as Corel sells
its Linux division, Stormix declares
bankruptcy and Ian Murdock-founded
Progeny stops development of its
distribution. Ben Collins is elected Debian
project leader, cryptographic software is
moved from the non-US repository to the
main distribution due to relaxing of US
Now chair of Progeny, Ian Murdock
has stayed close to the project.
2003 The port to AMD64 is
begun, while the
Dreamix project brings Debian to the
Xbox. Other spin-offs appear including
SkoleLinux and MEPIS Linux, while Bruce
Perens outlines the strategy for a new
business-focused Debian derivative,
UserLinux. Martin Michlmayr is elected
project leader.
appears complete with a plan for an
aggressive six-monthly release schedule.
2005 Branden Robinson
becomes project leader.
Debian 3.1 (Sarge) finally ships, with the
new Debian-Installer and hardware
detection by default. An unofficial version
for the AMD64 architecture is released.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 55

LXF70.feat_debian 55 11/7/05 11:01:30 am

 DEBIAN AT A CROSSROADS
THE DEBIAN
ADVANTAGE
“DEBIAN IS A NON-COMMERCIAL ENTITY,
SO IT CAN’T GO BANKRUPT AND IT
CAN’T BE BOUGHT – IT’S A SAFE CHOICE
TO BET YOUR BUSINESS ON.”
DEGREES OF FREEDOM
The Debian package archive is split into
three principal sections: main, contrib
and non-free. Only packages with a
DFSG-compatible licence can be placed
in the main section; packages with
non-DFSG-compatible licences go into
non-free. Additionally, packages in main
cannot require any non-free packages to
build or to run; free software that does
must be placed in the contrib section
(for example, open-source Java software
that uses the non-free Swing toolkit and
free emulators that require non-free
firmware must go in contrib).
56 LXF70 SEPTEMBER 2005
LXF70.feat_debian 56
Debian has some high-profile
adherents. HP offers solutions and
support based on Debian, the City of
Munich has rolled out Debian to its
14,000 desktop users, and the
government of the Extremadura region
in Spain has created its own Debian-
based distribution, Linex, for use in its
offices and schools.
In fact, Debian is becoming
increasingly popular as a base for
creating custom distributions.
Commercial Linux vendors Xandros
and Linspire both derive their easy-to-
use variations of Linux from Debian. A
recent Slashdot article even claimed
that there are now more Debian-
derived distros than there are those
based on Red Hat. Not only are there
more, but according to distrowatch.
com they are more popular than the
Red Hat-based distros.
What is it that attracts the millions
of users worldwide that it now boasts?
And why is Debian – a community-
developed distribution – so popular
with business and governments?
To aid package management,
packages are classified further into
‘subsection’ and ‘priority’.
Subsections group packages according
to the problem-space they address – so
there are subsections named mail, web,
math, science, games, and so on.
Priority groups packages according to
how necessary they are to create a
functioning Debian system – for
example, ‘required’ packages are
absolutely necessary, and ‘important’
packages are those you’d typically find
on any Unix system.
www.linuxformat.co.uk
The Debconf system provides a consistent means for Debian packages to
prompt the user during post-install configuration.
Actually, the affinity of business with
Debian isn’t that hard to explain.
Debian is attractive for many of the
same reasons that Linux and open-
source software in general are
attractive. Beyond any technical merits,
the key advantages of using Debian are
freedom and the strength of the
community that develops it. Debian has
been around for 12 years and it’s not
going to go away any time soon: not
unless all 1,000 of its developers
decide they’ve had enough and take
up gardening instead. Debian is a non-
commercial entity, so it cannot go
bankrupt and it cannot be bought.
Safety in numbers
Thus Debian is a safe choice to bet
your business on. As Ian Murdock
wrote on www.ianmurdock.com,
“This isn’t about technology at all… It’s
about freedom. It’s about controlling
your own destiny, building a future for
yourself that doesn’t bind you to the
profit motive of a single American
company.” This freedom applies as
much to individual users as to the
world of business. In a piece called
‘Why Linux? Why Debian?’ written this
summer, Debian developer Manoj
Srivastava exclaimed: “I do not want
my OS to be held hostage to anyone’s
business plan!”
More than this, Debian is serious
about using only free software, and its
definition of what ‘free’ means is more
rigorous than most. Only software
whose licence complies with the
Debian Free Software Guidelines
(DFSG) can be part of the ‘main’
Debian package archive (see Degrees
Of Freedom box, below). The DFSG
only permits licences that protect
users’ freedom to use, modify and
redistribute software. It doesn’t require
that a licence enforce a copyleft,
however. Thus the GPL, the X11
licence and BSD licences are all
compatible with the DFSG.
Last year, Debian developers even
voted to strengthen this commitment
to freedom and apply the guidelines
beyond software – for instance, free
documentation with no restrictions on
modification is now required too.
The first guarantee of this
amended Social Contract talks about
free ‘works’, not simply free ‘software’:
‘We provide the guidelines that we use
to determine if a work is “free” in the
document entitled “The Debian Free
Software Guidelines”. We promise that
the Debian system and all its
components will be free according to
these guidelines. We will support
people who create or use both free
and non-free works on Debian. We will
never make the system require the use
of a non-free component.’
Thanks to the DFSG, businesses
can be confident that as long as they
use the ‘main’ Debian packages, they
are using free software, unencumbered
by any licence issues, and unlikely to
meet legal repercussions.
People often attribute Debian’s
popularity merely to its APT package
11/7/05 11:01:33 am

NOT JUST LINUX
Debian offers more than just a slick
package manager. It has an experienced
community, a rigorous software
packaging policy and the tools to back
that up. So it’s only natural that these
attributes be applied to tasks other than
building Linux distributions. Thus various
official Debian projects have sprung up
to create operating systems based on
kernels other than Linux. The first such
project was Debian GNU/Hurd. This
replaces Linux at the heart of the
Debian system with the GNU Project’s
manager. This (or, more commonly, a
graphical front-end to APT, such as
Synaptic) is what lets you simply fetch
new or updated software – and,
crucially, any software required to use it
– and install them on your machine.
APT handles dependencies for you
automatically, and thus takes the pain
out of installing software. APT is
Debian’s ‘killer app’, they say. Is there
any truth to this? In short, no.
Beyond APT
While APT is an excellent tool, it is not
APT alone that makes Debian great. In
fact, Conectiva (now merged with
Mandrake as Mandriva) ported APT for
use with its RPM-based distro four
years ago, and it just doesn’t work as
well there. Other APT-like solutions
such as urpm and yum are available,
but again fare no better. So perhaps
RPM is inferior to Debian’s APT tool,
dpkg. Indeed, Mandriva developer
Claudio Matsuoka blogged (online at
http://tweek.dyndns.org:8080) on
the “top ten problems in RPM” and
complained about “the non-uniform
handling of file dependencies and the
large number of hidden dependencies
automatically added by RPM”.
Certainly, dpkg is more flexible than
RPM. For example, dpkg allows a richer
relationship between packages to be
specified – packages don’t simply
depend on one another. A package
can ‘pre-depend’ on another, if that
package must be installed and
configured for it to be installed; a
package can ‘recommend’ another as
a non-absolute, weaker dependency;
or a package can ‘suggest’ another, if
that package can enhance its
functionality, but is not required.
Dpkg also permits the use of post-
install scripts, which can perform all
manner of (possibly interactive) per-
package configuration. Debian provides
LXF70.feat_debian 57
microkernel-based Hurd kernel, fulfilling
GNU’s aim of creating a Unix-like OS
based entirely on GNU software. More
recently, other controversial Debian
ports have been initiated using NetBSD
and FreeBSD kernels. A port to
Windows, using Cygwin’s POSIX layer,
was under way at one point, but activity
on that seems to have fizzled out now.
Outside of the official projects, the
Debian packaging tools have been used
in the Fink project, which brings Unix
open-source software to Mac OS X.
the debconf system as a standard
means for packages to prompt the
user for information with dialogs
during such post-install configuration.
What’s more, dpkg supports multiple
interface types (text-only or GUI-
based using GTK or Qt toolkits) and
can be configured to ask the user
different types of question based on
priority. For unattended installs,
debconf can be told not to prompt at
all and packages will then be
configured with whatever the defaults
are for the questions that would have
been asked.
Actually, neither RPM nor dpkg is
perfect, although dpkg is more
versatile and more robust. The main
reason why package handling is so
much better in Debian-based
distributions than it is in RPM-based
distros is Debian’s comprehensive
policy on how software should be
Corel Linux
Demo Linux
Xandros
Morphix
Overclockix
Whoppix
Gnoppix
www.linuxformat.co.uk
DEBIAN AT A CROSSROADS
packaged. It is Debian’s policy
and the tools that support it
that are the distro’s ‘killer
app’. And, critically, a stable
Debian release will never ship
with packages that don’t match
Debian Policy. This strong policy
and the power of dpkg enable a basic
Debian install to be much smaller and
lighter than RPM-based distros. In
particular, Debian policy mandates that
shared libraries should be packaged by
themselves (not with binaries that
depend on them) and have a package
name that reflects the ABI (Application
Binary Interface) version of the library.
This does much to lessen ‘DLL hell’,
the pain caused when it’s impossible
to install different version of the same
library due to conflicts. With Debian
this is simply not an issue.
Technical advantages such as these
– along with the guarantees of
freedom that Debian provides – make
Debian perfect for customising and a
>>
popular distro on which to build your
own derivative.
The strengths of Debian make it an
ideal basis for rolling your own
distro, as the large number of
Debian derivatives proves.
Debian Linux
Progeny Debian
Knoppix Storm Linux Libranet Linux
Lindows OS
MEPIS Linux Linex Damn Small Linux
Kanotix Skole Linux
Guada Linex Linspire
Ubuntu Linux
User Linux
Kubuntu Linux
LXF70 SEPTEMBER 2005 57
11/7/05 11:01:34 am
 DEBIAN AT A CROSSROADS

TROUBLE IN
both of these problems, actual and
potential, Debian will be irrelevant by
the time Etch is out.”
So why did Sarge take so long?
There were no great technical
Steve Langasek, a member of the
release team for Sarge, wrote in an
email that has come to be known as
the Vancouver Prospectus that “it is
not sustainable to continue making

FREETOWN
The fact that Debian now has
glacially slow release cycles – Sarge
took three years and Woody took two
before that – is now firmly entrenched
in the open-source psyche.
The question is, why did Sarge take
so long, and how can Debian
developers learn from that to ensure
that Etch ships in a timely manner?
First, does it even matter how long
Etch takes? Debian is under no
commercial pressure so can release
when the code is in a fit state for
release, not when market forces
100,000
10,000
1,000
100
10
Packages
Developers
1
95
transitions between Woody and Sarge.
Introducing the new Debian-Installer
was the only large change, and that
has been in development for years. A
naive conclusion might be that Debian
has simply grown too big and the
release processes have failed to scale
with this growth. Earlier versions of
Debian, when there were many fewer
packages to deal with, were developed
and released much more quickly.
It is evident, however, that the
number of packages in the Debian
archive is not the sole issue. Sarge has
15,000 packages maintained by
around 1,000 developers, and this is a
better developer-to-package ratio than
either Fedora or Mandriva can field.
The tardiness of Sarge was not simply
due to package count.
Size isn’t everything
What’s more of a problem is the
number of architectures that Debian
supports. Debian developers put a lot
of effort into porting Debian packages
to the ‘minor’ architectures. For
instance, typically the upstream
developers of a package won’t
explicitly support ARM, HP-PA, MIPS
or m68k CPUs. If there are problems
in porting to these architectures, it’s up
to the Debian developers to correct
and fix them. In particular, stabilising
the kernel and the installer and getting
the infrastructure up and running to
support the Sarge release with security
updates were all factors that delayed
coordinated releases for as many
architectures as Sarge currently
contains… keeping 11 architectures in a
releasable state has been a major
source of work for the release team,
the d-i [Debian-Installer] team, and the
kernel team over the past year.”
This Vancouver Prospectus
advocates reducing the number of
architectures that Debian supports and
has sparked some controversy, partly
due to confusion and partly due to the
way it was presented as a fait accompli.
In fact it hasn’t yet been officially
adopted as Debian policy. According
to newly-elected DPL Branden
Robinson, “this is the beginning of a
process to figure out where we’re going
to, not the end of it.”
The confusion stems from a prior
proposal to split the supported
architectures for purposes of reducing
the bandwidth required to mirror the
Debian repository. There would be
primary architectures – likely to be
x86, AMD64, IA-64 and PowerPC –
and so-called Second-Class Citizen
(SCC) architectures, which have far
fewer users. The SCC proposal will be
enacted, since it’s a requirement to
provide enough bandwidth to add
additional ports such as the AMD64.
But the Vancouver Prospectus,
which essentially says that there won’t
even be releases made for the SCC
architectures, is still being discussed. It
states the conditions under which an
architecture will be considered for

96

97

98

99

00

01

02

03

04

05
n
n
n

n
n

Sarge’s release. release; and among these are that

Ja
Ja
Ja

Ja

Ja
Ja

Ja

Ja

Ja

Ja

Ja

The number of packages in Debian versions might have grown

dramatically, but so has the team of developers.
dictate. Surely users who want more
up-to-date software can use the
‘testing’ or ‘unstable’ trees or
one of the numerous Debian
derivatives such as Ubuntu?
One person who believes in
the importance of getting Etch
out the door at a predictable time
is Ian Murdock. In a blog posted on
his website shortly after Sarge’s
release, he wrote that Debian should
make a timed release cycle a priority,
as well as maintaining its position at
the heart of its growing family of
derivatives. “What’s at stake? Bottom
line: if we don’t do something about
TESTING, TESTING, TESTING
The Debian project maintains three trees
of packages, named ‘stable’, ‘testing’
and ‘unstable’. The stable tree
corresponds to the current stable
release of Debian (currently version 3.1,
aka Sarge) and this doesn’t change until
a new revision is released. Security
updates for the stable release are made
via a separate APT repository at
http://security.debian.org.
Unstable is the testing ground for the
next Debian release. New packages are
uploaded to the unstable repository and
after a certain period – if they don’t
accrue any additional release-candidate
bugs and they don’t depend on any
packages not already in testing – they
are automatically moved to the testing
repository. The idea behind the testing
tree is that it is always close to release.
End users can choose to install packages
from testing or unstable if they want (or
indeed mix and match packages from
stable, testing and unstable trees).
Using testing should be relatively safe.
There will be few show-stopping
problems with testing, although be
warned that it is not supported by
security updates from the Debian
security team as the stable tree is, so it
definitely shouldn’t be used for mission-
critical applications. Only the really
adventurous who enjoy living on the
bleeding edge should use the unstable
tree, since from time to time some
unstable update may break your system.

58 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.feat_debian 58 11/7/05 11:01:36 am


Mark Shuttleworth, father of Ubuntu: will the African Buzz Lightyear’s distro help or harm Debian?
hardware must be available to buy
new, that 98% of the non-
architecture-specific source packages
must build on that architecture, that
the Debian security team must be
willing to support it, and so on.
While some measure to increase
the manageability of the release
process seems to be necessary, it will
be a sad day for Linux if Debian no
longer supports the smaller
architectures, such as ARM and MIPS.
Other measures to speed up the
release process centre on better
communication and spreading the
burden. A lone release manager has
been replaced by a release team, and
Branden Robinson has been elected
as the head of a leadership team,
codenamed Project SCUD (another
Toy Story reference).
Unite and conquer
The late release of Sarge had an affect
of increasing the divide between the
stable version of Debian and those
distros based on Debian, such as
Ubuntu. Now that Sarge is here – and
thus a stable version of Debian with
relatively current software – the gap
between Debian and its progeny has
been narrowed.
LXF70.feat_debian 59
Thanks to Debian’s modularity, the
degree of compatibility between Sarge
and its derivatives (as well as Ubuntu,
Linspire and Xandros are well known)
is impressively high – especially when
you compare how incompatible Red
Hat and its spin-offs are. Still, many
say that Debian and derivatives should
“THERE IS TENSION BETWEEN THE
DEBIAN PROJECT AND ITS OFFSPRING,
PARTICULARLY UBUNTU.”
work much more closely. Ian Murdock
wrote: “We need to tap into Debian’s
unique asset, namely the collective
power of the Debian derivatives. On
their own, the derivatives aren’t
significant players; but, taken as a
group, they dwarf the individual
leaders (Red Hat and Novell).”
There is tension between the
Debian Project and its offspring
though, particularly the recent off-
shoot Ubuntu. Debian developers have
debated whether Ubuntu is actually a
derivative or a more damaging fork,
and whether Ubuntu developers do
enough to feed their changes back to
www.linuxformat.co.uk
DEBIAN AT A CROSSROADS
THE RISE OF
UBUNTU LINUX
The most popular Debian derivative is
also one of the most recent: Ubuntu
Linux, founded and funded by
ex-Debian developer, internet
entrepreneur and space tourist Mark
Shuttleworth. Shuttleworth is a native
of South Africa and ubuntu is an
African word meaning ‘humanity to
others’. Thus Ubuntu Linux bills itself as
‘Linux for Human Beings’ and is a user-
friendly desktop-oriented distribution
based on Debian.
The Ubuntu project echoes many of
the same commitments to freedom
that Debian does. You can even get an
install disc mailed to you anywhere in
the world free of charge, which is
perhaps one reason for its rocket-like
take off. Ubuntu also promises a
regular six-monthly release cycle,
another reason for its popularity.
Ubuntu is currently available for
x86, AMD64 and PowerPC
architectures, both in regular
installable and Live CD editions. Sparc
and IA-64 are supported as minor,
unreleased architectures.
■ Read Mark Shuttleworth’s thoughts
on Debian, Ubuntu and the
fork/derivative argument in the next
issue of Linux Format.
Debian package maintainers. Robinson
had this to say in a recent interview:
“Ubuntu is a mixed blessing, but one
with a net positive. It has helped raise
Debian’s profile in a mostly positive
way… The downside is that Ubuntu has
prompted some people to come to
doubt Debian’s ‘relevance’. As a
consumer of Debian’s OS with a
derivative product, I think Canonical
[Ubuntu’s sponsor] could do more to
emphasise just how much of a service
Debian does provide. If Debian were to
vanish tomorrow, Canonical would be
in a bit of a bind.”
Clearly better communication
between Debian and its derivatives is
needed. The various projects will
always have different goals, but
ensuring that Debian releases are
predictable will provide a solid, stable
basis for others to build on and thus
lessen the risk of incompatible >>
derivatives and forks.
LXF70 SEPTEMBER 2005 59
11/7/05 11:01:37 am
 DEBIAN AT A CROSSROADS
LOOKING
FORWARD
TO ETCH
“RELEASING ETCH WITHIN 18 MONTHS
WOULD DO MUCH TO REPAIR THE
DAMAGE CAUSED BY SARGE’S DELAY.”
THE UNFREE DOCUMENTATION LICENSE
The GNU Free Documentation License
(FDL), the Free Software Foundation’s
complement to the GPL for covering free
software docs, is considered by many to
be inconsistent and by others to be non-
free. The crucial problem is with
‘invariant sections’ that the FDL allows.
An invariant section cannot be modified
or removed – even when it becomes
obsolete – and so is a barrier to freely
60 LXF70 SEPTEMBER 2005
LXF70.feat_debian 60
Sarge is here and the dust is
beginning to settle, so it’s an
appropriate time to start thinking about
what will eventually become the next
stable Debian release, codenamed Etch
(named after the Etch A Sketch from
Toy Story). Although Etch isn’t even
three weeks old at the time of writing,
we still know a lot of the changes that
will be introduced during its
development. This is because many of
Documentation licensed under the GNU FDL cannot be included with Etch,
and already some FDL docs have been removed.
distributing modified versions of the text.
Debian developers have long discussed
the problems with the FDL, but have
finally voted on the issue and decided
that documentation licensed under the
FDL cannot be included in the main
section of the Debian repository. While
the effect of this decision has been
postponed so that Sarge could be
released, it will apply to Etch. Numerous
www.linuxformat.co.uk
these are issues that were known
about before but postponed to get
Sarge out the door.
Key transitions
First of all there are some software
transitions to overcome. This includes
updating the default Debian toolchain
to either GCC 3.4 or 4.0, which
involves a change in the C++ ABI,
causing backwards compatibility of
C++ shared libraries to break. Thus all
the C++ apps and libraries will have to
be rebuilt. This shouldn’t prove to be a
big problem, and plans are already in
place to handle this, but it will take
time to happen and for the results to
trickle through to the testing tree.
According to developer Bill
Allombert, the reliance on C++ in the
APT and Aptitude tools is one of three
main areas of Sarge that need work,
along with circular dependencies and
packages that “mess with conf files
causing useless dpkg conf file
handling”. In an email to the Debian
community Allombert said these
Debian packages contain documentation
licensed under the FDL. For each one,
either the offending text will have to be
relicensed under a new, DFSG-
compatible licence or the documentation
removed from the main section.
Some work has already been done,
notably a few Debian maintainers
removing FDL-licensed documentation
from their packages for Sarge.
problems were exacerbated by the
short time given to testing Sarge after
the feature freeze. “This needs to be
done automatically,” he wrote, “… [in]
roughly the same way as builds work.”
Another transition long overdue is
to replace the ageing Xfree86 4.3.0
server with the X.org server. Again this
shouldn’t cause problems, and the
Debian X Strike Force have been
working on packaging X.org for some
time. Other software updates currently
targeted for Etch include Gnome 2.10,
KDE 3.4 and OpenOffice.org 2.0.
Secondly, the AMD64 port will
definitely join the list of supported
release architectures in Etch, but as we
discussed earlier, it’s possible that the
overall number of release architectures
will be greatly reduced. Related to
supporting architectures such as the
AMD64, which provides multiple
execution environments (32-bit and
64-bit in this case), proposals on how
to fully support and manage multi-
architectural run times in Debian are
being examined (see Multiarch
Proposals box, right) and some testing
has already been done.
It’s too early to say whether these
changes will be implemented in Etch,
but if they are it will be a huge boon
for backwards compatibility and for
users who do cross-development.
More freedom
Thirdly, changes are required because
of the decision to strengthen the
Debian Free Software Guidelines. In
future they’ll cover not only software,
but supplementary material such as
documentation and firmware as well.
These changes were postponed to
allow Sarge to be released, but they
take effect now. One particular
problem area is the GNU Free
Documentation License, which has
been deemed to be non-free. This will
require a lot of documentation to be
re-licensed or removed from the main
section of Debian. Also, drivers that
require binary only firmware must be
removed from the kernel.
We can reasonably expect the
Debian task system and hardware
detection to improve with Etch. The
task system provides a way of speeding
up installs by allowing users to select
and install sets of packages rather
than having to select packages
individually. Sarge provided only a
handful of such tasks, so this is one
area Etch can definitely improve on.
11/7/05 11:01:39 am

Already a laptop task has been
proposed which will automatically
install power management tools
suitable for laptop use. The base
Debian install in Etch will be slimmer
and no longer require the C compiler
to be installed, so a C development
task will be necessary.
One suggestion for hardware
detection is the ability to detect and
react to changes in hardware
configuration (such as when an
installed graphics card or network card
is replaced). Debian should
automatically offer the user the ability
to reconfigure X or the network
settings. Such a feature has been
available in Kudzu, the Red Hat
hardware detection system, for aeons.
Security changes
Etch will also improve on security. One
change is the incorporation of
APT 0.6.0, which supports
cryptographically verifying the origin of
a package. This is so that when you
download a package from the net you
can be sure that it really is a Debian
package and hasn’t been compromised
during transfer. APT 0.6.0 has been in
the experimental Debian repository for
some time and is already in use in
Ubuntu, but the infrastructure changes
needed to roll it out in the Etch testing
tree are not yet in place.
Still on the security theme, full
support for the SELinux (Security-
Enhanced Linux) project is a oft-
requested feature for Etch. The
SELinux project was created by the US
National Security Agency and adds
support for mandatory access controls
to Linux. This allows for the fine-
The revamped task system that appeared with Sarge will be expanded
with more tasks in Etch to further speed up installation.
LXF70.feat_debian 61
grained configuration of security policy.
The libselinux library and associated
utilities are already part of the Debian
repository, but other components of
the SELinux project – such as the
patched versions of the standard file
manipulation tools (the coreutils
package) which supports access
controls, patched init and daemons –
are not.
The Debian release team have
pencilled in December 2006 as a
target date for Etch’s release, giving
them 18 months of development time.
Right now this deadline looks
achievable. Of course, this will mean
achieving certain milestones along the
way as the various transitions required
are performed, and we’ll have a better
idea of whether Etch will be released
on time as work progresses.
Critics warn that Debian is in
danger of becoming irrelevant if its
developers cannot get the release
processes in shape and get Etch out
on time. This is overstating the case,
but it’s clear that releasing Etch in
18 months would do much to repair
the damage caused by Sarge’s
interminable delay.
But as a reward for getting Sarge
out we think the Debian developers
deserve a little rest and recuperation,
and need to marshal their
strength for the long
road to Etch. LXF
www.linuxformat.co.uk
DEBIAN AT A CROSSROADS
MULTIARCH PROPOSAL
Many modern CPU architectures support
multiple ABIs (application binary
interfaces). The AMD64 can execute
IA32 or AMD64 code for example, the
IA-64 provides compatibility for IA32
and HPPA code, and the PowerPC even
supports big- and little-endian code. The
problem is that binaries require shared
libraries with precisely the same ABI,
and there’s only one standard place to
install libraries. To be able to execute
IA32 code binaries on the AMD64, you
need 32-bit versions of whatever shared
libraries that those binaries use, such as
the C library, X libraries and so on. How
are distros going to handle that?
The current FHS (File-Hierarchy
Standard) and LSB (Linux Standard
Base) have little to say about this issue,
so practice varies among distributions
and CPUs. For processors that can
execute 32-bit and 64-bit code, the
32-bit libraries wil typically be installed
in /lib, /usr/lib, and so on, while 64-bit
libraries will be in /lib64, /usr/lib64, etc.
The AMD64 is different. 64-bit code is
preferred, and so 64-bit libs will be in
/lib64, although Debian on AMD64
stores 32-bit libs in /emul/ia32-linux/.
The practice of using /lib, /lib64, and
so on is just too inflexible. What about
the IA-64, which can run IA32 and
HPPA code as well as its own instruction
set? What about big-endian CPUs?
What about ABIs from different
operating systems? And what about
software emulation provided by systems
such as Qemu?
The proposed solution is to specify
the ABI (CPU and OS) as a suffix to the
lib path. For example, you would have
/lib/ia32-linux, /usr/lib/ppc64-darwin7.0
to uniquely identify where the correct
libraries can be found. If necessary,
paths of the form /lib32 and /lib64
can be provided as symlinks for
backwards-compatibility.
The multiarch proposal for Debian
goes further and addresses the problem
of how to manage software for non-
native ABIs with the Debian packaging
system. Rather than force the user to
manually install and maintain the 32-bit
environment on an AMD64, say, why not
let them install libraries packaged for
the Debian i386 port? This is a cheaper
proposal than repackaging the libs
separately for native and emulated use,
but requires some changes to dpkg to
handle specifying the architecture in a
dependency. Happily, such changes have
already been implemented and tested.
LXF70 SEPTEMBER 2005 61
11/7/05 11:01:39 am
 BEAGLE INSTALLATION GUIDE

Go fetch!
ST
STEPEGP BY
U
install B IDE:
e
on Man agle
driva
Linux

It’ll sniff out documents and emails you thought you’d lost forever: is
Beagle man’s new best friend? Get it running here with Graham Morrison.

B
eagle, along with other
desktop search tools, is
threatening something of a
computing revolution. It’s such
a simple idea: enter a few
words or phrases into a search
window and Beagle will respond
immediately with a list of links that
contain your search term. Sometimes
these links are files, but the clever part
is that they can also be links to
conversations, emails, images and
music, and each one is located
somewhere in your home directory.
It’s similar to the way Google’s
Gmail service works. This offers
gigabytes of storage, and encourages
you to never delete a single email.
How are you expected to organise all
this data? Well, Google would rather
you just left it as it is. Gmail’s strength
lies in using Google’s redoubtable
HTML, popular image and music
formats, Gaim instant messages and
many, many others. Once you get
used to using Beagle, traversing vast
hierarchies of folders to get to a single
file will start to feel arduous and
antiquated. For more about the
technology and development history,
visit the Beagle homepage at
http://beaglewiki.org/Main_Page.
As we mentioned in our LXF67
review of Mandriva Linux Limited
Edition 2005, we were dismayed not
to find Beagle included in the copious
package list. Obviously, the people at
Mandriva were far too busy at the time
to include packages as wayward as
Beagle, so we’ve
decided to
cover each
of the
steps

“ONCE YOU GET

USED TO USING
BEAGLE, YOUR OLD
METHODS WILL FEEL
ARDUOUS AND
ANTIQUATED.”
searching algorithms to instantly
provide you with a list of emails that
match your search criteria. For
example, you may want to check an
online order: simply enter ‘compact
disc’ and ‘britney spears’ into the
search field, and any emails matching
the target will be listed immediately.
It’s this kind of integration that Beagle
aims to bring to your desktop.
But Beagle has one advantage
over Gmail: it’s not just your email that
can be queried, but a whole host of
Gnome-centric applications, including
OpenOffice.org documents, PDFs,
involved in installing Beagle
on to a Mandriva system. If
you use another distribution,
many of the stages will be
identical – just replace Mandriva’s
package manager with that of your
own distro, and things should work.
The main problem with getting
Beagle to work is Mandriva’s
sometimes bizarre naming and
dependency problems. Installing
Beagle isn’t for the faint of heart; it
can be rather stressful. It has dozens
of dependencies, requires a
compatible kernel and the manual
installation of some large
applications. Having said all
that, we’ve tried to remove
most of the pain from the
process and we’re going to hold
your hand through each step, so
it doesn’t require any specialist
knowledge. Just keep clear of the
moors and beware the moon!

62 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.feat_beagle 62 11/7/05 12:48:10 pm


STAGE 1 – LOW-LEVEL CONFIGURATION
Beagle actively watches for
changes in your filesystem. You might
expect this to consume a significant
chunk of your system’s resources, but it
doesn’t, becuase Beagle uses a kernel-
level event-monitoring system called
Inotify. This is included within many
modern kernels, but the problem is
that Beagle needs a specific version of
Inotify to be able to work. As we’re
going to install Beagle version 0.0.11.1,
released on June 15, we will need a
kernel that contains Inotify 0.23.
Earlier versions of Beagle
need earlier versions
of Inotify.
LXF70.feat_beagle 63
Compiling your own kernel is tough,
so it’s far easier to download one that’s
precompiled and nicely packaged for
Mandriva. The kernel we’re after is
called kernel-multimedia-2.6.11-9.mm,
and you can find it online by entering
the name into the search field at
http://rpm.pbone.net. Once you’ve
got hold of the file, install it by typing
the following into a console as root:
urpmi kernel-multimedia-2.6.11-
9.mm.4mdk-1-1mdk.i586.rpm
Urpmi also makes changes to your
system boot menu, allowing you to
select the newly-installed kernel at
boot time. It should also add the
necessary option for enabling Inotify,
which you can check by looking for
inotify=on in the append field of
/etc/lilo.conf. A simple reboot to this
kernel should now provide you with a
working Inotify system, which should
create a device node in your file
system located at /dev/inotify.
File secrets
There’s one other morsel of
low-level configuration that
Beagle relies upon, and that’s
enabling extended attributes
for your local filesystem.
Extended attributes are a
neat way of attaching
information to a file from
within the filesystem
structure itself, as opposed
to a database or within the
file. It could be something
as mundane as the file type
or when the file was last
accessed. Beagle uses
extended attributes to
transparently keep track of files
and their contents. Most modern
filesystems support extended
attributes, with a couple of
exceptions: Reiser4 and NFS.
Reiser4 implements its own extended
attribute interface, while NFS doesn’t
support them at all. So for now, both
are off limits for Beagle users.
To enable extended attributes, you
simply need to add user_xattr to
each partition in your /etc/fstab file.
As Beagle is intended to be run as a
user, you only really need to add this to
your home partition. Otherwise, if you
use a single partition, you can just
enable it for / (root). Your /etc/fstab
partition entry should then look
www.linuxformat.co.uk
BEAGLE INSTALLATION GUIDE
INSTALLING THE EASYURPMI WAY
The advantage with installing by hand is
that you get the latest release, and have
more control over how Beagle works on
your system. The flipside is that it can
take hours to get it to work. Mandriva
does offer a compromise though, using
its Contributions repository – this allows
you to install the whole shooting match
with a single command.
‘Contribs’, as it is affectionately
known, is a repository for the packages
that don’t make it into the final release.
They’re generally not as stable as their
disc-bound counterparts, and installing
software from the Contribs repository
can compromise your system’s stability.
But you do get to play with some of the
more avant-garde applications of the
Linux world, Beagle being one of them.
The easiest way to set up Contribs as
a source is to use a website called
EasyURPMI (http://easyurpmi.zarb.org),
which configures the various repositories
available to Mandriva users.
To add the Contributions repository,
first make sure that your system is the
one automatically selected (it should
read 2005, for 2005 LE). Then you
need to select a local server for Contribs
something like this:
/dev/hda7 /home ext3 noatime,user_
xattr 1 2
You can quickly remount the
affected partition without having to
reboot your machine with
mount -o remount /home
Dependency hell
The next step is to install the Mandriva
packages that Beagle depends upon.
You can install these packages from
the rpmdrake utility, or using urpmi
from the shell, logged in as root.
Before delving into the separate
packages, first ensure that all the
Gnome Workstation packages are
installed. You can do this from
Mandriva’s Control Center by selecting
Software Management, followed by
Look At Installable Software Packages
and selecting Gnome Workstation.
Clicking on Install should take you
through the long process of installing
the dozens of files that you’ll need.
“DOWNLOAD A KERNEL
THAT’S NICELY PACKAGED
FOR MANDRIVA.”
Installation couldn’t be any easier
with EasyURPMI.
in the Core Distribution section. Then
you just need to click on the Proceed To
Step 3 button, and the corresponding
urpmi command is generated
automatically. This needs to be pasted
into a console and executed as root.
After urpmi has finished downloading
the package list from the server,
installing Beagle is as simple as:
urpmi beagle
Next up is a list of specific
dependencies. Each one of these
needs to be installed, and will in turn
install many other related packages
automatically, depending on your
system’s configuration. Install them via
Mandriva’s Control Center, or on the
command line with urpmi.
■ bison
■ libglib2.0_0-devel
■ gcc-c++
■ libxml2-devel
■ libexif12-devel
■ libwv-1.0_2-devel
■ gnome-vfs-devel
■ mozilla-firefox-devel
■ libxorg-x11-devel
■ libpango1.0_0-devel
■ libgtk+2.0_0-devel
■ libsqlite0-devel
■ libgnome-vfs2_0-devel
■ libgnomecanvas2_0-devel
■ libgnomeui2_0-devel
■ libgnomeprintui2-2_0-devel
■ gtkhtml-devel
>>
LXF70 SEPTEMBER 2005 63
11/7/05 12:48:11 pm
 BEAGLE INSTALLATION GUIDE
STAGE 2 – ROLLING YOUR OWN
>> We have used as many readymade
Mandriva packages as we can, but
there are five that we need to
download, compile and install:
■ Mono 1.0.6 or newer Novell’s
implementation of Microsoft’s .NET
framework. http://mono-project.com
■ gtk-sharp 1.0.x A set of GTK GUI
.NET bindings for graphical interfaces.
http://mono-project.com/Downloads
■ gecko-sharp 0.6 A wrapper to
embed the Mozilla rendering engine
into GTK# applications. www.go-
mono.com/packagers/gecko-sharp/
gecko-sharp-0.6.tar.gz
■ gmime-2.1.13 A mime message
parser for reading all your emails.
http://spruce.sourceforge.net/
gmime
■ beagle-0.0.11.1 The star of the
show. We’ve used version 0.0.11.1.
http://ftp.gnome.org/pub/GNOME/
sources/beagle/0.0/
Of those five, it’s Mono that’s by far
the biggest and takes the longest to
obtain, so that’s a good place to start.
The download weighs in at close to
15MB, but can be easily grabbed from
http://go-mono.com or by using the
following wget command:
wget http://go-mono.com/sources/
mono-1.1/mono-1.1.8.tar.gz
Compiling Mono should be
straightforward, if a little time-
consuming. We first need to un-tar the
archive, run the configure script so
that the compilation can be optimised
for your system, and finally run the
make command, which should go off
and build the application. This should
take around 30 minutes on a modern
machine – just enough time to filter
some water, grind a little coffee and
make a refreshing beverage.
In our examples, we always use the
/usr prefix within the configure
command, as Mandriva uses this
location specifically. You may need to
change this to reflect your own
distribution’s filesystem.
tar xvzf mono-1.1.8.tar.gz
cd mono-1.1.8
./configure --prefix=/usr
make
sudo make install
Mono provides the development
platform for Beagle, and the other
components we need to install are all
related to augmenting the .NET
compatible framework that Mono
64 LXF70 SEPTEMBER 2005
LXF70.feat_beagle 64
configure scripts, you
just need to enter
make and make
install, as with Mono’s
installation.
Apart from Beagle
itself, the two
remaining packages
that need to be
downloaded and
installed (gecko-sharp
and gmime-2.11 . 3)
shouldn’t pose any
particular problems.
Install all the necessary Beagle packages using Just download the files,
Mandriva’s package manager. un-tar them, run
./configure --prefix=/
provides. The first package we need is usr from the package directory, and
gtk-sharp-1.0, which is mainly used for make then make install as root to
the GUI, and it’s important that it’s the install each package. If all has gone
older 1.0 version rather than the more
current 1.9 release that we use. In
common with all the packages that we
need to build, there are four stages to
“YOU CAN EVEN QUERY
each installation. These are: BEAGLE WITH YOUR
un-tar the archive, configure the build
process, build the application and, WEB BROWSER.”
finally, install.
tar xvzf gtk-sharp-1.0.10.tar.gz well, this should leave us with one
cd gtk-sharp-1.0.10 more package to manually install, and
./configure --prefix=/usr that’s Beagle.
There is one difference with gtk-
sharp, and that’s the output from the Installing Beagle
configure script. This output shows you This stage is actually pretty
which gtk-sharp components are going straightforward as long as all pre-
to be installed. If you’ve got each of requisite packages have been installed
the necessary development packages, first. After you’ve downloaded the tar
you should see something like this: file, un-compressed it and changed to
Optional assemblies included in the the newly-created directory, you may
build: be tempted to go straight ahead and
* art-sharp.dll: yes run the ./configure --prefix=/usr
* gnome-sharp.dll: yes command. If you do, you will be
* glade-sharp.dll: yes presented with an error that will look
* gda-sharp.dll: no something like:
* gnomedb-sharp.dll: no ‘configure: error: unable to find
* rsvg-sharp.dll: no grkmozembed; this can be
* gtkhtml-sharp.dll: no found in the mozilla/firefox/
* vte-sharp.dll: no thunderbird development
The only libraries we need are art- packages’.
sharp.dll, gnome-sharp.dll and glade- The configure script is complaining that
sharp.dll, so you can safely ignore the it can’t find grkmozembed, usually
rest. If one of those three required installed as part of the Firefox
packages is not included (that is, has development package. The problem is
‘no’ instead of ‘yes’ beside it) you need that Mandriva has already installed the
to make sure that you’ve installed all required package when we installed
the correct Gnome development mozilla-firefox-devel, but has changed
libraries. You can easily check by the name of the file the package
entering -devel as the search criteria configuration manager is looking for.
in Mandriva’s package manager. If In fact, there are two files we need to
you’re happy with the output from the provide for the configure script to work
www.linuxformat.co.uk
– and the least destructive way of
making the connection is to create a
symbolic link between Mandriva’s
version and the one that the configure
script will be expecting:
$ ln -s /usr/lib/pkgconfig/mozilla-
firefox-gtkmozembed.pc /usr/lib/
pkgconfig/firefox-gtkmozembed.pc
$ ln -s /usr/lib/pkgconfig/mozilla-
firefox-xpcom.pc /usr/lib/pkgconfig/
firefox-xpcom.pc
We also need to add a single
option to Beagle’s configure script to
include the new web service. This lets
you query the Beagle database directly
from any web browser. To execute the
configuration script including this
service, type the following:
$ ./configure --prefix=/usr --enable-
webservices=yes
Running the script should now
proceed without any errors. When the
script is finished, the last block of text
should inform you of which compatible
extensions it has found, thus:
Target OS: linux
Extended Attributes: libc
Prefix: /usr
GNOME Prefix: /usr
KDE Prefix: /usr
Evolution-Sharp? no
(missing dependencies)
gsf-sharp? no
gst-sharp? no
Epiphany Extension? no
(Epiphany not installed)
Mozilla Extension? yes
wv1? yes
libchm? no
Local SqliteClient? yes
Enable WebServices yes
Any further extensions you may wish to
add, such as Evolution-sharp (this
allows Beagle to index Evolution
address books) can be downloaded
and installed from Beagle’s own
optional prerequisites page at
http://beaglewiki.org/
Optional_Prerequisites.
Finally, type make and make
install as root to compile and install
Beagle on to your system.
11/7/05 12:48:13 pm

STAGE 3 – RUNNING BEAGLE
Beagle runs as a daemon from a
user’s own account, using the beagled
command. If you’re having problems,
adding the -debug parameter can help
determine what could be going wrong.
The most common problem
encountered when running Beagle
looks something like this:
[graham@stinky ~]$ beagled
ioctl: Bad address
Unhandled Exception: System.
IO.IOException: Attempt to watch /
usr/share/applications failed!
in <0x0026c> Beagle.Util.Inotify:
Subscribe...
This is caused by the kernel’s
version of Inotify being incompatible
with Beagle, and you’ll need to find a
new one (such as the multimedia
kernel from stage one). Otherwise, if
running beagled seems trouble-free,
you can check the running status using
the beagle-info command:
Scheduler:
Count: 114
Status: Working
1 Generator 0
File System Crawler
last=/home/graham/beagle/beagle-
0.0.11.1, uncrawled=964, dirty=964)
Now that Beagle is running, it will
automatically index your home
directory in the background. If for any
reason there are parts of your home
directory you do not wish to be
indexed, you can choose between two
options. The first one is to create a
hidden file called .noindex in every
directory you wish Beagle to exclude.
This will stop Beagle from indexing
THE INSTALLED BEAGLE AT A GLANCE
[1] This is the Bleeding Edge Search
Tool, or Best, a graphical utility for
querying the Beagle database and
displaying the results as a list.
[2] When message are logged from
the Gaim Instant Messenger, your
conversations can be queried.
[3] Firefox needs the beagle.xpi
extension, after which your
browsing history can also be
queried, letting you jump straight
to the link.
[4] Word documents, either from
Microsoft’s venerable application or
OpenOffice.org, can have their
contents indexed.
[5] Rather than waste time sifting
through your inbox, why not let
Beagle retrieve any important
emails for you?
LXF70.feat_beagle 65
either the directory or any of its
children. You can include the names of
specific files and directories within the
file and this will allow others that aren’t
specified to be indexed as normal.
The second option is to add a list
of files and directories to a central file
called .neverindex in your home
directory. This works in the same way
as the list of files and directories added
to the .noindex file, and they both
accept wildcards and regular
expressions. You shouldn’t notice the
indexing process too much, as Beagle
manages its CPU usage. This means
that CPU usage is kept to a minimum
when you’re at the computer, but
Beagle will absorb idle CPU cycles
when you’re not using your machine.
Console or GUI
Now that we’ve gone to such lengths
to get it to work, we may as well put
Novell’s pooch to task. Staying on the
command line for a few more
moments, you can query Beagle using
the aptly-named beagle-query
command:
beagle-query sausage
This will search your home
directory for any Beagle-compatible
files that contain the word ‘sausage’.
This includes Word documents, Gaim
instant messages, emails, text and
binary files. Unless you collect sausage
recipes, this isn’t going to find much,
but you get the idea. The best way to
query Beagle, however, is to use a
graphical tool called the Bleeding Edge
Search Tool – Best for short.
2 3
1 strength to strength as more filetypes
4 5
www.linuxformat.co.uk
BEAGLE INSTALLATION GUIDE
Once Best has been launched, all
you can see is a benign Beagle icon in
your toolbar. Clicking on this opens a
simple search interface where you can
enter your search query in the time-
honoured way. Results should appear
immediately, listed below. You can
launch a viewer for each result by
clicking on the link. If you leave the
results window open, it will update
whenever Beagle finds any new
content, such as extra lines in an IM
conversation, or a new email.
You can even query Beagle using
your browser. This is why we added the
configure option when we compiled
Beagle. In order for it to work you
You can search locally
using a web browser
such as Firefox.
need to run the daemon with an
extra option:
beagled --web-start
You then need to point your
browser to http://localhost:888/
beagle/search.aspx. Your browser will
then present an interface very similar
to that of the Best search tool. It’s also
worth noting that you can download a
Beagle toolbar for Firefox that contains
much of the same functionality.
A new way to search
Why does such a simple idea deserve
so much attention? The answer is that
you no longer need to worry about
organising your data so much since it’s
always available immediately via Beagle
– potentially transforming how you
index and store your data. Sending out
Beagle is a far more efficient retrieval
process than hunting out the file for
yourself; and what’s more, Beagle will
bring up files sharing a keyword that
you might never have thought of
looking for.
Beagle is still in the early stages of
development, and it can only go from
are added and stability issues are
addressed. It won’t be too long before
Beagle is as integral to your desktop
experience as a web browser or media
player, and hopefully easier to install.
With any luck, you’ve been able to get
Beagle working on your system so that
you can see for yourself. LXF
LXF70 SEPTEMBER 2005 65
11/7/05 12:48:14 pm
 WEB DEVELOPING

Playtime on the

WEB
While we’ve been getting excited by innovation on the desktop, web developers have been making
giant strides in interaction, content delivery and pure, raw speed. Amias Channer looks at three
pulse-quickening projects: an XML breakthrough that promises a quicker, smoother user experience,
an exciting opportunity for hackers from Google and, first up, a new service at the BBC, whose
openness is being rewarded by some thought-provoking applications of its RSS content.

BBC BACKSTAGE FEEDING INNOVATION

W
e love web programming –
sitting up late, drinking
coffee and watching Buffy,
waiting for a visit from the
elusive muse… the thrill of
a sudden idea, and the
fevered tapping into the night as it
becomes reality. But in the morning,
all you have is a jazzier way to show
off your holiday snaps online. If only
there were some source of real, hard
content; important stuff worthy of your
programming skills.
Well, now there is. The BBC is
opening up its content for free,
non-commercial use at a new site at
http://backstage.bbc.co.uk for
anyone interested. Only the RSS feeds
and several content-access APIs are
available at the time of writing, but
much more is expected soon. The
project is meant to engage the web-
development community in the huge
amount of content that the BBC
provides, and also to provide beta
testing for its RSS feeds service.
RSS feeds are available covering
news, sport, BBC community projects,
travel, weather, entertainment, radio
and TV listings. Some of the feeds are
more complete than others but the
service is growing: the listings service,
for example, previously limited to
BBC4 information, has just added
seven-day listings from other BBC
channels and radio stations.
Before using the service you
should definitely read the terms and
conditions, but don’t fret: the licence is
fairly simple and easy to comply with.
For those who read licences a lot, this
will be a breath of common-sense
flavoured fresh air compared with the
usual legalese. The gist of the
agreement is that you aren’t allowed
to charge site visitors (hardly the first
thing open-source developers think
of) nor summarise the content, and
you must provide a link back to the
originating BBC site – easy.
Though you’re not allowed to
charge for the content, you will be

66 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.feat_bbc 66 8/7/05 5:15:00 pm

 WEB DEVELOPING

control of the environment and group of people to

CASE STUDY: EXPERIMENTING
WITH BBC BACKSTAGE
“I’ve been experimenting with
audio content, blogs and RSS for a
while, but the real insight came
when I realised the power of what
Dave Winer and Adam Curry were
developing, with podcasting. I
could download a feed, synthesize the text then
redistribute this enhanced feed as a podcast. The
Holy Grail for me has long been to get legal access
to a world-class news source. So when the BBC
Backstage project came along, I was overjoyed. I
crank out all my feeds, from the weather in
Prudhoe Bay, Alaska to BBC news, on a $600
Linux box in my basement. I use the Festival speech
synthesis system, a wonderful piece of open
source software. I program in Perl, using dozens of
TED GILCHRIST
modules developed by the Perl community, and
Rebotcast of BBC
world news feed distributed at CPAN.org. My use of the feeds
http://feeds.feedburner. themselves depends on the kindness of strangers,
com/bbcworldRebotcast either explicitly or via Creative Commons.”
allowed to display adverts on your site The launch is also expected to
but you cannot claim that the content feature some of the results from the
is yours or charge for the use of it. BBC’s recent flirtations with podcasting,
The official launch of the as well as the rest of the Open Tech
http://backstage.bbc.co.uk site programme, which is all about
will happen on July 23 at the Open technologies that everyone can have a
Tech 2005 meeting in Hammersmith, go at – be it familiar open source
London, organised by the UK Unix and
way it’s
presented and
the context it’s
presented in to
the end user,” he
says. “However,
that obviously
provides numerous
benefits on the flipside – such as
improved reach, innovation around
your content, and so on.”
Since the Backstage site was
opened it has attracted an active
developer community and some truly
innovative prototypes. There have
been no abuses – yet – of the terms
and conditions, but the prototypes are
checked when they’re submitted, and
regularly after that.
To learn more about BBC
Backstage you can subscribe to the
mailing list, which is rapidly becoming
a high-quality source of RSS hacking
information for UK-based projects.
The mailing list is also a great way to
get a feel for what’s going on at the
cutting edge of web development.
Duncan Metcalfe, a recent A-Level
work with, and makes me more
comfortable when thinking about how
technology could improve in the future.”
Some of our favourite submissions
are in the Five Of The Best box, below.
“We have been very impressed with the
quality of what’s being produced,” says
Metcalfe. “We will certainly be looking
into how we can incorporate some of
those ideas back into the main
bbc.co.uk content. People who
submit prototypes to us retain the
intellectual property of their work, so
we’ll be working with them on that.”
RSS goal
The British Broadcasting Corporation is
unique in the field of broadcasting
because of its charter and not-for-
profit status, both of which make it
very internet-friendly. It has always
used open technologies where
possible and even runs a few of its
own open-source projects.
The corporation’s reputation on the
internet is strong, and the Backstage
project shows its commitment to
making sure that its web content is as

“IT IS A REALLY ENJOYABLE,

Open Systems User Group – a huge
tip of the BBC hat to the open-source
community. There is much speculation
in the geek media and blogs that the
BBC will allow the downloading of its
huge library of TV series. Think Dr
Who, The Tripods, Bagpuss…
While this would seem likely to
dent the BBC’s profitable DVD and
video sales, it would be an enlightened
response to the explosion in illegal
downloading that threatens to do the
same anyway; at least this way the
BBC would be able to monitor what
was going out, rather than leave things
to the pirates.
Such a move would also be in line
with the BBC’s charter. Licence payers
have already paid for the shows, so
they could argue that they have the
right to watch them again without
having to pay again.
ENCOURAGING ENVIRONMENT AND
GROUP OF PEOPLE TO WORK WITH.”
software, or ‘repurposing’ everyday student who has submitted several well respected as the rest of its
electronics hardware like, say, an Xbox. prototypes using the RSS feeds, says services. The project also provides
he’s been inspired by the ideas and somewhere for it to grow into following
Best behaviour support in the project. “As my first time the success of the online news service
The Backstage project lead is Ben being involved in any way in a – it is aiming for 10% of its
Metcalfe, who spent months working developer network, it is a really news.bbc.co.uk traffic to be driven
on it in the BBC’s new media and enjoyable and encouraging by RSS by the end of the year.
technology division. He says parts of
the BBC initially needed some
“persuasion” to make its content
available. “You’re actually asking quite
a lot from a content stakeholder for
permission to let their content be
reproduced on other people’s websites,
where they will lose the physical

FIVE OF THE BEST

BBC BACKSTAGE PROTOTYPES
■ www.chatbots.co.uk/newsbot Have a virtual newsreader read news to you.
■ www.assistmedia.com/trafficlite Traffic updates from the BBC on your mobile.
■ www.missingwords.co.uk Online version of the Have I Got News For You missing
words round, with new headlines generated every 15 minutes from RSS.
■ www.backstage.min-data.co.uk/sport Find news about your nearest British
football team.
■ www.justinflavin.com/tagsoup/tagsouprss.php Tag soup of BBC technology news.
www.missingwords.co.uk: just like having Ian Hislop in your living room.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 67

LXF70.feat_bbc 67 8/7/05 5:15:13 pm

 WEB DEVELOPING
AJAX PUSHING FOR CHANGE
Used by Google Maps, Gmail and
parts of the Flickr photo-sharing
program, the technology known as
Ajax has been gaining notice for some
months now. At first, we were unclear
about what it actually was, though we
were pretty sure it was neither a
cleaning solution nor a Dutch football
team, and we did know that the sites
it’s being used on were pretty cool, so
by extension we assumed Ajax was
too. Now we know a little more about
it, and we’re pretty excited.
The term Ajax was coined by long-
time web developer Jesse James (no,
really) Garrett, pictured here, as he
related earlier this year. “I needed
something shorter than ‘Asynchronous
JavaScript+CSS+DOM+XMLHttpRequ
est’ to use when discussing this
approach with clients,” he wrote on the
website of web
consultancy
Adaptive Path,
which has been
using the
technology for a
while but didn’t
invent it as such.
In fact, Ajax
isn’t particularly
new or even original technology. It’s
PHOTO AND DIAGRAMS COURTESY ADAPTIVE PATH
really just a clever new way of using
some old favourites. Specifically, Ajax
uses mature, open-standard
technologies (HTML, CSS, JavaScript/
ECMAScript and XML), so minimal
extra learning is required and you don’t
need any new browser plugins. What is
new are the ideas behind Ajax, and
new ideas can take a little getting
used to – but it should be an
interesting process, and the results are
seriously impressive.
It’s actually useful!
Conventional web applications have a
very simple client–server structure,
with user requests, or HTTP ‘calls’,
having to go back to the server, almost
always resulting in a complete page
refresh. Traditionally, the server does
all the logic and presents the client
browser with simple interface pages,
though many people have tried to find
ways to allow web pages to manipulate
their own information or to allow a
page to receive more data without
68 LXF70 SEPTEMBER 2005
LXF70.feat_bbc 68
having to visibly reload. Cacheing and
clever layout design can make the
client>server>client journey smoother,
but there is always a noticeable delay
or break for the user.
The Ajax approach brings in a third
layer for a structured data stream. This
is usually XML, but can be JavaScript
Object Notation. You might think
adding an extra layer would slow the
process down even more, but it
actually speeds some things up. This
“THIRD PARTIES CAN
ENHANCE YOUR SITE OR
ADD EXTRA DATA SOURCES.”
separated data layer allows logic to be
applied to data before it goes back to
the server, or even for the intervening
Ajax machine to handle some basic
tasks by itself, and for new data to be
collected and dropped into the user
interface. This is very useful for
validation, but really comes into its own
when you need high-speed interaction.
Design pattern aficionados out
there will recognise this as MVC
(model view controller) architecture.
Design patterns are definitions of
common types of program structures
that almost all programs fit into, and
they allow you to refer to best practice
for whatever it is you are trying to do –
you shouldn’t have to reinvent the
Browser client
User activity
HTTP request
HTTP(S) transport
HTML+CS data
Web server
Data stores, back-end
processing, legacy systems
Server-side systems
Classic
web application model
www.linuxformat.co.uk
wheel every time you go for a drive.
One of the main problems with web
architectures is that they tend to follow
very heterogeneous structures, so best
practice is often hard to isolate and
reproduce in other scenarios. Noticing
that your work follows a design pattern
is a skill in itself. Design patterns won’t
necessarily help you build the best
code, but they do give you a set of
useful pointers on how not to trip
yourself up.
The MVC model is one of the most
common design patterns used in
programming today, with almost all
graphical operating systems using it. Its
main advantage is that it allows
separate development of the interface
and back-end components, whereas
the conventional web development
paradigm, as we’ve mentioned, keeps
the two heavily intertwined.
XML in the mix
If you’ve been around web developing
for a while you might think this sounds
rather similar to DHTML, and indeed it
is. But there’s something extra special
about Ajax – XML! Conventional
DHTML was limited to manipulating
data supplied with the interface via the
Browser client
User interface
Browser client
HTML+CSS data
Ajax engine
HTTP request
HTTP(S) transport
XML data
Web and/or XML server
Data stores, back-end
processing, legacy systems
Server-side systems
Ajax
web application model
Document Object Model (DOM),
whereas Ajax applications can request
data directly. This trick is usually
performed with a new JavaScript
function, XMLHttpRequest(), which
has only recently become part of
mainstream browsers’ JavaScript
implementations. With this lovely little
function, a page can request any XML
feed from anywhere on the internet via
HTTP. The possibilities here are
immense: you could get an up-to-date
list of blog articles, news headlines,
traffic announcements, syslog
messages from a server or, if you have
no imagination at all, stock quotes.
It is also trivially easy to generate
an XML feed from a database for use
with Ajax.
And the fun doesn’t stop there –
because the XML data is structured, it
can easily be put into the page from
which it was called, with judicious
manipulation of CSS and the DOM.
This will happen in the background via
the JavaScript events mechanism,
which can even make use of
eXtensible Stylesheet Language
Transformation (XSLT) to turn your
XML into HTML within the browser.
Traditional web paradigms would
require the page – or some horrible
hidden page – to reload. Either way it
tends to make the browser busy and
unresponsive, in the worst cases it can
even hide the page entirely. Add to
that creation and rendering of PNG
images and you can start to see why
the word ‘website’ suddenly seems a
bit wishy-washy… ‘application’ seems
more appropriate, with all the
interaction that’s going on.
Future of the web
Because there’s no interface reloading,
the Ajax approach (if done correctly)
should yield massive speed benefits
over more old-school methods. After
the initial page load, each call to your
server will just be fetching data without
having to fill out a template or make
decisions about interfaces.
Examined on a per session basis
this seems a small benefit, but over
the whole site it adds up to a
significant decrease in server load,
which in turn makes the data retrieval
even faster.
Another benefit of Ajax is its clean
separation of the data, interface and
logic layers, which means that third
parties can drop in replacements to
any of the layers without the need to
8/7/05 5:15:20 pm
 WEB DEVELOPING

FIVE OF THE BEST AJAX-BASED WEBSITES Ajax paradigm with a robust back-end
– Ruby on Rails appears to be the
■ http://maps.google.co.uk most popular so far. These toolkits can
Google’s jaw-droppingly amazing Ajax-powered mapping site. be a great way to get used to Ajax, but
■ http://gmail.google.com Google’s Ajax-driven über-email service. will not be so useful if you are hoping
■ www.google.com/webhp?hl=en&complete=1
to add to an existing site.
Google’s new search interface with autocomplete.
With the continuing provision of
■ www.monoplylive.com
Online version of Monopoly using GPS and wireless data from cabbies. useful open content such as RSS
■ www.basecamphq.com feeds under flexible licences, Ajax
Project management site that was so good it spawned Ruby on Rails. apps will soon become the norm,
leaving conventional websites looking
know the internals of any of the acceptance in the world of uninteractive and dull. While there are
others. If you wrote your website’s web development. The second is that, because its many competing technologies,
initial scripts with JavaScript classes, But there are two other factors that standards compliance is not owned by paradigms and buzzwords out there,
third parties will be able to subclass or have set Ajax apart from its peers. The any particular entity (unlike Java), there Ajax deserves your consideration
override your methods to enhance first is that with Ajax, unlike Flash, you can be no sudden changes of direction because of its great combination of
your site or add extra data sources. can view the source, which leads to or lock-in. Already there are several tried and tested component
Already Google Maps has spawned rapid community-based growth as toolkits for producing Ajax-based web technologies and the proven potential
several modified versions, each using developers examine each other’s code. applications that claim to integrate the it offers for innovative user interfaces.
the mapping engine to present
interfaces to various geographical
Classic web application model (synchronous)
keyed data such as business finders.
Before you allow third parties
access to your data, it would be wise
CLIENT
User activity User activity User activity
to create an acceptable use policy and
> > > > > >
some sort of developer program (such
Data transmission

Data transmission
Data transmission

Data transmission
as a Wiki or mailing list).
>

>
>

>
Disability drawbacks
You’re probably thinking: “This sounds Time
>

>
>
>

great – what’s the catch?” Well, there

are a few, but they’re easily avoided.
While the standard web paradigm
degrades nicely in situations where the
>

>

>

>
client is not the latest web browser, or
even a web browser at all, Ajax > > > >
applications aren’t nicely behaved at System processing
Syste ocessing System processing
stem pr sing
all and can be a real nightmare when SERVER
it comes to accessibility. Speech
readers and Braille printers will be
Ajax web application mode (asynchronous)
utterly mystified by an Ajax application
and will not be able to handle CLIENT
XMLHttpRequest() for some time.
For these browsers you will need to Browser UI
provide alternative access, or ensure
User activity
Input display

Input display

Input display
Input display
>

>

>

>

that everything can be done via the

>

>

>

>

normal forms or with standard links.

Not all of the core Ajax functions Ajax engine
>

>

>

>
>

>

>

>

are available in all browsers. To get

around this, several JavaScript libraries Client-side processing
Data transmission
Data transmission

Data transmission
Data transmission

>
>
>

>

>

>
>

>

are available that provide a unified API

across all browsers and help degrade
nicely in situations where Ajax support
cannot be used.
Time
>
>
>

>

>

>
>

>

Only now that the underlying

standards have matured has Ajax
moved from being interesting but
unusable in the real world to the point
>
>
>

>

>

>
>

>

where it’s powering some of the most

impressive sites around. Its
Server-side Server-side Server-side Server-side
namechecking on Google Maps, processing processing processing processing
Google Suggest et al has proven its
suitability and helped it gain
SERVER

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 69

LXF70.feat_bbc 69 8/7/05 5:15:28 pm

 WEB DEVELOPING
GOOGLE MAPS A HACKER’S PARADISE
Google Maps’s aerial maps are certainly useful, but add the information coming in from Ajax and it all gets much
more interesting. Here, bus stop information is tied to the map’s co-ordinates and overlaid.
If you’re looking for an example of
something cool that you can do with
Ajax, you could do a lot worse than
taking a look at Google Maps. If you
haven’t seen it yet, hurry along to
maps.google.co.uk for a treat.
Anyone familiar with Keyhole.com,
which used satellite images, 3D
geography and maps, will probably
experience a touch of déja vu: Google
recently bought Keyhole, and it’s the
Keyhole system that provides the
underlying mapping engine. Google
took the Windows-only application and
ported it to the web, making it much
more versatile, accessible and free.
The really clever thing, though, is
what Google did next. It used Ajax
methods to retrieve data and plot it on
its maps. The result is a truly
groundbreaking system for displaying
data from XML files that contains
70 LXF70 SEPTEMBER 2005
LXF70.feat_bbc 70
geographical co-ordinates (known as
geolocated data) over maps or satellite
images. It has all the comfort and
usability of a native application, so
much so that it’s hard to believe that
you’re looking at two separate systems
occupying the same screen space.
Two techniques
So how does it work? Interestingly, only
the query data uses Ajax methods: the
map image-fetching is done using the
more conventional IFRAME inline
frame method, so you can still use the
Back and Forward buttons – each
IFRAME refresh causes a History entry
to be created in the browser. This is
also the reason why the map will
display in all browsers but the query
functions work only in the latest ones
(though Internet Explorer has
supported them since version 3).
www.linuxformat.co.uk
The map images are impressively
handled, but the real magic is in the
overlaying of data points and polylines.
The user request is sent to the
server in the conventional way, with
the server returning a page in a hidden
frame that contains lots of embedded
GOOGLE MAPS COPYRIGHT
Google has been intentionally vague
about what it will allow with respect to
reuse of its mapping system. As long as
Google and the original copyright
holders of the map image are credited
on your site, it look like you can use it
for anything within the normal non-
porn, non-terrorist, non-offensive
boundaries. So far there has only been a
single takedown notice served on a
Google Maps derivative project – which
was a website that automated the
sticking together of map tiles to create a
XML. This XML is fed at load time to
XMLHttpRequest() via a call to a
function on the parent page. Now
here’s the odd thing – Ajax could have
been used to do this directly. We don’t
know why it isn’t being used to the full,
but it could be that Google’s
developers didn’t want to expose the
XML server directly, so they’ve put a
cacheing proxy server in front of it.
Given the maturity of HTTP cacheing
technologies they may have found
more scalability in having the XML fed
through a more conventional HTTP
request than requesting raw XML
alone with no protocol-level cacheing.
Once this data is received it is fed
to the _load() function from the
parent window, and an XSLT transform
is applied to the XML to turn it into the
local info panel at the side. This reload
also triggers the creation of those cute
pointers, speech bubbles and route
lines of the Seattle Busmonster shot.
Because of Ajax, the data can be
sourced from anywhere, but it will only
be meaningful if it has geographical
positions within it.
Hack me happy
So we know how it works – let’s hack
it! To generate a new map, we don’t
actually need to build much ourselves
(more details are available on the
Google Maps website), as an informal
API has evolved due to the emergence
of several mailing lists and Wikis for
Google Map hacking. But unless you
like having to rebuild your site every
day it might be a good idea to wait for
the official Google Maps API.
If you really can’t wait, the simplest
way to get started is to generate a map
at www.mygmaps.com using the
online interface, which will get you a
basic data file, host the resulting map
and probably give you a few ideas for
what you might achieve. You could
large poster-size image. In doing this,
the site broke the licence under which
Google obtained the data. Extracting the
map images and using them for
something else infringes the copyright
of the map data owners, so this site had
to be stopped to prevent the copyright
owners taking it out on Google.
Incidentally, the takedown notice itself
was typical Google. It was written as if it
was a note from a friend – it even
apologised to the site’s creator for
having to shut him down.
8/7/05 5:15:36 pm
 FIVE OF THE BEST
GOOGLE MAPS HACKS Quality, low-cost Co-Location
■ www.chicagocrime.org
Chicago puts in a bid to be the next Grand Theft Auto city.
■ www.busmonster.com
Hopefully the final nail in the coffin of the Seattle motorist.
■ www.gtraffic.info
Make sure the police behave by watching them on London traffic cams.
■ www.dynamite.co.uk/local
Simplify your stalking with geostamped Flikr photos.

Co-locate your server

■ www.housingmaps.com
Find houses and then check them out in satellite photos.

with us from just

£27.50 per month

for 1U of rack space, with 100GB of

high quality monthly data transfer.
Down and out in Chicago? No problem – here’s where to hustle
your money back.

then use the code from http://stuff. The word from the bloggers is that
rancidbacon.com/google-maps- Google is developing an API for site
embed-how-to to embed a map in developers to integrate Google Maps
your page. Simply call a wrapper into their own sites, but as yet this is
function and enter the name of the not finalised. Google has continued its NEW Quality Rack Mountable Servers
DIV tag that you want to display your tradition of being annoyingly good at now available to buy or lease
map. The first hacks were very doing websites, and seems to be
unreliable – Google Maps is still in getting even better despite having not
beta and the underlying API can much monetary incentive to do so –
change without notice. They did, all involved have (or should have)
however, demonstrate perfectly the already made Croesus-like amounts of High quality bandwidth with 99.9% SLA
potential for hackers, and should now money. With the release of a
be very stable. That potential is being commercial API for Google Maps they Modern Manchester-based data centre
tested by creative types the world over: will probably double this fortune. More
for our favourite hacks. power to them – and to us. LXF Pricing from just £27.50 + VAT per month
Friendly, approachable support
CASE STUDY: EXPERIMENTING
WITH GOOGLE MAPS Remote reboot switches available
DUNCAN BARCLAY “I decided to try to make something that mapped
Firewalled connection as standard
News by county
http://backstage.
BBC news stories to roughly the right place on a
Google Map by doing it by county. It was fairly
24/7 Emergency Telephone Support
min-data.co.uk/news simple to get the data from the RSS feed and on
to the map. It was simply a question of working
Online monitoring of your bandwidth
out the longitudes and latitudes of the places,
downloading and parsing the data (I used PHP),
then putting the values into the XML for the
Google Maps scripts to deal with. Getting the
Google Map on to the page was a bit more
difficult, but basically involved looking at the
source code for Google Local, and cutting out all
the code that didn’t involve the map. There have
been a few problems with it, such as when Google For more information, call us free on
updated their maps script, making me go through
the whole process again!” 0800 915 8771 or visit NorthernColo.co.uk

LXF70.feat_bbc 71 8/7/05 5:15:43 pm

 INTERVIEW GAËL DUVAL
Le
Comeback
Gaël Duval, founding father of Mandrakesoft, has
seen his baby go from the brink of disaster to
becoming part of a multinational Linux empire. How
did he do it? What next for Mandriva? And how can
two coffees and a diet Coke come to 33?
R
ising from the ashes of
bankruptcy, Mandrakesoft has
changed its name and more
than doubled in size through
acquisitions in the last six
months. At its head remains
Gaël Duval, the quiet Frenchman with
a penchant for Jimi Hendrix and a fear
of flying. Linux Format took a trip to
Paris, meeting Gaël at the Louvre for
lunch and a chat over a cup of coffee
to ask about the future of Mandriva,
the pressure from its competitors, and
just what it was that almost drove the
company to destruction...
LXF: A year ago, Mandrake was
almost bankrupt. Now you’re
coming out fighting with the
Conectiva and Lycoris deals. What
brought about these changes so
soon after bankruptcy?
GD: Maybe Mandrake lost some time
with the financial issues in the past, so
when all these problems were solved
we had to have things go faster…
LXF: To catch up?
GD: Yes. So the merger with Conectiva
was an opportunity to show that first,
we are still alive and well, and second,
to show that we can expand the
company and grow the business and
become strong and so on.
LXF: Whose decision was it to
merge with Conectiva? Was it a
number of people’s?
GD: The first step is to collect
opportunities, so there was Conectiva,
but there were other opportunities as
well. The second step is to discuss with
72 LXF70 SEPTEMBER 2005
LXF70.iview 72
everybody and the third step is to see
if it makes sense to merge with or to
acquire that company. Then once most
people in the management staff agree
that it would be a good idea –
LXF: On both sides?
GD: Yes, on both sides. Then we enter
a formal process, which is to have the
board of Mandriva give their opinion
about the merger, and then we have
to have all the shareholders vote yes
or no. So it’s a long process, and it
took nearly one year – nine months
from the beginning.
LXF: How close was the match
between Conectiva and Mandrake?
Was it quite similar in terms of
company strategy, or did you want
“I THINK THE RESPONSE FROM THE
COMMUNITY WAS THE KEY FACTOR THAT
HELPED US TO NOT STOP EVERYTHING
AND SAY, ‘OK, WE GIVE UP’.”
something quite different?
GD: It was interesting for us to merge
with a company which was not exactly
in the same market. It was not an
exact competitor to Mandriva because
they have a local market in South
America so it’s interesting for us
because it’s a complement to
Mandriva, it’s not a… I don’t know how
to explain that. And it was a company
with a similar size with approximately
the same number of people in place.
There was a good feeling when we
www.linuxformat.co.uk
THE LXF
King
INTERVIEW
were in discussion with them, so…
Before the merger we already started
to work together, we started the LCC,
the Linux Core Consortium, so there
was already a good feeling with them.
And on their side, they were interested
to buy the deal because they were
quite small and not very famous, a
marginal player, so it’s an opportunity
for them to have more visibility.
LXF: Did the LCC actually help get
you anywhere?
GD: There is not much activity on this
side currently. There shouldn’t be any
impact – it’s just that two members
have become one member.
LXF: As a result of the merger, do
you see Mandriva’s goals being
closer to Conectiva’s goals, or
closer to Mandrake’s goals, or a
mix of the two?
GD: Basically they are the same. We
do a Linux product and we sell it to
individuals and corporations – they do
it in South America, we do it especially
in the USA and Europe.
LXF: Do you think Mandriva will
continue to be all the things
Conectiva used to be? Are you
moving your market a little bit?
GD: I think the markets are quite
similar. There is just the question of
the product line, so we have to merge
the product lines, the brand names,
because it’s more efficient to have
only one product line and one brand.
That’s one of the reasons for the new
name, because Conectiva plus
Mandrake gives Mandriva.
We are in the process of merging
the products into only one line of
Linux solutions, and maybe they will
have some specific solutions for South
America if there is a particular need
for them in the local market, but
globally we will merge everything. We
take the best of Conectiva Linux and
we put it in Mandrake, and everything
becomes Mandriva.
LXF: What are the best bits of
Conectiva Linux?
GD: They have some good technology
in the Smart software. It’s like urpmi,
the dependency software of Mandrake,
but they have better algorithms and
it’s more sophisticated. We are going
to merge urpmi with Smart into a >>
great separate package.
11/7/05 3:13:24 pm
 INTERVIEW GAËL DUVAL

TOP TRUMPS
LXFtrumps
Gaël Duval

COURTESY: WI
Created the Ma
nd
as a fork from Re rake distro
1998, turning it d Hat in
int
profitable compa o a

NNING
ny
boxed sets. Stee providing
re
from near-bank d it away
ru
Edge-IT and Lyco ptcy to buy Conectiva,

MOVES UK LTD
ris.
AGE
NATIONALITY 32
YEARS USING Frenc h
LINUX
PROGRAMMIN 10
G LANGUAGES
NUMBER OF PC 8
S
DAILY COFFEE 12
INTAKE 2 x decaf
SANDALS OWNE
D
HE SAYS “People 1
who claim that
100% ready to Linux is
replace Window
don’t know wh s
at they are talkin just
g about.”

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 73

LXF70.iview 73 11/7/05 3:13:29 pm

 INTERVIEW GAËL DUVAL
>> LXF: Is Conectiva Linux
APT-based?
GD: No, no, it’s RPM, but it can
understand APT to a certain extent.
LXF: Do you plan to visit
Conectiva’s offices?
GD: Maybe one day. Not soon,
because I have to find a boat.
LXF: You don’t like to fly?
GD: I can’t fly, I’ve got a plane phobia...
LXF: Had you considered
merging with SUSE instead?
They’re a lot closer!
GD: Well you usually take planes to
Germany, so, same problem!
LXF: Tell us about the Lycoris buy-
out – what prompted the move?
GD: Mandriva started as a desktop
Linux distribution, although we have
diversified on the server. But we think
that it’s important for us to go further
on the desktop because it’s part of our
image and also because we want to
be ready for when the Linux desktop
takes off.
LXF: So you don’t think Linux is
ready for the desktop just yet?
GD: I think that Linux is technically
ready. I mean, the technology is here. was more than €20 million, and of
But we have to wait for
independent software
vendors to release
more often for Linux, to too many people. In the middle of
consider it as a viable 1999 we were approximately 12
market. Of course this
is a virtuous circle.
74 LXF70 SEPTEMBER 2005
LXF70.iview 74
On the other hand, there are good
signs with the recent release of Nero
for Linux and the new Acrobat Reader.
And maybe the move of Apple towards
Intel [see News, page 6] could have a
positive effect as well, by reducing the
costs and simplifying the work of
software vendors to port their
applications to Linux. On the other
hand, if we wait enough we will
certainly see every piece of Windows
software have an open-source
equivalent, so this time Linux will be
really ready for the desktop – and it’ll
be too late for commercial software.
LXF: Can we expect another
name change?
GD: No!
LXF: Going back before the
merger with Conectiva, what led
Mandrake into the problems that
it had, and what have you learned
as a result?
GD: I think it’s a problem of timing: we
started in 1998, and Mandrake was
successful very quickly. The problems
came after, in the noughties, and they
were connected to the dotcom crash.
In 1999 and 2000 we got plenty of
cash coming into the company
– investment from venture capitalists. It
course when you’re a young company
and there is a lot of cash coming in,
you start to spend too much, to hire
employees at Mandrakesoft, and one
year after, we were 150.
www.linuxformat.co.uk
LXF: Wow! OK, now we know
where the money went.
GD: Yeah. You know, in France the
laws to fire people are very strict, so it
took a very long time to decrease the
number of employees and to break
even, so between 2001 and 2003 we
progressively decreased expenses. But
this took time – three years – and we
ran out of cash. When you run out of
cash you have to close the company
and find some solutions.
LXF: Was poor decision-making
part of the problem?
GD: Yes, a little bit. It’s hard to know if
something was wrong or not, because
you have to try another path and to
know if it’s a good path, but I think the
decision to hire too many people was
a wrong decision. After that there were
decisions that were taken, like starting
the e-learning business, which had a
cost, and the management at this time
took some decisions that were very
costly because they thought money
would always come in to the company
and cover all the costs.
LXF: Not very realistic.
GD: No, and all the management staff
had to leave, because it wasn’t
possible to go on like that in the long
term. We started to clean up
everything, but it took a few years.
LXF: Of those 150 employees at
the height, how many were
engineers for programming, and
how many were sales,
administration and management…?
GD: Maybe 60 programmers. We had
a team in the US also; that was very
expensive. A sales team but engineers
also. Maybe a third were developers.
LXF: It must have helped your
finances enormously to have
people in the community buying
MandrakeClub memberships.
GD: At the end of 2002 we knew that
there was a very serious money issue,
and we knew that if we wanted to
grow further we had to find some
money. We were unsuccessful at
finding more investment so we tried to
explain to the community of users that
we had serious problems with money
and that if they wanted to help
Mandriva, they could do so by buying
our products, subscribing to
MandrakeClub… I think the response
from the community was the key
factor that helped us to not stop
everything and say “OK, we give up”.
LXF: So they kept you alive?
GD: Yes, yes, we realised that
Mandrake was very important to them
and at the same time many people
subscribed to the club to help us
financially, so we realised we had to at
least try to continue and find solutions.
LXF: Are you worried that you’re
expanding too quickly and might
repeat your earlier problems?
GD: [Lycoris] is our third acquisition in
one year and we haven’t spent much
money – [it’s been] mostly through
the stock exchange. Additionally,
Edge-IT and Conectiva were profitable
companies, like Mandriva. So the risk
and our situation are really not the
same as before when our expenses
were three or four times our revenues.
LXF: Lycoris was originally based
on Caldera Linux [forerunner of
SCO] – do you think that kind of
association may harm Mandriva?
GD: I don’t think so – the Lycoris
image was not really associated with
Caldera or SCO.
LXF: Do you foresee further
consolidation of the Linux market?
Do you think that’s a good thing?
GD: We’re very happy to be in a
situation where Mandriva is buying
other companies instead of being
absorbed by a bigger one. And that’s
a good thing for Mandriva to be able >>
to expand.
11/7/05 3:13:38 pm
 INTERVIEW GAËL DUVAL

“WE DON’T THINK THAT

LINUX IS READY YET FOR
THE MASS MARKET.”

LXF69 AUGUST 2005 75

LXF70.iview 75 11/7/05 3:13:41 pm

 INTERVIEW GAËL DUVAL
>> LXF: So have we seen the end of
the Mandriva spending spree for
the time being?
GD: It’s likely that there’s going to be a
pause because it’s already a big job to
have Conectiva and Lycoris join
Mandriva in terms of organisation and
processes. But who knows?
LXF: There’s only been one limited
edition Mandriva release so far in
2005, whereas previously we had
PowerPack, PowerPack Plus… will
there be only one release in future
or will there be different levels?
GD: The limited edition is really a
special version, so there will only be
one. The name of the next version will
be in track with this version, because
we will add 2006 in full, but there will
be the demo pack, and PowerPack
Plus, and Discovery.
LXF: Mandriva has switched to a
yearly release cycle; but there has
also been a recent announcement
that every two months there will
be a club release with new
features. Is that yearly, or every
two months, or a bit of both?
GD: There are the official products on
the one hand – the product line that
you find in boxes with documentation
and support, so people can buy it in
stores and stuff, and on the other
hand there are club subscribers, who
have access to all this product and
more. The message is that if you want
all official Mandriva products, you have
to be a member of the club. Club
members can download an updated
version every two months, with new
features such as the new KDE and the
new Gnome.
LXF: Does that update itself
continually, or do you have to
keep on downloading new ISOs
and installing, installing, installing?
GD: You have the choice. There are
updated packages, but you can also
download the complete distribution.
LXF: Was this decision to have a
new release every two months
related to your competitors
having more frequent
releases? The new
SUSE 9.3 has Gnome
2.10, KDE 3.4, Xen,
Beagle, F-Spot,
Mono, Java – it’s got
all these things, and
76 LXF70 SEPTEMBER 2005
LXF70.iview 76
people saw Mandriva come out
two weeks after SUSE and yet it
had older Gnome, older KDE…
GD: Traditionally Mandrake was at the
cutting edge of software. SUSE and
Red Hat and others were more
conservative and they tried to be sure
that they only offered the most stable
software. And now they are trying to
do what we did before so it’s a bit
funny, but if we release every two
months it’s not because of SUSE or
anyone else, it’s because our users tell
us that they like to have things
updated.
LXF: So why is Mandriva 2005
behind SUSE in terms of numbers?
As you say, Mandriva has always
had all the latest software and
now suddenly it doesn’t. What
happened there?
GD: Compared to what we did before
we have more formal processes in the
company because we want the quality
to be very good. When you have
formal processes you have to choose
which version of software you will put
on the product, maybe three or four
months before the release, so it’s hard
to put the very latest version on it. I
don’t know how SUSE does it!
There are two types of Linux users.
There is the geek, who always wants to
have the latest version. And then you
have the corporate market, and
individuals who don’t know much
about Linux and just want to try it,
maybe to use it, but they don’t want
the latest version. We want to address
both these people. I think it’s good to
www.linuxformat.co.uk
offer very stable software in
official releases, and on the other
hand we can offer more cutting-edge
software to club users and people who
always want the latest. I think maybe
SUSE takes some risk to put… I don’t
know how stable it is, but…
LXF: The new Mandriva release
has had a lot of criticism among
our readers because the graphics
are very cartoony. You’re trying to
appeal to the business market, but
those graphics just don’t work in a
business environment.
GD: This version in particular is a
special version. We had planned to
have an official version in March with
Mandrake, Conectiva and many other
things, but it was not possible, so we
decided to release a special
transitional version. It’s not targeted at
the corporate market or the regular
market – it’s just for power users.
Power users have no problem with
cartoonish pictures.
LXF: Where do you think
Mandriva’s key markets are?
What’s the biggest place for you in
terms of revenue?
GD: We have a strong core of home
users. We can rely on this market – it’s
a big part of our revenues. But we
think that the corporate market, the
business market, offers more
opportunities for us to expand. And
maybe later another step would be to
address the mass market, single users,
but we don’t think that Linux is ready
yet for this market. We will be ready
for when it starts to explode but for
now we have to offer products to our
core user base and we have to expand
the business in the corporate market.
LXF: So how about around the
globe? Obviously Mandriva’s got a
presence in France and Brazil, but
where else do you have a strong
market presence, and where
would you like to be stronger?
GD: We are strong in the corporate
market in France, because our
presence in Paris makes it easier to
meet people and to go to big
corporations and governments and
places. But speaking about individuals
… yes, we have a good user base in
France, but we have a good user base
in the UK, in the US, and in many
other places in Europe. For individuals
we make a big part of our business
through our online store, and we ship
to approximately 150 different
countries, so the market is big but we
can’t have offices everywhere.
LXF: Do you think the market in
South America is very different
from the Linux market in Europe?
GD: There is a big Linux market in
South America, especially in Brazil,
because I think that more than here
the governments really want to get rid
of Microsoft – they don’t like Microsoft
too much and I don’t think they like
the USA too much as well. So there is
a big market and it’s different because
11/7/05 3:13:45 pm

it has its local rules, but it’s a Linux
market for big corporations and
government agencies.
LXF: How about some of
Mandriva’s non-core products, like
Mandriva Move? A while back you
did a TransGaming boxed product
incorporating WineX; plus you did
an Xbox Mandrake distro. These
aren’t really desktop Linux, these
are non-core products. Do you
expect them to continue?
GD: Ah yes, Mandriva Move will
continue because it’s the easiest way
for Joe User to try the product –
“I NEVER PLANNED TO DO
A CAREER, I JUST WANTED
TO DO WHAT I LIKE.”
everybody who can switch on a
computer can try Linux just by putting
the CD in the CD-ROM. So yes, we
are going to continue with that. But I
don’t think we are going to release
another gaming box. Yes, we have
Xbox versions but it’s very… it’s not a
market decision. If you take the
example of the Xbox version with the
latest Mandriva, I’m pretty sure that
our CEO didn’t know about it before
the release!
LXF: What are the areas in which
Mandriva is behind its
competitors? Where do you think
they are better than Mandriva?
GD: If we talk about technology, I think
the core is Linux, so choosing a Linux
distribution is a bit like choosing a car
– you just choose a colour and brand,
but most components are the same
inside, and it’s true for most Linux
distributions. There are things that
make Mandriva easier to use than
other Linux distributions. But maybe
you will find some technology, a
feature, in another distribution that will
make you prefer that over Mandriva.
LXF: Mandriva has always had
great configuration tools. Do you
see that as a selling point for
Mandriva, or an area in which
distros should work together?
GD: I think that it’s better that all
Linux distributions rely on a common
configuration layer, such as
configuration files in the same place
LXF70.iview 77
and so on. This has to be handled by
common standards such as the LSB
[Linux Standards Base]. If each Linux
distribution builds their own graphical
tools to address this configuration layer,
there is no problem, and it’s actually a
plus for Mandriva.
LXF: Does Mandriva plan to
compete in all the same market
segments as Red Hat and SUSE?
Red Hat has Enterprise Linux,
which goes up to IBM mainframes
– really high-level machines. Does
Mandriva plan to compete in all
the same areas?
GD: Yes, if we have the opportunity
we will do so.
LXF: Also, Red Hat has a lot of
support in place. They guarantee
support for seven years – the
lifespan of the product. Is
Mandriva capable of doing that?
GD: We offer support for two or three
years at least, but we are going to
increase that because there are very
big opportunities in the corporate
market for us.
LXF: Do you see a viable future in
selling your boxed distros?
www.linuxformat.co.uk
GD: It depends on where you’re
talking. Sales are decreasing
everywhere that people have
broadband access: they prefer to
subscribe to the Mandriva Club and
get instant access to our latest
commercial products. But in other
areas, they sell well.
LXF: What’s your opinion on
software patents?
GD: Most small proprietary software
companies are going to be equally
hurt by software patents. At the same
time I’m confident that the European
Parliament is going to adopt the good
amendments so that we reach an
acceptable law.
LXF: Obviously you’re not
proprietary, but would you class
Mandriva as a small software
company? How many employees
do you have?
GD: Employees… currently there are
around 60 in France and appriximately
60 in Brazil. So 120 in total.
LXF: And how many of your
employees are engineers?
GD: 35 in Paris and about 30 in Brazil.
LXF: Thirty-five just in Paris? Wow.
So what kind of projects do they
work on? Or is it really bugfixing
and packaging?
GD: You mean for software projects?
Most engineers, when they find a bug
or something they will report it and
contribute a patch to any free software
INTERVIEW GAËL DUVAL
project – this is the primary developer
contribution. And the guys who
package KDE and Gnome at Mandriva
are also directly involved in the KDE
and Gnome projects so they work on
this project in their work time. For a
long time we had David Faure, who
did a lot of work for KDE.
LXF: Do you think that the
Mandrake and Conectiva merger
will result in some layoffs?
GD: No plans. Maybe if we… who
knows what can happen in the future
but there are no such plans.
LXF: At what point did Mandrake
stop being a hobby and become a
proper career?
GD: On December 15, 1998 I guess,
when I became the first employee of
the just-born Mandrakesoft! Anyway I
don’t know if we can call that a career.
I never planned to do a career, I just
wanted to do what I like, and that’s
one of the reasons why I preferred to
start a business when I finished at
university, instead of finding a job in an
existing company.
LXF: Do you think you’ll work on
Mandriva forever? Do you enjoy it?
GD: Yes, of course. I started the
Mandrake project alone, and then it
became a company, and I started with
development, and then obviously I
worked on the website and so on, and
now I’m in charge of communications,
and I think I am going to change to
something else, so... who can say? LXF
LXF70 SEPTEMBER 2005 77
11/7/05 3:13:47 pm
 TUTORIAL First Steps: Upgrades

FIRST STEPS SYSTEM MAINTENANCE SERIES

Upgrades
Get the best from Gnome 2.10
PART 3 Andy Channelle dons a pointy red hat and grabs his toy fishing rod
to look at some of the new features of Gnome 2.10.

Last month we looked at
some of the aesthetic and
practical updates in KDE 3.4
including the new RSS
reader aKregator. If you
missed the issue, call 0870
8374722 or +44 438794 for
overseas orders.
While KDE is winning users the Linux world over by
offering everything but the Kitchen sinK, the GNU
Object Model Environment, or Gnome as it is
commonly called, remains a popular choice for its clean looks,
innovation and admirable goal of promoting simplicity on the
desktop. Gnome is the default desktop environment for several
distributions, including Fedora and Ubuntu, and seems destined
to play a larger part in Novell’s desktop plans, as the company
has become one of the largest corporate backers of the project
via its Ximian acquisition.
In the past, Gnome has faced criticism for its reductive
approach and some of the controversial decisions taken in its
name. However, Linux use is about choice, and that’s true even
after you’ve chosen your desktop environment – there’s no rule
that you must use the exact application suggestions of the
developers or the recommended method of working with
Gnome once you’ve installed it.
One of the more contested innovations introduced into
Gnome recently is the spatial mode of the Nautilus file manager.
This tutorial will look at the pros and cons of ‘spatial’ navigation
and show you the quickest way to go back to a more traditional
– non-spatial – way of managing files with Nautilus. I will also
examine some of the customisation options and new features
available in the latest 2.10 version of the desktop.
A successful desktop is one that doesn’t require the user to
go through complicated procedural hoops in order to
accomplish simple jobs, and one where the user can adapt the
environment to suit them – not the other way round.

80 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.tut_begin 80 11/7/05 9:57:27 am

 TUTORIAL First Steps: Upgrades

PART 1 – CONFIGURE THE DESKTOP

The typical desktop in Gnome 2.10 – prior to any

user-led changes – can look incredibly stark. Ubuntu, for
example, goes for the completely bare look, relegating even the
trash/recycle can to the lower taskbar, while SUSE removes
trash, Home and Computer from the desktop and just provides
a link to them instead.
At the top of the screen is a taskbar, which houses Gnome’s
equivalent of KDE’s kpanel, the system tray and a clock. A
second taskbar at the bottom of the screen allows one-click
access to all running applications. Just as in KDE, there are a
few ways for you to alter the look and feel of these panels or
remove them altogether. One word of warning though: the
presence of three menus in Gnome’s top bar (see Part 2)
means that attempting to decamp this to either side of the
screen makes for a very ugly and intrusive panel, and it is not
Gnome ships
a straightforward process to get things back to normal. with a range of
Start by right-clicking on the top panel. This provides applets suitable
access to various resources, quick-launch applications and for putting in
the system tray and clock. Right-clicking will bring up a new The desktop the panel.
menu with which we can add or remove items from the panel, background tools demonstrate one of the
change its size or look and generally move things around. There ways Gnome is changing itself and following a ‘simple = good’
is quite a decent range of things we can add to a panel beyond approach. Earlier versions had an option for setting backgrounds
the ordinary, such as a mini-command line, a drawer for holding for individual virtual desktops, but this has been dropped.
files and folders that might otherwise be dropped on the
desktop and swift access to your Tomboy notes (see Tomboy Virtually unlimited
Speaking of virtual desktops, these allow us to open, for
Configuring the example, communication applications on one desktop and a
Pager is word processor on another without having to deal with a
straightforward.
confusing array of open windows. They are also useful for those
times when a spouse or boss interrupts a game of Frozen
Bubble and you need to switch quickly to a work screen! Most
distros tend to install with four virtual desktops and these are
accessed with the Pager, which is usually placed somewhere on
the bottom panel. There is, of course, no reason why we should
stick with four desktops. If you think you’ll only need two
desktops (or 10 or 16, for that matter) simply right-click on the
pager and select Preferences to open the relevant dialog.
There are three ways you can get an application window on
to a particular desktop. The most obvious is to select that
desktop with the mouse and then open it. The second is to click
on the window’s Properties icon, which is usually at the top left
of the window. Just choose Move To Another Workspace and
select the workspace name or number. However, the coolest
way to do it is to grab the window’s icon within the Pager and
simply drag and drop it on to the right desktop. This is great, but
box, page 83). Simply select Add To Panel and choose the a little difficult if you have loads of windows open.
required applet. It should be added exactly where you did the
right-click. If you want to move it, right-click on the icon and
select the Move option.
More fundamental changes can be made to any panel by
selecting the Panel Properties option. In this dialog it is possible
to force panels to follow the style of the Gnome theme, use a
background image or use a solid colour, complete with the usual
transparency options including the ability to set different levels
of transparency on each panel.
Change the background of the desktop by right-clicking on
any open space and selecting Change Desktop Background.
Any images or patterns already on the system will be shown in a The easiest way to
list. There are options here for adding images and changing the move a window to a
way they are displayed, such as tiled (for patterns), centred or different workspace is
filling the entire screen. It’s also possible to remove the image – through the window’s
just select No Image – and then choose a plain desktop colour
right-click menu.
>>
in the dialog’s lower section.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 81

LXF70.tut_begin 81 11/7/05 9:57:31 am

 TUTORIAL First Steps: Upgrades
PART 2 – SYSTEM NAVIGATION
The best way to gain entry to your Gnome desktop is via
the three menu entries on the top bar. These are Applications,
Places and System (or Desktop in SUSE).
The first of these gives you access to the applications
installed on the system – no surprise there. These are sensibly
divided into task-centric submenus. So, under the Internet
entry, we find applications such as Evolution for email and Gaim
for instant messaging. You can launch these applications simply
by selecting them with the mouse, and you can also drag and
drop their icons on to a taskbar or the desktop for rapid access.
The Places entry is discussed below, so we’ll turn to System as
this is where we can change many of the ways in which Gnome
has been set up by your distribution’s engineers.
The most important entry under this menu is for the Gnome
Control Panel or Preferences, where we can change most of the
Gnome-specific options. Below this you should find more distro-
centric entries and these might include YaST or some other
administration tools. If your distro has set up a non-standard way
of getting into the various options, you should be able to launch
the control panel by selecting Run Application from the
Applications menu and typing gnome-control-center (note the
Americanised spelling).
Look and feel
In this large utility, we are most concerned with the Look And
Feel section, which lets us choose desktop themes and set up
fonts for the system.
■ Windows This section allows us
to set up how the mouse interacts
with windows – for example
making windows ‘active’ just by
moving the mouse pointer over it
for a pre-defined time, or ‘rolling
up’ an open window with a
double click on the title bar. It is
also where you can dictate which
key can be used in tandem with
the mouse to move windows
around. By default this is the Alt
key, but it’s a good idea, if your
keyboard supports it, to use the
Super/Windows key as this may
otherwise be redundant.
Gnome can deal with any ■ Theme This is where we can define the visual
sort of key presses you’d elements of Gnome including window borders and widgets –
care to throw at it. buttons, menus, scrollbars, and so on. Each distribution ships with
PART 3 – FILE NAVIGATION
Gnome’s developers and maintainers caused some
consternation recently by changing the way Nautilus works when
accessing files. The big change in the addition of a so-called
spatial navigation mode, which basically launches a new window
every time you open a folder.
So far, so Windows 3.1. But Nautilus is a bit cleverer than
that. It remembers, for example, both the position of each
window on the screen and the view mode selected when the
window was closed. So while you may want to view a folder full
of images as Thumbnails, it is better to browse a folder full of
spreadsheet files as a list complete with creation and
modification details. The downside is that it is possible, when
navigating to a document at the bottom of a stack of folders, to
82 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk
LXF70.tut_begin 82
Make use of the otherwise redundant ‘Windows’
key in this dialog.
a range of pre-made themes, including their own house styles,
but these are just a starting point as each of the three main
parts – controls, window borders and icons – can be individually
customised using the Theme Details tab.
This tab is divided into three parts and new elements can
easily be downloaded and installed from websites such as
http://art.gnome.org (which is also a useful place for desktop
background art, by the way) and, once you’re satisfied with the
setup, can be saved under a suitable name using the Save
Theme option in the main dialog box.
■ Font The Font dialog box lets us set up the textual look of the
desktop, applications (the menu bars and stuff), window titles
and the terminal. It is also where we can set up the font anti-
aliasing to make your screen look beautiful; there are options for
monochrome CRTs, LCDs and so on.
■ Desktop I have already mentioned this, but briefly this is
where we set the desktop image. In this dialog box, you will also
find, under the Personal section, an entry called Shortcuts, which
allow us to set up some keyboard shortcuts to access particular
applications, processes or folders. To set something up, simply
choose the entry from the left side of the dialog box and hit the
key (or combination) used to initiate the command. This is where
you might set up a non-standard keyboard to access the
multimedia features of Gnome: volume, play/pause and so on.
The last thing we’re looking at here is the Session dialog,
which is where we can kill naughty applications (under the
Current Session tab) and the Startup Programs tab where we
can set applications to be launched automatically on startup.
end up with a ton of open windows on the desktop, and that’s
just annoying! The good news is that it is quite simple to change
this behaviour either universally or individually. Doing it
universally used to involve going into Gnome’s GConf editor,
which was the equivalent of using the Windows Registry – and
we know how intuitive that is – but the latest versions have
made things easier.
To change the annoying default, open any Nautilus window
by double-clicking on a folder to open it in spatial mode. Now
do Edit > Preferences and go into the Behaviour tab. The entry
that makes the difference is Always Open In Browser Windows.
Select this, close the dialog box and all subsequently open
windows will follow the browser paradigm.
11/7/05 9:57:36 am
 TUTORIAL First Steps: Upgrades

appropriate tag. Again, Gnome updates automatically

Navigating through a network volume can leave the
desktop full of windows.
There are still a few other things we can do to make the
setup even more Windows/KDE-like. Chiefly this means adding a
sidebar that we can use to access the complete file tree. To do
this select View > Sidepane. By default this sidepane displays
information about the currently selected folder or file, but we
can change this by selecting the Information drop-down and
changing it to Tree.
If you quite like the spatial method but occasionally need a
less cluttered desktop you can right-click on a particular folder
icon and instead of choosing Open, select Browse Folder and
this will open up the folder in the browser interface – the best
of both worlds.
One feature that tends to be forgotten or overlooked is
Gnome’s emblems. These can be added to individual files to
give us an extra method of sorting files in sensible ways. For
example, it is possible to add an ‘Art’ emblem to images in a
folder and then select View > Arrange Items > By Emblem to
corral all the pictures together. Of course, you could also
arrange items By Type but then the GIF and PNG files would be
potentially far away from the JPEGs. Using emblems depends –
like so many things – on diligently tagging files with the
appropriate emblems. To add an emblem, select the image or
file with a single right-click and choose Properties from the
menu. Click on the Emblems tab and then select the
to include your changes.
Of course, as these emblems are just normal Quick browsing
PNG files (usually with transparent backgrounds) it is You can speed up Gnome for
entirely possible to make your own emblems that file browsing by going into
reflect your particular situation. Once you have Edit > Preferences and
designed an emblem you can either drag it into the ensuring all the entries under
/.nautilus/emblems/ folder within /home or open a the Preview tab are set up for
window, do Edit > Backgrounds And Emblems, select Local Files Only. Slower
the Emblems tab and click on Add A New Emblem. hardware would also benefit
from removing the
You’ll notice in this dialog that you can also change
background image and just
the background colour or texture of windows and this
having a solid colour.
is really where the spatial thing comes
into play, because alterations made to
individual windows (such as
/home/Documents) are persistent.
One other cool usability feature is
the Zoom tool, which enables us to
reveal more of the contents of a file
without having to open it up. This
makes obvious sense when it comes to images
(in which case why don’t Windows or OS X do
it?) but Nautilus can also do the same job with
text files. By default Nautilus displays previews
at 100%. Just as with the behaviour of
particular windows we can change this on a
window-by-window basis or universally. The
simplest thing is to keep the default and then,
when needed, use the Zoom options within
the Nautilus user interface to change
individual windows. Using the spatial mode
you can zoom in or our using the View
menu, while ‘browser’ users can do the same with the + and Say it with me: spatial
– icons on the toolbar. Nautilus remembers these settings so it navigation is an OPTION!
is safe to set up bigger previews for image-filled folders while
keeping small previews for more general locations.
One of the newer ideas in Gnome 2.10 is the Places
menu entry, which provides easy access to storage devices,
significant locations and network volumes on the NEXT
computer, and it all does pretty much what you might
expect: Floppy is linked to a floppy drive, Desktop opens MONTH
We’ll wade into
a folder with the contents of the desktop, and so on. The
the often confusing
Computer menu item opens up a Nautilus window with
world of file
icons for filesystems attached to the system. LXF
compression.

FROM THE F-SPOT AND BEAGLE STABLE: TOMBOY

Provided with many of the Gnome 2.10-based
distributions currently available, Tomboy is one of a new
breed of Gnome applications that are coming into the
desktop as a result of the Mono project. Others include
Time to rethink the ‘My Favourite
Application’ award.
F-Spot, an excellent photo management application, and
the Beagle desktop search tool.
To call Tomboy a note-taking application is much too
weedy; this is a notes app with the muscle and
simplicity of a world-class bodybuilder. It is a kind of
personal Wiki system which makes linking notes and
thoughts brilliantly easy. What’s more, it integrates with
a Gnome panel so that you can open, browse and
search your notes straight from the desktop.
On first run, Tomboy consists of a single note called
Start Here. On the surface this is the same as any other
note-taking app, but there’s far more going on beneath.
To demonstrate its power, let’s make a new note called
Tomboy. Simply type Tomboy somewhere on the Start
Here page and highlight it with the mouse. Now select
the Link icon at the top of the window. A new note will
be created called Tomboy and you can begin adding
text to it straight away. Unusually, there is no ‘saving’ or
‘loading’ involved. Everything is saved automatically.
We now have two notes, Start Here and Tomboy.
Close Tomboy so that Start Here is active, and begin
typing some text. Now the clever bit: if, in the course of
your typing, you enter Tomboy as a text string, it will
automatically become a link to that page. A return link
can be created simply by typing Start Here anywhere in
the Tomboy page.
Access to notes is provided under the Tomboy icon on
the taskbar, and amazingly, the application remembers
the position on the screen that each note occupied just
as Nautilus does.
As well as the usual selection of text formatting
options, Tomboy also has a decent and fast search tool
to make note management easier.
In my opinion, Tomboy is reason enough to make sure
that whatever distribution you opt for, both Gnome and
Mono are installed. In the future, it may be linked up
with Beagle and F-Spot and other Gnome apps to
completely change the way we use the desktop.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 83

LXF70.tut_begin 83 11/7/05 9:57:43 am

 TUTORIAL Gimp
I grafted a dog’s head on to a
man’s body to show what an
important role transparency
plays in merging images. If
you missed the issue, call
0870 837 4773 or +44 1858
438795 for overseas orders.
84 LXF70 SEPTEMBER 2005
LXF70.tut_gimp 84
ILLUSTRATION & IMAGE EDITING
Gimp Fixing smiles
Michael J Hammel helps you brush up on masks, scissors, the Clone tool and
other Gimp techniques to create a dentist’s vision of the perfect smile.
A fairly easy way to get started using Gimp in a small
graphics business is to create print adverts for local
magazines and newspapers. Actually, most advertisers
don’t really know how to create graphics that draw eyes to an
advert that eventually leads to revenue for the business. You’ll
often find poor designs done by local print shops that crowd
images with too much text or bathe an image in a rainbow of
colours that cost an arm and a leg to print but end up bleeding
and fading on cheap paper. Picking out poor adverts and
approaching the advertiser with alternative designs can often
lead to the start of repeat business with a loyal customer.
This month’s project comes from just such a real-world
encounter. My wife created some adverts for a local catering
business, which happened to do a job for a dentist’s office party.
The dentist, just getting started in the area, needed to produce
advertisements of his own and contracted my wife to produce a
print ad for his office. The requirements for the ad included
using a picture of a male model sporting a big, bright smile. She
www.linuxformat.co.uk
found several potential images in stock image collections and
purchased the one I’m using here. I did the patchwork for her
to clean up the smile and she submitted the ad, along with
some other designs, to the client.
The client loved the modified image but in the end rejected
the ad in favour of a completely different design that removed
the smiling person completely (something she had
recommended from the start, mind you). Still, the modified
image came out so well that I felt it worth sharing the design
process with LXF readers.
This tutorial is more advanced than some of the others I’ve
done recently. You won’t be able to just follow along and do
what I do. The goal here is to put into practice what you’ve been
playing with all this time. At some point the wheels have to hit
the pavement, and this tutorial is a good example of how that
happens. What is more important is that this is what Gimp is all
about – doing real work. Eventually you have to stop exploring
your tools and just use them. Now’s the time.
8/7/05 4:11:29 pm
 TUTORIAL Gimp

THE PROJECT

To be fair, he’s not a bad-looking guy as he is, and the original

image was reasonable, but this ad was for a dentist’s, so
anything less than the perfect smile just wouldn’t do. There are
three main fixes required: clean and straighten the top teeth;
clean and straighten the bottom teeth; and adjust the bottom lip
to make it more symmetrical.
The bottom lip is the easier fix, but we can only do this after
we’ve fixed the teeth. This is because the lower lip will be
adjusted using the IWarp plugin, and we’re better off doing that
after the new, improved top and bottom teeth have been
merged into a single layer.
The upper teeth present three problems:
■ They aren’t symmetrical.
■ They aren’t even at their lower edges.
■ They aren’t white.
None of these will be especially difficult to fix, but it will take
some skill to create the initial selection. To make the teeth
symmetrical I’ll be using some large patches created from
selections. To fix the lower edges I’ll be using the Clone tool on a
separate layer. Cleaning the teeth needs to be done after
patching and cloning, and then only after merging any layers
that make up the visible upper teeth. Wait till you’ve done all
that before you brighten the teeth, and it should help guarantee
that you don’t end up with strange blotches where colour
changes weren’t uniform.

THE SOLUTION

1 2

1/ Use the Scissors for an accurate copy
The first thing we want is a copy of the upper teeth: we’ll use the Scissors tool
for this. This tool works by looking for natural edges between anchor points. In
this example I’ve clicked very close together, creating a large number of anchor
points – this enables me to create a selection that matches the teeth very
closely. We’ll be blending the copied teeth in with the original image, so we don’t
need an exact match here.
Once the outline is created (remember that the anchor points can be moved
before you create the selection), click inside the anchor points to convert them
to a selection, then save this selection to a channel.
2/ Create a new layer for the selection
Feather the selection by five pixels, then paste a copy of the selection in as a
new layer. The layer window will be snug around the copied selection, and you’ll
need some extra space to work in here, so expand the layer size a bit (Layer >
Layer Boundary Size). Turning off the background layer, you can see what you
need to work on! What I noticed most about this image is that the left-hand side
of the top row of teeth is a little better off than the right; so we’ll copy left to right >>
first, and work from there.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 85

LXF70.tut_gimp 85 8/7/05 4:11:34 pm

 TUTORIAL Gimp

>> 3/ Replace imperfect teeth

Create a rectangular selection around the left half of the teeth,
copy and paste the selection, convert the floating layer to a new
3
layer, then flip it horizontally. The selection doesn’t have to be
perfect – it gets blended into the other teeth with a mask
anyway. Then place the copy over the right-hand side of the
teeth, add a layer mask and blend the copy into the original
teeth using the mask. The two teeth furthest right are better
than the two furthest left, so we can mask out the copies of
those left-side teeth completely.
This image shows the flipped and masked layer and the
teeth with the masked copy overlaid on the original teeth. To
complete this process, we need to copy the right-hand side
teeth on to the left and merge with a mask again. The result is a
more balanced appearance to the shape of the teeth.

4 4/ Straighten out the smile

You should now merge these three layers (original copy of
upper teeth, right side flipped and left side flipped) together.
Employ the Clone tool to straighten the lower edge of the front
teeth. Pick a soft-edged brush to give weight to the lower edge
of these teeth, then a hard-edged brush to clone the straight
edge. Take the clone source point from just above where the
cloning is started and draw a straight line (hold down the Shift
key while dragging the mouse) with the Clone tool. Turn the
brush opacity down
slightly to add
blending to
the clone
process.

5/ Tidy up the bottom row

The same process is applied to the bottom teeth: 5
1 Selection.
2 Copy.
3 Paste to new layer.
4 Copy right side.
5 Flip.
6 Move to left.
7 And finally, merge.
You don’t have to do much to the bottom teeth, however,
because their cutting edge is hidden behind the top teeth. You
can also see that the amount of upper gum showing on the
left-hand side is greatly reduced, and the gap between the teeth
on the right is also reduced. On the lower teeth, the left side no
longer appears to be receded from the front of the mouth.
After all this, the alignment of the teeth is greatly improved.
Next we should attend to their colour.

86 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.tut_gimp 86 8/7/05 4:11:42 pm

 TUTORIAL Gimp

6 QUICK TIPS
Touching up images
■ Copy. Paste. Add Layer Mask… Work on layer
copies and then merge back into the original.
■ Blend, blend, blend. Layer masks, reduced
opacity on clone operations and soft-edged
brushes help merge bits and pieces of images
together seamlessly.
■ Colour in context. Colour adjustments in
portraits, especially those in flesh, bone or
body tones, should be subtle. Reducing
yellowing in teeth is a reduction in saturation
and increase in lightness.
■ Symmetry is ideal, but not real. Copying
left-to-right sides of the upper teeth in this
project helped provide some symmetry in the
shape of the teeth, but blending those copies
with the original helped provide enough
variation to keep the copy-and-flip from being
plainly obvious. Nature is nothing if not a wee
bit random.
■ Why did we save that selection? Because we
just don’t know when our experimentation with
copying, cloning and merging will work out just
right. It can take many attempts at trial and
error to get it right. And all of them – in this
6/ Adjust curves and saturation to whiten the teeth project in particular – start with getting a copy
So far I’ve focused on the teeth, ignoring the context of the picture. When adjusting the smile’s colour you need to
of the original upper teeth. I didn’t show that,
step back and look at it within the colour range of the whole image. If you don’t do this you risk making the teeth but now you know why you sometimes do
too bright, causing the effect to look fake. things that don’t seem necessary.
The yellow tint and dark stains in the teeth can be fixed in a number of ways. It’s easy to adjust the curves, but When you don’t do
in this case it would leave the top teeth too bright. We can adjust the bottom teeth with curves, though, so they those things,
have the same lighting appearance as the upper teeth, which should bring them out from the shadow and into line it’s inevitable
with the upper jaw. You’ll need to make only minor adjustments to the saturation (reducing it slightly to remove the that you’ll
stains and yellow tint), and brighten them up slightly to make them appear whiter. need them.
Perfection is achieved with one last tweak. The lower lip has slight bulges on either side. These are easily fixed
with the IWarp plugin after merging all the layers together. To get to IWarp, right-click on the canvas and go to
Filters > Distorts > IWarp.

THE RESULT

BEFORE

The result is obvious: a little

dental work can go a long way.
This tutorial shows how
important transparency is to NEXT
the Gimp artist, but we’ve also
used a number of simple MONTH
techniques to carry out this
project. Real-world AFTER Come with me on a trip
through text effects to go
beyond your everyday, canned
applications like this are a
logo effects.
good opportunity to practice
what you’ve learned, and we’ll
do the same next month.

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 87

LXF70.tut_gimp 87 8/7/05 4:11:49 pm

 TUTORIAL Emacs
GET MORE FROM EMACS TEXT EDITING SERIES
Emacs The custom buffer
PART 2 Biagio Lucini adds extras to your newly compiled text editor.
LAST
TIME
We went through a detailed
guide to installing Emacs from
source. We’ll use the version
in that tutorial (21.4) for the
rest of the series. If you
missed the issue, call 0870
8374773 or +44 1858
438795 for overseas orders.
Console prompt
By default, at each startup
Emacs loads some
initialisation files that
influence its behaviour by
assigning value to a bunch of
variables. If you think that
Emacs is not reacting to your
changes as it should, try
calling it from the command
line in the following way:
emacs --no-init-file --no-site-
file
where --no-init-file inhibits
the loading of ~/.emacs and
default.el, while --no-site-file
prevents the loading of site-
start.el. The meaning of
those files will be explained in
future instalments.
88 LXF70 SEPTEMBER 2005
LXF70.tut_emacs 88
You might feel inspired to use Emacs by this month’s
Roundup, and one of the many reasons to pick Emacs
as your editor of choice is its unbelievably high level of
customisability. Extending Emacs with options is a huge topic
that could easily fill a book, so my goal here is to introduce the
general strategy and basic concepts so that you can use them
to go further on your own. A minor point: to make them clear in
the text, I’ll present each option in upper case, such as
Reset To Saved, even if it’s written slightly differently as Reset to
Saved on screen.
OK, let’s start by opening up the Options menu.
The Options menu
A handful of the most used and most useful setup options can
be accessed through this menu, for quick customisation (Fig 1).
Here’s what they do in detail, from top to bottom:
■ Syntax Highlighting This gives you coloured keywords when
writing code or text in any supported language (basically, all
programming and markup languages are supported).
■ Active Region Highlighting When it’s turned on, this option
uses a different background colour to differentiate a region
marked for an operation such as cutting.
■ Paren Match Highlighting Particularly useful when
programming, this shows with colour whether the parenthesis
under the cursor is balanced correctly.
■ Truncate Long Lines In This Buffer Use this to prevent text
belonging to a single line from being pushed on to more lines,
independently of the extension of the line itself.
■ Word Wrap In Text Modes This is the reverse of the
previous option: whenever the characters in a line exceed a
default number, that line is automatically broken.
■ Case-Insensitive Search Select this to toggle between a
case-sensitive and case-insensitive word search.
■ Use Directory Names In Buffer Names Here, you can give
a unique name to buffers referring to files with the same name
located in different directories by adding the name of the
directory to the buffer name.
■ Save Place In Files Between Sessions If this is selected
when you reload a previously edited file, the cursor will return
to where it was when the previous session was closed.
■ Automatic File De/compression Decompress files on
loading then re-compress them on saving.
■ Enter Debugger On Error and Enter Debugger On Quit
These are mainly useful for software developers.
■ The Mule submenu The gate to customising the default
language, encoding and input methods.
Each option (excluding the submenus) has a box next to it.
To set an option is just matter of checking this box. However, by
default, these settings are lost on exit. If you want to save them,
www.linuxformat.co.uk
the solution is to click on Options > Save Options. This will
create a .emacs file (if not already there) in your home directory
where the desired features will be recorded, which is loaded at
startup by the editor.
Customising with Custom
The last entry in the Options menu opens the Custom buffer.
This provides an intuitive interface for tweaking just about every
aspect of Emacs. The buffer is navigable, pretty much like an
HTML page in a web browser, and highlighted boxes show what
look like hyperlinks. Various options are grouped according to
the general topic they refer to. A given customisable option can
be reached from within any of the groups to which it is
reckoned to be relevant. Apart from the Options menu, it is
possible to access the customisation buffer with the command
M-x custom
This opens the top-level customisation buffer, in which all
parent groups are displayed. A specific option is reached by
following a logical path through the groups down to the
appropriate entry. It can take some time to familiarise yourself
with this approach, but the level of personalisation it allows you
to achieve is a good reward for the time spent customising
Emacs in this way.
Each screen has a set of general options for setting and
saving globally the customisation you’ve performed in the buffer.
They are:
■ Set For Current Session This applies the new options to the
current session.
■ Save For Future Sessions Ensures the new values are the
default for future Emacs sessions and saves the options for the
current one.
Fig 1/ The Options menu with the tooltip box for Active
Region Highlighting. Note the different state of the check
boxes for options that are set (such as Save Places In Files
Between Sessions) from option boxes that are unset such
as Case-Insensitive Search).
8/7/05 5:37:23 pm
 TUTORIAL Emacs

■ Reset Sets all your currently edited options back to their
previous values.
■ Reset To Saved This sets all modified options back to the
previously saved values.
■ Erase Customisation Use this to reset all options in the
current buffer back to their default settings.
■ Finish Takes you to the previous Custom buffer (or the buffer
from which Custom was launched).
Clicking on the Status button next to each option opens a
drop-down menu that gives you access to those very same
options, but here you configure not the whole buffer but that
single option. Backwards navigation is provided by the Go To
Parent Group option, which highlights one or more parent
groups (depending on the current group). An editable option
can have several forms: if a value is required, you can just enter
it; if different or multiple choices are offered, they can be
entered by selecting the one you want from a drop-down menu YOUR
(if they are mutually exclusive) or by checking the box next to FEEDBACK
them (if more sub-options can be set at the same time). I’d welcome your comments
If you need help for the various options, click on the Show on this series and any Emacs
button (clicking on Hide will decrease the level of verbosity). questions. Send them either
The strategy for setting a value is straightforward: navigate the to lxf.letters@futurenet.
custom buffer to the option you want to change, then perform co.uk or directly to me at
biagio.lucini@gmail.com.
the wanted changes and finally press Save For Future Sessions
or Save For Current Session, according to whether you want the
modification to be temporary or permanent. Reset and Reset To
Saved can be used when the end result looks worse than the
initial configuration.
Work through the following mini-tutorial for an example of
using Custom.

SET THE LEVEL OF UNDO

1 2

1/ Go to the customisation window 2/ Identify the present level

Open a new Emacs window, type M-x custom and click on the Go To Group link Scroll down the window to the Undo group and click on Go To Group. You
after Editing Group. should see that the undo level is set at 20,000 actions, which is the default
value. The green text after the State button tells you that no modification has
been performed.

3 4

NEXT
3/ Edit the undo level 4/ Save as the default
MONTH
I’ll discuss advanced usage
Change the value from 20,000 to 30,000. The green text after the State button Click on Save For Future Sessions in the of Custom and give you
shows that the value has been edited. Also the global state of the buffer shows Operate On Everything In This Buffer session. alternative techniques for
that some options (only one in our case) have changed. Clicking on Finish will take you back to the customising Emacs.
Editing Group. If no more customisation is required,
the window can be safely deleted. LXF

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 89

LXF70.tut_emacs 89 8/7/05 5:37:26 pm

 TUTORIAL Perl
LESSONS IN PERL PROGRAMMING SERIES
Perl Sort your code
PART 2 Scared by abstruse Perl operators and regular expressions? Even more
scared of the word abstruse? Explain everything Marco Fioretti can.
In part 1 of this series we
defined and tested out Perl’s
variables, getting used to all
those $s, @s and #s. If you
missed the issue, call 0870
8374773 or +44 1858
438795 for overseas orders.
Calculate the size of an array
How do you know how many
elements there are in an array
or hash? Very simple: just
assign it to a scalar. Since the
latter can only hold one
value, Perl puts just the
number of elements into the
scalar. A similar trick works
with hashes. The keys
function returns an array
containing all and only their
keys, like so:
$HOW_MANY_JEDI =
@JEDI;
$HOW_MANY_ITEMS_INTO_
AN_HASH = keys %SOME_
HASH;
90 LXF70 SEPTEMBER 2005
LXF70.tut_perl 90
Before we go on with our journey through Perldom,
you’ll need to remember the three variable types of
scalars, arrays and hashes. Marked with a $, scalars
hold a single piece of information like a string. Arrays, which
begin with an @ sign, are just ordered scalars. Hashes, marked
unsurprisingly with a # sign, are groups of scalars defined by
other scalars referred to as keys. In this second tutorial we’ll
learn how to manipulate the most complex variables – arrays
and hashes – and then introduce the real black magic of Perl,
its regular expressions.
Sorting and listing
Remember how arrays order scalars? Perl list literals work in a
similar way. These are unnamed, ordered sequences of scalars
enclosed in parentheses. A good use of list literals might be to
assign values to two or more variables in a single instruction, or
to swap them around in some way.
($X,$Y,$Z) = ($Y,$Z,$X); # Circular shift
($Name,$Surname,$Phone) = (‘John’, ’Smith’,5556791);
($DARTH_VADER,@JEDI) = (‘Anakin Skywalker’, ‘Yoda’, ‘Obi-
Wan’, ‘Mace Windu’);
The first two lines are pretty self-explanatory. In the third,
the list on the left collates a scalar ($DARTH_VADER) and a
named array (@JEDI). Everybody knows what happens if we
assign it to the list on the right: to the Dark Side of the Force
young Anakin goes alone. This works because $DARTH_
VADER, being a scalar, can only hold one value. That value is
the first one out of our four Jedi knights. And since the only
other thing in the left-hand list is one array, @JEDI, all the other
knights in the right-hand list go there, in the same order.
Let’s now look at the splice() function, which is used to
delete, add or replace elements inside an array. First, we’ll
define some planets:
@STAR_WARS_PLANETS = (‘Naboo’, ‘Tatooine’, ‘Geonosis’);
Use splice() to add Coruscant and Alderaan right after Tatooine:
splice (@STAR_WARS_PLANETS, 2,0, (‘Coruscant’,
‘Alderaan’));
www.linuxformat.co.uk
In this line, the first argument is the array name: STAR_
WARS_PLANETS. Next comes the index (starting from zero!)
from which we wish to splice: in this case, Tatooine. The third
parameter is how many elements must be removed. Here I
don’t want to remove anything, just add more planets, so I type
zero. The last, optional argument is the list to be added in the
position we’ve just specified. If you don’t include this argument,
splice() wouldn’t add anything.
If you want to order things alphabetically, Perl has a sort()
function which, by default, considers all elements as strings,
even when they are numbers. So for example, if you type the
following code at the prompt:
perl -e “@A_LIST = (‘Dominions’, 180, 3, ‘10, Downing
St.’,Admiralty’); print join( \”\n\”, sort @A_LIST), \”\n\”;”
what you’ll get is an alphabetically-sorted list:
10, Downing St.
180
3
Admiralty
Dominions
sort() function can order elements according to other criteria:
@SORTED_LIST = sort AS_I_WANT @UNORDERED_LIST;
AS_I_WANT is a subroutine that takes two scalars as inputs and
returns -1, 0 or 1 depending on which of them comes first by the
desired criteria. We’ll look at subroutines later in the series.
HOW TO RUIN YOUR LIFE WITH
REGULAR EXPRESSIONS
Perl probably has more regular expressions than any other
computer language. To see with your own eyes how perverse, er,
powerful, they can be, check out the longest regex ever seen at
www.ex-parrot.com/~pdw/Mail-RFC822-Address.html. Rumour
has it that it validates email addresses, but I wouldn’t look at it for
too long. To write your own instead, the ultimate reference is
Mastering Regular Expressions by Jeffrey Friedl (O’Reilly, 2002).
11/7/05 12:01:24 pm

One last thing about arrays. There is one that no Perl hacker
can live without, even if it doesn’t really look like an array. I’m
talking about our beloved STDIN, the input text stream of every
well-behaved Unix program. Luckily, it’s very easy to use. I’m
mentioning STDIN here because, believe it or not, it can be
loaded into an array in a heartbeat:
@LINES = <STDIN>;
There. In just one instruction, you’ve loaded every line of input
text in one separate element of @LINES. Handy, isn’t it?
Pure hash
In part 1 of this tutorial I introduced hashes and their concept
of indexing scalars through other scalars (keys). As we know, a
hash consists of keys representing values. Once you have a
hash, it’s easy to manipulate it using just the keys, rather than
the underlying values. After all, if you wanted to get to those
values to put them in a fixed relative order, you would have just
used a regular array, wouldn’t you?
Unlike arrays, Perl hashes are not arranged in any specific
numerical order, nor in the order in which the keys are added.
The result of this is that hash values must always be addressed
by key instead of position. This is true even when you want to
delete a key and its associated value. The way to do this is with
delete(), which takes as its argument the key of the element
you want to get rid of. Using this function is necessary because
if you, for example, simply assigned an empty value to it
nothing would be removed. The element would still exist, just
with a null value.
As a matter of fact, sometimes the first thing to do with a
hash element is to find out if it exists at all, and if any value has
been assigned to it. To do this Perl provides two functions
named (who would have guessed?) exists() and define(),
which we use like this:
if exists ($STAR_WARS_ACTORS{‘Leia’}) { # do something...};
if defined ($STAR_WARS_ACTORS{‘Leia’}) { # do something
else...};
The first command is executed only if the hash includes a key
equal to the ‘Leia’ string, regardless of what the associated value
REGULAR EXPRESSION CHEAT SHEET
This is a list of the most common special characters found in Perl
regular expressions. Photocopy it and keep it close to your
keyboard; it will be a real time-saver.
. ...................... Any single character except a new line.
^ ..................... Beginning of the string.
$ ..................... End of the string.
* ...................... Zero or more of the previous character.
+ ..................... One or more of the previous character.
? ..................... Zero or one of the previous character.
\n .................... New line.
\t ..................... Tab.
\w ................... Digits and alphabet letters, regardless of case.
\W ................... Every character that is not a letter or a digit.
\d .................... Old fashioned digits: 0, 1 and so on up to 9.
\D .................... Everything but digits.
\s ..................... Whitespaces: space, tab, new line…
\S .................... Any non-whitespace character.
\b .................... Word boundary.
| ....................... Alternative between two values (eg A|B).
[] ..................... Square brackets delimit a character class.
() ..................... Normal brackets remember the enclosed substring.
Note: when you need to match one of the characters literally, for
example a plus sign, it must be preceded (escaped) by a backslash
to distinguish it from its role as a regex.
A+ # matches one or more capital As
\+ # matches one plus sign
\++ # matches one or more plus sign
www.linuxformat.co.uk
LXF70.tut_perl 91
TUTORIAL Perl
is. The second goes a bit further: it will be true only if there is a
key equal to ‘Leia’ and if its associated value has been explicitly
defined before.
Regular expressions
To manipulate great quantities of text, Perl has developed maybe
the most complete and powerful set of regular expression
facilities of all programming languages. A regular expression, or
regex, is a description of the structure of a string of text, made
with a custom syntax. The description is made up of regular
text and special metacharacters corresponding to the string’s
properties. Some of these metacharacters are listed in the
Regular Expression Cheat Sheet box, below left.
The beauty and ultimate purpose of all this is that if you can
describe any string in full detail you can also tell a script how to
find specific patterns and edit them automatically. This is best
explained with some examples:
/Jedi/
/\bJedi\b/
/^Jedi$/
/Jedi/i
/Jedi|Sith/
The first regex is true whenever a string contains the Jedi
substring, perhaps as part of a longer word. If you only want to
find ‘Jedi’ when it appears as a complete word, enclose it
between word boundary markers (\b) as in the second
command. The third regex is even more precise: it will be true
only when ‘Jedi’ is both at the beginning (^) and at the end ($)
of a line, or in other words when ‘Jedi’ is the only word of that
line. A Perl regex is case-sensitive, so if you don’t feel like
typing JEDI|jedi|JedI and so on, use the i modifier, as in the
fourth line here, for a case-insensitive match. Finally, the last
command is true whenever Perl finds either ‘Jedi’ or ‘Sith’.
All this is worth knowing because when Perl recognises the
text pattern described by a regex it can take any actions you
want or modify the corresponding string according to your
instructions. Here are the two corresponding formats:
if ($STRING =~ m/some regex here/) {do something}
$STRING =~ s/some regex here/some other text pattern/;
The actual regex is delimited by slashes. Strings are
associated to it by the =~ operator. When the slashes are
preceded by an m character it means, ‘does the string match
this regex?’. When there is an s instead of an m and some other
text between slashes right after the regex, it means ‘take
$STRING and, inside it, substitute the regex with whatever is
written between the two last slashes’.
Regexes can contain scalars and store their results in special
variables. This makes them an extremely flexible tool.
$JEDI = ‘Anakin’;
s/Master $JEDI/The future Darth Vader/g;
s/Master (Obi-Wan|Yoda)/The Jedi Knight $1/;
Here we start by saying that all the occurrences of ‘Master
Anakin’ (Anakin because Perl will substitute the $JEDI variable
with its current value) must be replaced by ‘The future Darth
Vader’. The g modifier at the end means globally: without it
only the first match would have been replaced instead of
all of them.
The second regex shows another really neat feature. It
finds a match whenever one of the two Jedis mentioned is NEXT
called Master. As the names are in parentheses, they are
not forgotten when found, but saved in the special MONTH
The flow of a Perl script, how to do
variable $1. Therefore, the same regex will substitute
‘The Jedi Knight Yoda’ for ‘Master Yoda’ or ‘The Jedi Knight file I/O, and techniques for
presenting data in the most
Obi-Wan’ for ‘Master Obi-Wan’. If more matches must be
readable way.
remembered they are associated in the same way as $2,
$3 and so on until $9. LXF
LXF70 SEPT 2005 91
11/7/05 12:01:27 pm
 TUTORIAL Subversion
VERSION CONTROL
Subversion
Branches, tags and mergers
PART 3 Graham Morrison explains how to manage your repository once development starts to
expand.
In LXF69’s tutorial we looked
at downloading, modifying,
uploading and analysing work
using Subversion as a client. If
you missed the issue, call
0870 8374722 or +44 1858
438794 for overseas orders.
92 LXF70 SEPTEMBER 2005
LXF70.tut_sub 92
This is the final instalment in our three-part tutorial
dedicated to the Subversion version control system.
After covering the basics of setting up and
administering a server and using Subversion from a client’s
perspective you should now have a good idea of some the
benefits that using Subversionn can bring to your project; but if
you do take that leap, it won’t be long before you need what
we’re about to cover in this month’s final instalment, which deals
mainly with ‘branches’.
I’ll be using some of the concepts from last month to
introduce some good server practices. These will have an
impact on how you run your Subversion repository, but it’s all
going to be done using the client. We’ll be drawing on the
lessons learned in the two previous tutorials, so it might be a
good idea to refresh your memory before continuing with the
final instalment.
Arboriculture revisited
Real trees (the kind that grow in forests) have branches.
Branches grow off the main trunk of the tree, and so do those
of the Subversion variety. But there is a difference: some
www.linuxformat.co.uk
Subversion branches can go on to become the main trunk of
development, which would make a real tree fall over. Subversion
branches can strengthen the development process rather than
unbalance it.
There are numerous reasons for creating a branch from the
main development trunk. The most common reason is to
facilitate development of a new version, while still allowing for
critical updates to be made to a previous one. To put this into
context, every major release needs to be updated. The KDE 3.4
release, for example, was followed by an update called 3.4.1.
This fixed many bugs and added several translations, but it didn’t
add any new features or functionality. Those are reserved for
any upcoming major release.
Using two separate branches makes it much easier to
maintain both a stable release version and a development
version. Bugfixes can be applied to both, while new features
need only be applied to the development version; significant
changes can be merged back into the stable version. This
means developers can forge ahead with new ideas using the
development branch, safe in the knowledge that their work
won’t affect the stability of the release version.
11/7/05 11:35:44 am

PART 1 – MAKING A BRANCH
Look at a branch as nothing more than a copy of the main
trunk made at a certain point. There’s nothing stopping you
using a simple local copy to implement this but there’s no need.
Subversion treats a branch as nothing more than a copy, albeit
with an initial history common to that of the trunk, and this is
reflected in how you create a branch.
For these examples, we’re still using the simple ‘helloworld’
example from the previous two months. This simple project is
nothing more than a small source file (helloworld.cpp) and a
corresponding Makefile. Both files are currently in a single
directory, but as we’re moving to a branched development cycle,
they need to be moved to their own branch.
$ svn mkdir branches
A branches
$ svn mkdir branches/stable_1_0
A branches/stable_1_0
$ svn mv helloworld.cpp branches/stable_1_0/
A branches/stable_1_0/helloworld.cpp
D helloworld.cpp
$ svn mv Makefile branches/stable_1_0/
A branches/stable_1_0/Makefile
D Makefile
We’ve just created a branches directory in our local working
copy, followed by another called stable_1_0 to hold the stable
development branch. We then moved the two source files into
stable_1_0 in preparation for taking a branch from the stable
directory. The next stage is to copy the stable_1_0 branch
directory to the directory we’re going to use for the unstable
development branch, but we first need to commit the previous
changes before Subversion will allow us to copy a directory.
$ svn commit -m “Created new branch structure.”
Deleting Makefile
Adding branches
Adding branches/stable_1_0
Adding branches/stable_1_0/Makefile
Adding branches/stable_1_0/helloworld.cpp
Deleting helloworld.cpp
Committed revision 5.
The -m added to the commit command lets you enter your
comments without Subversion having to launch an editor. Now
that the previous revisions have been committed, we can make
a development branch by copying the stable_1_0 directory:
$ cd branches/
$ svn copy stable_1_0 HEAD
A HEAD
$ svn commit
Adding branches/HEAD
Committed revision 6.
In our example, we’ve used a copy of the stable branch to
create a development branch suitable for adding features and
fixing problems with. As this is at the ‘head’ of the development
effort, we’ve called the branch HEAD.
Managing access
Creating branches is easy; the most important part of branching
is how you manage them, which is usually down to project
policy. Now that the branches have been made, any developers
joining the project would work on the stable branch for bugfixing
and the HEAD branch for adding new features.
If security is a concern, you may want to restrict access to
specific branches to certain people, or only allow certain people
to create branches in the first place. If you only want to grant
certain developers the ability to modify specific branches, you
www.linuxformat.co.uk
LXF70.tut_sub 93
TUTORIAL Subversion
Tag 1.0
Branch 1.1
1.0 1.1 1.2
Subversion history
need to be using Apache for repository access. The reason for Branches and tags are
this is that the best way to provide per-user access rights is by taken from the main
installing the mod_authz_svn Apache module. development trunk.
A tag can’t be modified,
A new working copy of the development branch can now be but is just a snapshot of the
checked out by downloading from the HEAD directory. trunk taken at a certain
$ svn checkout file:///usr/share/subres/branches/HEAD point in the development
A HEAD/helloworld.cpp process (to record
significant releases, for
A HEAD/Makefile
example). See page 95.
Checked out revision 6. A branch, on the other
The history information for the HEAD directory will only go hand, is a snapshot that
back as far as the branching. But the history for the files within can be worked on, either
each branch will have been brought forward from their previous independently of the trunk
or with the intention of
location. The difference now is that the helloworld.cpp has merging parts back into
been copied into two separate directories (the two branches). the trunk.
The history for helloworld.cpp will also branch to reflect the
changes made to each file individually. You can easily check this
using the log command – here is a truncated version of the
output from svn log helloworld.cpp in the HEAD branch:
r7 | Added a cutting-edge feature.
r6 | Added HEAD development branch.
r5 | Moved project into a branch structure
r4 | Resolved conflict by incorporating both changes.
You can see that helloworld.cpp has inherited the history
from before the creation of the branch at r6 (revision 6).
Depending on how a project is organised, the process of
adding new features to a project’s development branch can
include bug- and stability fixes that should be ported to the
stable branch. For this, we need to merge the changes into the >>
stable branch.
BUBBLE-UP
Subversion stores every single revision as a new Changes are managed through a process
filesystem tree, creating a logical copy of all the known technically as bubble-up. Every change is
files in a repository – mostly a series of links to copied to a new file – the start of the bubble.
the previous version. This is what makes Each revision is actually a complete file tree,
Subversion so efficient: it doesn’t copy made up of copied versions of any changed files,
everything, only the changes from one revision to using links to previous files that haven’t changed.
the next. Subversion creates a new revision link between
In Linux terminology, it’s a similar concept to a the edited file and its parent directory. From
hard link. A hard link looks and behaves just like a here, the links continually bubble up through the
file, but is just a connection to the location of the file tree until the process reaches the root, at
file on disk. In fact, filenames themselves are which point the new revision is complete.
hard links, as they don’t contain any document By comparison, CVS needs to read every file in
data – just a name and a link. the directory tree whenever there’s a branch.
LXF70 SEPTEMBER 2005 93
11/7/05 11:35:50 am
 TUTORIAL Subversion
>> PART 2 – MERGING ONE BRANCH WITH ANOTHER
Working on the development branch of a project often
involves solving problems that should be applied to the stable
branch, especially when working with security problems. In our
simple helloworld.cpp example, the development branch has
been modified to include another output line containing ‘a
cutting-edge feature’. Of course, changes will be significantly
more complex in the real world – but the principle remains
the same.
Despite their sharing the same ancestry, Subversion sees the
two files as totally separate. Last month we used svn diff to
check the differences between various revisions of the same file.
This time we need to rationalise the differences between the
same file but now copied to two branches. For this we need svn
merge to apply the differences between two sources. First, from
the stable working directory, we need to check the differences
against those changes made within the unstable HEAD branch.
If we examine the log of helloworld.cpp from the stable
branch, it’s pretty obvious that none of the changes that have
been made to the development branch:
r5 | Moved project into a branch structure
r4 | Resolved conflict by incorporating both changes.
We’re missing revisions 6 and 7 from the HEAD branch. As
you can see from the previous log messages from HEAD,
revision 6 was the process of copying the files to the new
branch, while revision seven was the addition of some fantastic
new feature (r7 | Added a cutting-edge feature).
You can see the differences between the two revisions with
the svn diff command in the next column:
“BRANCHES ALLOW A NEW
VERSION TO START WHILE THE Rewind
OLD ONE CAN STILL BE UPDATED.”
The code within a development tree is inherited by each successive version.
[1] The original release (1.0).
[2] Updates to the original release should just add fixes (1.1).
[3] Features are reserved for a significant revision update (2.0).
2 held in the repository.
1 3
94 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk
LXF70.tut_sub 94
$ svn diff -r 6:7 file:///usr/share/subres/branches/
Index: HEAD/helloworld.cpp
=================
--- HEAD/helloworld.cpp (revision 6)
+++ HEAD/helloworld.cpp (revision 7)
@@ -6,6 +6,7 @@
{
cout << “Hello World!” << endl;
cout << “Both modified additions.” << endl;
+ cout << “Cutting edge feature.” <<endl;
return(0);
}
The only difference is the addition of the “cutting edge
feature”, signified by the + symbol at the beginning of the line.
As we saw last month, we could use the svn diff output to
make a patch. But there’s no need with Subversion, as you can
use merge to apply the differences to your local working copy
immediately. From a local working copy of the stable branch,
svn merge will add only the changes specified by the revision:
$ svn merge -r 6:7 file:///usr/share/subres/branches/HEAD/
helloworld.cpp
U helloworld.cpp
$ svn status
M helloworld.cpp
Our code tells us that the changes made to the HEAD/
helloworld.cpp file are merged into the local copy of the same
file, represented by the M in the first column of output from svn
merge. The developer is now free to examine the changes
made to the helloworld.cpp file, and commit them to the
stable branch. There is a chance that there is going to be a
conflict between the merged and original version, so take care
when merging between branches.
One unconventional side-effect to using the revision numbers to
specify which changes are to be merged is that you don’t have
to use them to update from one revision to another. You can
also turn them around – using 7:6 instead of 6:7, for example,
has the effect of winding back any revision 7 changes back to
revision 6. Using the previous example, we would type:
svn merge -r 7:6 file:///usr/share/subres/branches/HEAD/
helloworld.cpp
G helloworld.cpp
The G is to show that Subversion has successfully merged
the repository’s changes into the local file. This is a good
moment to mention the svn revert command, which is a much
safer method of reverting any local changes back to the version
Another neat trick you can perform with branches is to
switch the branch that your local working copy references on
the server. The command for achieving this magic is svn
switch. In fact, it doesn’t actually do anything all that clever – it
just changes the URL that your working copy references. You
can see the current URL for your working copy using svn info:
$ svn info
URL: file:///usr/share/subres/branches/stable_1_0
From the above example, we can change our working copy
branch from stable_1_0 to HEAD using svn switch:
$ svn switch file:///usr/share/subres/branches/HEAD
U helloworld.cpp
Updated to revision 7.
$ svn info
URL: file:///usr/share/subres/branches/HEAD
11/7/05 11:35:52 am
 TUTORIAL Subversion

REMOTE ACCESS THE PROJECT WITH SVNSERVE

The first tutorial showed you how to
configure the Subversion server to use an
Apache module for remote access. But we
have also used the svn:// protocol URL for
accessing remote servers. The server
application responsible for servicing this
protocol is called svnserve, and is usually
part of a typical Subversion installation. It’s
not used by default, because it’s meant for
smaller and less demanding access, but it is
a worthwhile alternative to Apache.
Getting the server running is actually
incredibly straightforward: you just need to
execute the command with a couple of
parameters – the operating mode and the
path to the repository. For example:
svnserve -d -r path_to_repository
This runs the server as a daemon (the
other operating modes use Internet
Daemon, AKA inetd), and points to the
repository with the -r parameter. Once the
server is running, you can access your
repository directly using the svn protocol:
svn co svn://localhost
By default, svnserve provides only
read-only anonymous access to the
repository. You can change this by editing
the svnserve.conf file located within the
repository’s conf directory. The
configuration file is well documented,
and you can revoke or allow anonymous
write access under the [general] section.
To add per-user permissions, enable it
in the Auth-access field, point
password-db to the location of a
password file and create the password
file yourself. For example:
svnserve.conf:
[general]
anon-access = none
auth-access = write
password-db = passwd
passwd:
[users]
graham = grahampassword
You will then be able to access the
repository as a user by specifying the
username before the server address and
entering the password, as in installed, and a user account on the server Once svnserve is running
svn co svn://graham@localhost machine. Svnserve is run as the connecting you can access the server
The only problem with the above method user, so there’s no need to have it running with Konqueror.
is that passwords are sent as plain text, and beforehand, but the client needs to specify
therefore pose a security risk. The answer is the location of the repository on the server
to use the ever-versatile SSH to harden the filesystem. For example:
connection between the client and the svn co svn+ssh://graham@localhost/usr/
server. All you need is to have OpenSSH share/subres

PART 3 – VENDOR RANCHES AND TAGGING

Vendor branches let you integrate third-party projects
into your project tree, like external libraries that might provide
essential functions for your application. By using a vendor “SCRIPTS MAKE IT EASY TO UNDO
branch you can keep on top of any changes in these external
projects, but (more importantly) you can also make sure all your
CHANGES MADE ACCIDENTALLY
developers are using the same version. CVS has specific support
for vendor branches, but Subversion is versatile enough to
TO THE TAGS BRANCH.”
integrate them without too much difficulty.
A vendor branch usually exists as a separate folder structure To create a tag from our previous example:
under the root of the Subversion repository. This is often under a $ svn mkdir file:///usr/share/subres/tags
directory called vendor – hence the ‘vendor branch’. You need Committed revision 8.
to import the entire third-party project into this folder. Whenever $ svn copy file:///usr/share/subres/branches/stable_1_0 file:///
the third-party project releases a new version, you need to usr/share/subres/tags/release_1
Structure your repository
merge these changes against a local working copy so that your Committed revision 9.
The way the filesystem is
own changes won’t be lost. You can then commit these changes There’s only one problem with this approach: there’s nothing structured is is entirely up to
back to the repository in order to update the version of the to stop a developer changing the contents of the tagged the project maintainer, but it’s
third-party application for other developers to use. directory. Most of the time this won’t be a problem – it’s usually worth bearing a few things
project policy to leave the contents of the ‘tags’ directory alone. in mind.
Playing tag But there is the option of hardening access if you need to. If you’re developing
To be able to use vendor branches effectively, you need to mark This takes us right back to the first episode in our Subversion different applications within
the third-party branch to show it can’t be modified. This is series (LXF68), because you can solve this problem using the same repository, you’ll
probably want to separate
known as ‘tagging’ in version-control speak, and to Subversion, Subversion hooks. In case you can’t remember, these are scripts
them at the top level into
it’s simply a branch that shouldn’t be edited. Like a branch, a tag that are executed when a specific Subversion event is triggered.
different directories. You also
is a complete copy of the repository at a certain point in time. Using a script in this way makes it easy to undo any changes
need to consider how you’re
This may seem like any revision point within the repository – that may accidentally have been made to the tags branch. You going to manage branches
and it is – but creating a tag is often a good way of marking a can also restrict access using the mod_authz_svn, as and tags. Most projects use a
specific point in the development cycle. mentioned earlier. folder called trunk for the
One of the main reasons for creating a tag is to mark a And that’s it! The end. C’est tout. Je suis un sandwich! After main development source
significant release version, such as the stable_1_0 in our this brief series of three tutorials, I hope you now have the tree, and create separate tags
example. Just like branches, tags are nothing more than copies confidence to not only run your own server, but also to and branches directories at
of the repository, and it’s the svn copy command we use to contribute to a project. If the project is a success, you will no the same level. It doesn’t
have to be this way of course
create them. The only differences are that the comment for the doubt need to merge and manage changes as we have
– the filesystem structure you
revision should reflect the creation of the tag, and that covered this month, hopefully bringing the whole development
use is left entirely up to you.
developers mustn’t edit the tagged development branch. cycle full circle. LXF

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 95

LXF70.tut_sub 95 11/7/05 11:35:53 am

 TUTORIAL PHP
SCRIPT PROGRAMMING
PHP The Monty Hall problem
Having problems with last issue’s LXF bounty? Paul Hudson shows you how to prove such problems
using brute force, a dash of brain power and a highly-trained team of goats.
Last month was all about
curses – the tricky but
rewarding graphical library. If
you missed the issue, call
0870 837 4773 or +44 1858
438795 for overseas orders.
“THIS IS A RELATIVELY SIMPLE
QUESTION THAT HAS CAUSED
MANY PEOPLE GREAT FEAR.”
96 LXF70 SEPTEMBER 2005
LXF70.tut_php 96
Anyone who has studied advanced mathematics has a
deep – if somewhat unconventional – appreciation of
numbers. We lovingly bestow names on certain types
of numbers to show our admiration for their magic: harmonic
numbers, perfect numbers, semi-perfect numbers, hyperperfect
numbers, amicable numbers and even, believe it or not, weird
numbers. We love them because they’re predictable,
unchangeable and exact, and therefore can be relied upon for
logic, building complex proofs and pretty much anything else
except paying the bill when you take one out to dinner.
In the rare times when I venture into the outside world, it
has come to my attention that the great unmathed do not have
the same respect and love for numbers that the rest of us do.
Numbers scare them, and, as we all know, fear leads to anger,
anger leads to hate and hate leads to suffering. How wrong they
are! However, we can at least glimpse their irrational terror: early
in your numerical romance you too may recall being
overwhelmed by their immense strength, or bamboozled by
their exotic attractions.
Take, for example, the Monty Hall problem. This is a
relatively simple question that has caused many people great
fear – until they come to terms with the magic of numbers. It is
named after Monty Hall, the host of an old US game show
www.linuxformat.co.uk
called Let’s Make a Deal, where something similar to this
problem was done on most shows. The problem goes like this:
you are on a show, and asked by the host to choose one of
three doors. Behind one door is an expensive car – the star
prize – but behind the other two doors are goats. Once you’ve
picked a door randomly, the host (who knows which door leads
to the prize) opens one of the other two doors to reveal a goat.
He offers you the chance to stick with your chosen, unopened
door, or change to the other unopened door. Do you switch, or
do you stick with your original choice?
The answer – naturally – is that you should always change,
as there is a greater chance that the other door contains the
car. This issue we’re going to write a PHP script that proves the
correctness of the ‘switch’ decision in the Monty Hall problem
and, once that’s done, I’ll add some options for raising the
difficulty a little.
My brain hurts
Before we write any code, it’s important to understand why the
Monty Hall problem (MHP from now on) causes people such
anguish. There are a few other things we should also nail down
now to avoid confusion: “win” means to pick the door with the
car; “lose” is to get one of the two doors with a goat; the doors
are numbered 1, 2 and 3; “host” is the game show host; and
“contestant” is, well, the game show contestant – duh. With that
out of the way, let’s analyse the problem. In MHP you get to
pick a door to win a prize, and because there are three doors
and cheating is ruled out we can safely assume that there is a
one in three chance of winning.
Say you were to choose door 1. At this point, there is a 1 in
3 (1/3) chance that you have the correct door. The host knows
11/7/05 9:58:10 am
 TUTORIAL PHP

where the car is, and so opens door 2 to reveal a goat. This If I have a 2/3
leaves door 1 and door 3 closed. It’s at this point that most chance of winning a
people go wrong, because they will now believe that there’s an goat by sticking, then I’m
sticking – I want a
equal chance of door 1 and door 3 containing the car. “After all,”
goat!
they will explain, “there are two doors, no clues or cheating, so
we can just follow the basic rules of probability – it’s just the
same problem again with one fewer door.”
Of course, that’s a whopping great big fallacy – the basic
rules of probability won’t help you much here. Instead, you need
to know about the rules of probability when something else has
already happened. If you pick door 1, you know that there is a
1/3 chance of having picked the door that leads to the car.
However, it is equally correct to say that there is a 2/3 chance
that the car is behind one of the other two doors.
Now, when the host opens one of the other two doors (we’ll
say he opens door 2), there is still a 2/3 chance that the car is
behind one of those two doors, except now you know that one
of them contains a goat. This effectively gives door 3 a 2/3
chance of holding the car. So, if door 1 has a 1/3 chance of
leading to the car, and door 3 has a 2/3 chance, what do you
do? Switch, of course.
This is shown in the diagram below right. If we presume you
pick door 3, there are three possible outcomes. In possibility A,
door 3 contains a goat, and the host will open door 2. If we
stick with door 3 we lose, but if we switch to the other door
(door 1), we win. In possibility B, door 3 contains a goat, and
the host will open door 1. If we stick with door 3 we lose, but if
we switch to the other door (door 2), we win. In possibility C,
door 3 contains a car, and the host will open either door 1 or
door 2 – it doesn’t matter, so let’s say he opens door 1. That
leaves door 3 (our choice) and door 2. If we stick with door 3
we win, but if we switch to door 2 we lose. So in possibilities A
and B, we need to switch to win; in possibility C we lose if we
switch. Switching, therefore gives us a 2/3 chance of winning,
and sticking gives us a 1/3 chance of winning.

Cracking the code

The way to empirically prove this is to write a script that opens
doors thousands and thousands of times, switching and not
switching, to see what results we get. This calls for a PHP script
that accepts user input (whether to switch or not; how many 1-3 are the doors, and
attempts we get), runs the calculations, then prints the results. A-C are your choices.
In terms of code, we need three parts: a function that runs
a single Monty Hall problem; something that parses user input,
runs the problem multiple times, tracks results and prints them
1 2 3
out; and some HTML that allows users to specify the test to run.
The whole thing comes in at under 100 lines, but we’re going to
take it in chunks, starting with the easy bit – the HTML:
<form method=”post”>
<p>Number of iterations: <select name=”Iterations”>
A
<?php
/// generate all these using a loop, rather than lots of typing
for ($i = 50; $i <= 10000; $i += 50) {
if (isset($_POST[“Iterations”]) && ($_POST[“Iterations”]
B
== $i)) {
print “<option value=\”$i\” selected>$i</option>”;
} else {

}
print “<option value=\”$i\”>$i</option>”; C
}
?>
</select></p> <option value=”No” <?php if (isset($_POST[“DoSwitch”]) &&
<!-- This sets the box to “No” if that was its previous value -- ($_POST[“DoSwitch”] == “No”)) print “selected”; ?>>No</
> option></select></p>
<p>Should the contestant switch? <select <p><input type=”submit” value=”Prove it!” /></p> >>
name=”DoSwitch”><option value=”Yes”>Yes</option> </form>

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 97

LXF70.tut_php 97 11/7/05 9:58:15 am

 TUTORIAL PHP

>> OK, so I lied a little bit: there’s a bit more to it than just the initial score to zero, then run the function multiple times
HTML, but it’s definitely the easiest part of the script. There are (passing in the should the contestant switch doors?
two comments in there to explain what’s going on, but it’s all variable). Each time the function runs, we catch the return value
very easy: it prints out a form for the user to choose how many and alter the scores appropriately. Still with me?
times we should execute the problem, and whether the Now on to the serious code: the MontyHall() function itself.
contestant should switch doors or not. As you’ve just seen, we need to make this accept a parameter
Moving on, we need to handle the form we’ve just written to store whether the user wants to switch or not, but that’s the
using PHP, then run the problem a suitable number of times. easy part. Here’s the rest:
Here’s the code for that: function MontyHall($DoSwitch) {
if (isset($_POST[“DoSwitch”])) { /// make an array of the doors, then randomise its order
/// only run this if the form has been completed $doors = array(1=>”Goat”, 2=>”Goat”, 3=>”Car”);
$wins = 0; shuffle($doors);
$losses = 0; /// pick a door, then remove it from the $doors array
/// run the function many times. This returns either “Win” $our_choice = array_rand($doors);
or “Lose” $our_door = $doors[$our_choice];
for ($i = 0; $i < $_POST[“Iterations”]; ++$i) { unset($doors[$our_choice]);
$result = MontyHall($_POST[“DoSwitch”]); /// now to “open” a door
/// keep track of the results /// get the first remaining door, and check whether it’s a
if ($result == “Win”) { goat
++$wins; if (reset($doors) == “Goat”) {
} else { // first door in the array contains a goat,
++$losses; // eliminate it from the list of choices!
} // get the current key
} $cur_key = key($doors);
/// print out the results unset($doors[$cur_key]);
print “<h1>The Amazing Monty Hall Empirical Proof } else { // first door not a goat, so remove the next door
Device</h1>”; next($doors);
print “<p>Number of iterations: {$_POST[“Iterations”]}</ $cur_key = key($doors);
p>”; unset($doors[$cur_key]);
print “<p>Contestant switches: {$_POST[“DoSwitch”]}</ }
p>”; /// should the contestant switch?
print “<p>Number of wins: $wins</p>”; if ($DoSwitch == “Yes”) {
print “<p>Number of losses: $losses</p>”; /// yes, switch. Is the remaining door the car?
} if (reset($doors) == “Car”) {
Again, very easy stuff: if the form has been submitted, set // yes - we won!
return “Win”;
} else {
CALLING ALL SUDOKU SOLVERS! // a goat :(
We started a coding bounty last month that columns, etc. Anyone can write a program that return “Lose”;
asked people to create programs to solve Sudoku guesses: it takes skill to write a group of }
puzzles. Sadly, no one here at LXF Towers is algorithms that mimic human behaviour. } else {
allowed to enter (we are the judges, after all!), ■ Once you have a script that’s able to solve // contestant shouldn’t switch
but I can at least give you some tips: problems in a human-like manner, you can get if ($our_door == “Car”) {
■ You must create a GUI for your app. That’s down to the business of making it generate // we picked the right door first time
fine: PHP makes GTK GUIs very easy. If you don’t puzzles. This is important, because you need to
return “Win”;
have a complete set of LXF back issues (why be able to a) pass those puzzles through your
} else {
not?), then you missed an 8-page special tutorial solver to make sure that it works, and b) grade
// another goat! :(
on how to use PHP to create GTK apps back in the puzzles. By solving the puzzles like a real
LXF40. We’re lovely people here, so we’ve put a person would, your solver can count the steps it return “Lose”;
PDF of the tutorial on to our coverdiscs this took – the more steps, the harder the puzzle. }
month so you can swot up. ■ Use Glade, and you’ll be able to make much }
■ Don’t bother creating a complex file format. nicer interfaces. }
If you have PHP 5, you should be using XML, with That’s enough hints for now: using PHP you Again, there are comments throughout that should explain
SimpleXML to load and save. have a big advantage over coders in tougher everything – I even used meaningful variable names, so there’s
■ Think about how humans solve Sudokus. We languages, so the best way to win the puzzle is to no excuse if you don’t get it! The one area where you might get
work by getting the easiest answers first, then spend more time thinking about the algorithms confused is in the handling of the $doors array. In the code,
looking for twins, marking possible answers on a than you do actually programming. Good luck,
I’ve used reset(), next() and keys() – functions that aren’t used
scratch pad, trying to complete rows and and if you win, I want a cut of the prize money!
all that often. The first of them moves the array cursor (the

“THE WAY TO PROVE EMPIRICALLY THAT MONTY HALL

WORKS IS TO WRITE A SCRIPT THAT OPENS DOORS
THOUSANDS AND THOUSANDS OF TIMES.”
98 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.tut_php 98 11/7/05 9:58:17 am

 TUTORIAL PHP

HOME IMPROVEMENT
It’s time to separate the sheep from the goats. If this
has been enough programming for you, then go off and
enjoy the summer of numbers with your pals. But if
you’re a real programmer, you can think about how to
improve the script.
There are at least three changes you should try out to
flex your coding muscle:
■ Fun… Make the script run n number of tests with
switching turned on, then n number without, to save
the user having to toggle the select box. These results
should be kept separate on the screen.
■ Tricky… Our present script makes no attempt to print
out any sort of result percentages based on the
number of wins and losses. For simple values of n this is
quite easy – if there are 1,000 test runs you can see
what’s going on at a glance. But if you do 19,350 runs,
let’s say, the answer is less immediately obvious.
Calculate the win and loss percentages and display
them in the output.
■ Taxing… MHP extends above and beyond a three-
door problem. We could theoretically have a hundred
doors, with 99 goats and one car – the host could close
98 doors after we choose ours, and we’d have a 99/100
chance of winning if we switched. This can’t be tested
with our current script, but you could do it by adding an
extra form element to let people choose the number of
doors, passing that element to the MontyHall() function,
constructing the array with many goats and adding a
single car to the end.
■ Mayhem! Assume the game show host doesn’t know
which door contains the car – what happens to the
chances of winning now, and why? Answers on a
postcard to the usual

internal pointer that tells the array which element is being read)
to the start of the array, and returns the value of that element.
So our line if (reset($doors) == “Goat” means ‘go to the
start of the $doors array, get its value and if it is “Goat”, the
condition is true’. Then we use the key() function, which returns
the key of the element that is currently pointed to by the cursor
(ie the first one, because we just used reset() to move to the
beginning), and uses that value to remove the element. This is
quite a big hoop to jump through, but remember that we don’t
have any other way of indexing in to the array because we don’t
know which value we removed. If the value of the first $doors
element is not “Goat”, we remove the second element by
calling next(), which moves the array cursor forward one place.
Top and tailing
We now have one door in $our_door, and one element in the
$doors array. The remaining question is: should the contestant
switch? Of course, this was passed in as a parameter, so we just
check that and act accordingly. So, if we’ve opted to switch, we
return “Win” if the value in $doors is “Car”, and “Lose”
otherwise; if we’ve selected not to switch then we return “Win” if
the value in $our_door is “Car”, and “Lose” otherwise. Simple.
You’re not done just yet: you should add some top-and-tail
HTML code to present it nicely, but more importantly you
should also add a call to srand() near the top of the script.
Without this, PHP might not bother randomising things much,
and your results are likely to be very predictable between runs.
Now all that remains is to go ahead and run the thing using
your web browser of choice – copy the script from your home
directory to /var/www/html (this may vary according to your
distro), and go for it. You may want to compare your script
against mine, just to make sure it’s in the right order.
That’s it for this month, but hopefully you’ve learned a bit
about numbers and probability, a bit about PHP’s array
functions, and a lot about how PHP can be used to test all
possibilities in a problem to find the right answer. What’s more,
we did all that without me having to goat to town on goat
jokes. Miraculous.
Many people disagree with the solution of switching their
decision when they’re shown the first goat; and of course they
are wrong and should be mocked publicly. But if the
unenlightened are people you actually care about, it’s much
better to show them the error of their ways and spread some
number love. So use this script to prove to your friends just how
Monty Hall works – and why not slip them a copy of Abbott’s
Flatland to show them what they’re missing? LXF

DOOR 2

NEXT
MONTH
Next month I’ll be sneaking in
more hints on how to win our
Sudoku competition, by
covering file handling
A car is yours if you can find it: with XML.
do you switch doors or stick?

LXF70.tut_php 99 11/7/05 9:58:18 am

 ANSWERS

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

OUR EXPERTS The magic’s gone

Q
I recently installed
Whatever your question is, we can Fedora Core 3 on to my
find an expert to answer it. From Dell 510M laptop, in a
installation and modem woes to
network administrations, we will get dual-boot setup with
the answer for you – just fire off a Windows XP, and it’s giving me a
letter or email and it'll all be taken problem. One of the Linux
care of. partitions (/) is nearly full, while the
other one (/home1) is almost empty.
I used Partition Magic 8.0 on the
LXF Answers guy
David Coulson is XP partition to resize it before
a networking and installing FC 3, but I’ve used PM on
security guru with another PC to resize a Linux
plenty of sysadmin partition and it messed up the
experience to boot.
whole Linux partition – it didn’t
boot anymore and I had to reinstall
everything from scratch – so I’m
reluctant to use PM again.
Nick Veitch is the
editor of the On the Linux partition of my
magazine, and laptop I have installed QtParted,
DSL modems are a pain to get working but with easy-to-install tools such
answers your easy but this doesn’t allow me to as these for the BT Voyager, much of the headache can be avoided.
questions! Or manipulate a mounted partition. I
indeed anything to
do with GRUB, LILO, netatalk, Vi...
can see the settings, but can’t
change them. How do I safely partitions, or even to copy everything change the partition structure around,
transfer unused space from the to a second disk (or the Windows but it’s hit or miss whether you
/home1 partition to the / partition partition if it is not NTFS-based) and actually adjust the partition correctly
Hans Huberland without destroying my Linux then go back to the beginning and this way.
is Rackspace installation or losing data? I’ve install Linux again. To resize the filesystem on
Managed Hosting’s
Linux expert and a
spent so much time installing and I would recommend that when /dev/hda1 to 512MB, you would
real-life system configuring everything on this installing Linux for the first time you issue the following command as root:
administrator on machine and I really don’t want to create a separate /home partition in resize2fs /dev/hda1 512M
call for your questions. have to re-do it from scratch. addition to the mandatory / (root) and Unmounting is a good idea prior to
Send your questions for our experts to: Vincent Borreux swap partitions. In this way data is making any changes that modify the
As you are using partitions separated from the system so you can structure of the filesystem. DC

A
Linux Format, Future Publishing,
30 Monmouth Street, Bath BA1 2BW rather than volumes, it’s do an update or a complete reinstall
or email: lxf.answers@futurenet.co.uk. very difficult to resize them while keeping your data safe. Radeon redux

Got a sysadmin query? Send it to Hans
at sysadminqa@rackspace.co.uk.
without wiping everything
out. A smart option is to back
everything up prior to altering the
Q
If QtParted isn’t working, you might I’m having a problem
consider using resize2fs to shrink with the Linspire 5.0 Live
ext2 or ext3 filesystems then distro. I can run it as a
Live CD with no problem,
but when I try to install it by typing
startx at the prompt the cursor
appears and the screen turns grey –
then it just stops. I’m not sure what
to do to get it to work. Any help
would be appreciated. My hardware
specs are:
Pentium 4 3.00 GHz
512MB DDR2 RAM
200GB ATA hard drive
ABIT AA8 DuraMAX motherboard
ATI PCIE Radeon vGuru series
RX600 graphics card.

LXF70.answr 100 11/7/05 2:42:31 pm


Linspire has a voiced
tutorial on how to run
Linux – but not how
to install it!
PJ
It sounds as if
A
Linspire doesn’t like your
video card. It might not
have recognised it correctly,
or even support it.
As you have a high-end and fairly
new Radeon chipset, the chances are
that Linspire doesn’t even know what
it is. Video card drivers for X are
updated frequently but many
distributions are a few revisions behind,
and some are a long while overdue for
an update.
You may want to try another Linux
distribution, such as SUSE (which lists
your card as fully supported in
version 9.2 and 9.3), Fedora or
Mandriva, just to see what they do
with the card. These distros tend to be
updated a little more frequently than
Linspire, plus they have more
community support to get updates in
the mix early.
Linspire may also have updates
available, but you’re kind of stuck if it
won’t even start a graphical desktop.
DC
Now, Voyager
Q
I tried the Knoppix Live CD
from the cover of your
magazine [LXF63], and it
recognised almost all of
my hardware except for my ADSL
modem – for me, the most important
bit of hardware by a mile.
A QUICK REFERENCE TO:
For many years, administrators used
good old Telnet, or even rsh, to access
systems. However, as internet security
has become a concern for more and
more individuals and corporations, it’s a
growing liability to use platforms and
protocols that use plain text
authentication. Both rsh and Telnet send
everything back and forth between the
client and server without any form of
encryption or verification, so attacks that
involve an intermediate system collecting
information and replacing it with
malicious content are a real problem.
SSH is an excellent alternative as it
performs encryption of the
authentication stages of the connection,
as well as for the actual terminal session
itself. SSH can also perform
authentication using keys rather than
passwords to avoid the system password
LXF70.answr 101
Knoppix from LXF63 may
not recognise the Voyager modem.
I have a BT Voyager 100 USB
ADSL modem, which I received when I
first went to broadband with AOL, and
now use for my current ISP,
Central Point. I have no idea where to
go for drivers for this modem or how
to go about installing them.
The current drivers from BT also
launch into the sign-on interface –
would this come with any Linux-
compliant drivers? I would like to
install either Fedora Core or SUSE.
James
The Voyager 100 is a
A popular modem, but it isn’t
supported by many Linux
distributions. However, you
can obtain drivers that will work with
most systems.
I’d suggest you head over to
http://eciadsl.flashtux.org/
download.php?lang=en and scroll
down till you find the EciAdsl Nortek
section. Now download the .bz2 file,
and reboot to Linux. Copy the .bz2 file
to somewhere like /usr/src and start a
shell prompt:
tar -jxvf eciadsl-usermode-0.10-
nortek-alpha.tar.bz2
cd eciadsl-usermode-0.10-nortek-
alpha
./configure
make install
SSH
even being typed into the client. This can
help to secure systems where passwords
may have been compromised, as
password-based authentication can be
completely disabled and everyone forced
to use keys.
You can forward other TCP sessions
over an SSH connection, encrypting all
information as you go – essentially
creating a secured proxy service. For
situations where a mail server has been
unavailable – or all you can do is SSH
out – being able to forward a TCP
connection from the SSH server, or
within the network it lives on, can help
work around local firewall issues or
configuration problems. Anything you
forward over SSH will appear as a TCP
port on a local system, avoiding the need
for any major configuration changes on
the client.
www.linuxformat.co.uk
ANSWERS
ALL ABOUT RACKSPACE
LXF Sysadmin
Answers – in
association
with Rackspace
Managed
Hosting
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
hosting company to guarantee 100% committed to ensuring customer
network uptime and we will even satisfaction, and they will not rest until
pledge to replace faulty hardware (let’s a client problem is resolved.
face it, it does happen) within an hour. Since Rackspace was established in
As you would expect from Red Hat’s 1998, it has specialised in managed
only Advanced Hosting partner you will hosting – nothing else. We’re not an ISP,
receive Fanatical SupportTM and instant we don’t sell domain names, and we
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.
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
All our customer platforms are improve your web hosting, please
housed in state-of-the-art, secure data contact us at www.rackspace.co.uk
centres and can be fully customised to and ask any questions you have about
meet any requirement. A dedicated the ultimate managed hosting service.
account manager is allocated to you at See page105 for Rackspace’s star letter.
make cfg modem, type 10 and enter the
cp GS7470_SynchFiles/gs7470_ chipset as GS7470. When choosing
synch01*.bin /etc/eciadsl synch files, you want gs7470_
Now you have to configure the synch01.bin. There is some other
driver with the settings BT gave you. information in the tarball in case you
The VPI is 0 and the VCI is 38, and run into problems, but this should be >>
the password can be anything you enough to get you up and running. DC
want. When it asks you to choose a
Clients
for SSH are
available for
almost any
operating
system you
can think of,
including
mobile
devices and
Windows.
The standard Unix SSH The sshd service is a reliable and
implementation is known as OpenSSH secure way to permit remote
(www.openssh.org), and ships with access to Unix systems over the
nearly every Linux distribution. For public internet.
Windows users, PuTTY (www.chiark.
greenend.org.uk/~sgtatham/putty)
is particularly popular, and supports
all of the features offered by SSH,
including the TCP port forwarding and
key-based authentication.
LXF70 SEPTEMBER 2005 101
11/7/05 2:42:36 pm
 ANSWERS
FREQUENTLY ASKED QUESTIONS KERNEL
FAQ How do I find out what
kernel I’m running?
These days, most up-to-date
distributions ship with 2.6 kernels,
although some are still using the
2.4 kernel. You can easily check what
version you’re running with the uname
command. Simply do
$ uname –a
and you’ll get a line that should look
something like this:
Linux cr2 2.6.11.6 #1 SMP Mon
Mar 28 18:48:32 BST 2005 i686
GNU/Linux
This system is running 2.6.11.6,
which is the sixth revision of 2.6.11.
Your distro’s creators might have
applied their own patches, so it won’t
be a simple 2.6.x.y kernel version.
FAQ Where do I download a
kernel from, and what
do I have to get?
You can download a tarball of the
latest kernel from the Linux Kernel
Archives at www.kernel.org.
Generally, you’ll just want to grab
linux-2.6.11.6.tar.gz from
linux/pub/linux/kernel, which is the
complete tarball of the kernel source.
If you already have a kernel source
tree available, you can obtain the
patch between that and the next
revision – which saves you having to
download the entire source tree.
>> Reaching Apache
Q
I am a newcomer to
Linux, and am trying out
Mandriva Linux before I
make the switch from
Windows. On my Windows setup I
have installed Apache, PHP and
MySQL – all of which I use to test
my websites offline, before I expose
them to the public. Everything
works fine. In Linux, Apache auto-
loads (I noticed this from a previous
Mandrake install), and I have also
installed PHP and MySQL.
So far, so good. The problem is
that all my data is on my FAT32 /
Windows drives. I don’t want to
copy over the files to a Linux
partition because then I wouldn’t
have access to them from Windows.
So I would like to know if it’s
possible for Linux to access the
102 LXF70 SEPTEMBER 2005
LXF70.answr 102
FAQ Do I have to use the
kernel that came with
my distribution or can I go
and build my own?
It’s probably best to answer this with
an example. Red Hat distributes a
kernel that’s heavily patched, and
some of its tools and utilities don’t
work properly with a kernel that
doesn’t have its patches. Likewise,
some third-party patches will conflict
with the contents of the Red Hat
kernel, as they will be expecting a
clean kernel tree.
Unless you need specific kernel
capabilities that render a Red Hat
kernel useless, due to incompatibility,
it’s often good to stick with distribution
kernels. Of course, if you don’t need
any of the patches within the
distribution’s kernel, a vanilla kernel
tree should work perfectly well.
FAQ I’ve got my kernel tar
file. How do I compile it?
The general build location is
/usr/src/linux, so you need to cd into
/usr/src then untar it with
tar xzf ~/linux-2.6.11.6.tar.gz
Once the kernel tarball is
unpacked you will have a linux
directory, which contains the source
code for compiling the kernel.
The general routine to configure,
build and install a kernel is
Windows partitions for Apache, and
what and where to write in the
Linux Apache config file httpd.conf.
I noticed the httpd.conf setup
file for Apache is considerably
shorter in Linux than it is for the
Windows machine, so that’s doubly
confusing me about how to edit the
file. Editing Linux line-ending files
when in Windows is no problem, as
I have an editor that will read and
write both Windows and Linux
format text files and line endings.
From the LXF forums
You can edit the config file
A httpd.conf and adjust the
DocumentRoot option to
point to your web directory
mounted from your FAT32 filesystems.
You can check with mount which
filesystem is mounted where on the
Linux file structure and point to the
appropriate directory. Bear in mind
www.linuxformat.co.uk
cd linux
make mrproper xconfig dep clean
bzlilo modules modules_install
First, you need to configure your
kernel. The defaults will give you
something that will boot but you’ll
probably need to select other
options for hardware devices, such
as soundcards and SCSI interfaces.
As long as there is a backup kernel
available via LILO, it shouldn’t be a
major issue to try different options to
decide which supply the required
capabilities for the system.
Running bzlilo will build the kernel
and copy it to /vmlinuz. Run lilo to
install the new kernel. You’ll have to
modify LILO to use /vmlinuz as the
image, rather than /boot/vmlinuz,
although you can hack the Makefiles
to install the kernel into /boot/.
The last two make options in the
above code build the loadable
modules and put them in the right
place for the kernel to use. Once you
have a kernel built with loadable
module support, you can just go in
and reconfigure the kernel and install
the modules.
FAQ Can I add capabilities to
the kernel source tree?
There are many patches available for
the Linux kernel, which enable it to do
things that a standard kernel can’t.
Configuring Apache can be complex for beginners, but a well-commented
httpd.conf will get you off to a good start.
that Apache doesn’t much like files
with spaces in them, and is case-
sensitive under Linux, so your existing
websites may not work exactly as they
do in Windows.
One of the main projects that
distributes combined kernel patches
is WOLK, the Working Overloaded
Linux Kernel. Rather than having to
apply individual patches one at a
time, WOLK provides a single patch
that adds a whole bunch of different
patches. It is available from
http://sf.net/projects/wolk.
There is also the -ac kernel series,
distributed by Alan Cox. These
kernels contain many updates that
haven’t yet made it into the
mainstream kernel, along with some
basic fixes for things broken in
existing kernels. Many people use
-ac kernels instead of the regular
kernel tree; although, as with
distribution kernels, other kernels
may object to being applied to an
-ac tree.
You can find httpd.conf in
/etc/apache, which is generally well
commented under Linux so you can
easily adjust configuration options
without worrying what they will do. DC
11/7/05 2:42:38 pm

Ping pong
Q
I have three machines
connected on my
network, and problems
working between them.
They are:
LinBox running Mandrake 10.0
WinBox running Windows 2000
WebBox running Mandrake 9.1 as
an internal web server.
The network seems to be
working only partially. The
connection seems fine between the
Linux and Windows boxes – I can
ping between them and WebBox
can ping to both – but I can’t ping
to WebBox from either of them.
However, I can open the default
web page on WebBox (running
Apache) at http://192.168.0.3:80, and
using LinNeighborhood I can see the
Linux and Windows boxes on all
three machines, but not the web
box (they are all in the same
workgroup INTEGRANE). Can you
suggest anything?
Mike
It sounds to me like your
A web server box is not
participating in the Windows
network properly, even
though its IP configuration is sound. It
may be that there is a firewall running
on the system, and it’s blocking the
SMB traffic from the network.
To verify whether a firewall is
running, run iptables –nvL.
If you can ping the system by IP
address, in this case 192.168.0.3, I
think it is unlikely to be a routing issue.
The fact that you can access the web
server directly pretty much discounts
any IP problems.
What I’d suggest is that you
compare the smb.conf files from
WinSCP is a great tool allowing Windows systems to copy files securely to
Linux servers over a network.
LXF70.answr 103
LinBox and WebBox to establish if
there is any configuration variation
within Samba on the two systems, and
to ensure that Samba is running
correctly on your server. DC
Which CMS?
Q
I work for a company
with about 75 people in
different departments,
some of them technical,
some of them not. I would like to
have a content management system
to hold all the data that could be
shared within the organisation,
ideally with the minimum
technical input possible.
Maybe text files or
Word documents. Can
you recommend one?
Lindsay
As a matter of
A fact, I have looked
into this issue recently
and I’d like to share some of
my findings. Mambo (get it from
www.mamboserver.com) takes the
cake: it’s the most popular, most
versatile and most configurable CMS.
The only snag is that it has by far the
steepest learning curve around, and
from what I know, customisation
involves delving into the code.
I won’t comment on the usual
suspects – PHP-Nuke, Postnuke and
the like – because they’ve been amply
discussed elsewhere. I would like to
mention TWiki here (http://twiki.org)
because it is a nice little system,
though before customisation it looks
overwhelming to the non-techie.
But the CMS I liked the most was
Exponent (www.exponentcms.org).
It’s the most configurable, the most
straightforward and the most docile to
the untrained eye. Just set up an
www.linuxformat.co.uk
underprivileged user in it, log in as
that user and you’ll see that in Edit
mode it is nowhere as scary as the
others – it’s quite inviting and intuitive.
One last thing – I’d invite you to
look at www.opensourcecms.com,
which has live websites of web systems
that are reviewed and rated to help IT
managers make up their own mind on
the subject. HH
File transer
Q Q
I have a networked web
server set up with Apache,
which I am going to use
for testing websites. I can
access the web pages via port 80
over the network, but at the moment
only the default index.html file is
available in /var/www/html.
What is the best way to upload
files? Can I set up FTP, and if so, how
do I do it?
Mike Davies
Actually there are a number
A of ways to get remote file
access with Linux, the most
popular of which is the SSH
protocol. SSH offers secure transport
for shell or file transfers, although
there is some overhead due to the
decryption and uncompression that
takes place (more of which in Snail-
paced Backups). Of course, on a
public network security is a major
concern, so the safety of SSH offsets
many of its throughput problems. SSH
works with Windows using a client
such as WinSCP for file transfers and
PuTTY for shell access.
If you are going from a Linux
workstation to a Linux server on a
ANSWERS
Tools such as iptraf provide valuable
information for anyone checking
performance issues on a network.
local network you can use NFS to
share a filesystem on the server and
access it on the workstation as if it
were local. This is usually the best
configuration, as NFS runs at near to
line rate. However, it is insecure on
anything close to a public network.
For a Windows client, you can
always install Samba on the Linux
server and access the web directories
using a share. Copying data back and
forth is simple, although Samba might
take a little coaxing to work exactly as
you want it to. DC
Snail-paced backups
I’m trying to set up
Unison so that it will
synchronise backups on
a small LAN. It all works
fine, except that the transfer rate is
very slow compared with the LAN’s
normal speed.
The PCs are both using Unison
on SUSE 9.2, connected by SSH. I’m
getting about 80KB/sec on a
100MB/sec LAN that normally
manages about 10MB/sec over NFS.
Is the lack of speed due to SSH
or to Unison? Is there any way to
tweak whichever is causing the
slowdown to improve matters?
Neil Lewis
Depending upon the speed
A of the systems transfering
data, the encryption
overhead can cause the
throughput rate to drop significantly,
but for modern workstations and
servers, it probably won’t be noticable.
However, when you’re accessing the
drive and possibly zipping up the
contents, things can run slowly and
start to drag. 80KB/sec is pretty
dismal – we doubt SSH has anything >>
to do with this. You could try to run
LXF70 SEPTEMBER 2005 103
11/7/05 2:42:44 pm
 ANSWERS
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.
>> Unison over rsh or rsync and see if
you’re able to squeeze any more life
out of it. Trying to copy some data
using SSH directly will help to indicate
if SSH is really any burden at all. DC
Share where?
Q
I’ve got a Samba web
server on Fedora Core 3,
which is fine for
browsing and generally
works OK, so I decided to try a
Debian server as well. I set up a
very simple Debian Stable system
and Samba, and the storage share is
working fine, but I can’t browse to
it. I can open the share by IP
address, so it seems to be a
NetBIOS problem. I don’t
understand this, because I’ve been
going over and over the smb.conf
and comparing the two. The only
difference is that Debian is using
domain authentication and I used
SWAT on it instead of hand editing
as I did with Fedora.
104 LXF70 SEPTEMBER 2005
LXF70.answr 104
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
Also on Fedora, I’ve noticed that
I can’t ping any Windows systems
by hostname, although they can
ping me by hostname. Both systems
have nmbd and smbd running and
both these machines have their
hostnames set as well as other
general networking settings (IP, gw,
netmask, broadcast, etc).
From the LXF forums
You can always add a Linux
A system to the hosts file on
Windows, which can be
found using the Search
option. Here you can set specific
hostnames for IP addresses on your
network, which will work in the event
of it becoming unreachable via
NetBIOS.
You can also add the following
option to the global section of your
smb.conf. Of course, this needs to be
unique for each of the Samba servers.
netbios name = SambaServer
DC
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.
No limits
Q
I would like to use HTTP
to upload large files (in
excess of 200MB) to a
server, because some of
our customers aren’t able to use
FTP due to firewall restrictions
(using Perl or PHP as the upload
handler). What issues are there
with Apache handling this amount
of traffic in an occasional burst, or
even frequently? Should there be
any tuning done to the server that
would help? Or are there any
amends I could make to our Apache
configuration file that would help?
John Moylan
Indeed, there are a number
A of configurable limits to how
big the file size is. The ones
that might affect you are
Apache’s LimitRequestBody, though
it defaults to unlimited in version 2.0,
and PHP’s upload_max_filesize and
post_max_size. However, what
generally affects Apache the most in
file uploads is the memory
consumption and how long the httpd
process stays up. The latter limits the
number of requests httpd can handle,
either because of the number of
processes that have to be in memory
or the memory footprint of the
processes themselves.
Also make sure you’re not allowing
uploads via SSL, to avoid the
overhead that encryption would incur.
If you’re using Perl, you should fork so
that you release the httpd process
and have your child process handle
the rest of the transaction. I couldn’t
tell you anything about that in PHP as
I haven’t used it for quite a while. HH
Look don’t touch
Q
I have discovered the
wonders of Knoppix
while waiting for my new
PC to arrive. The HD
from my Linux PC is on /dev/hdb
and all my data is there. I wanted to
copy a few files from /mnt/hdb7 to
/mnt/hda5 but I get a read only
message. I have done su and
checked the /etc/fstab and listed
the permissions – it all looks fine.
I have never worked with
Knoppix before; I just booted it up
as a novelty. Now that it is a truly
useful tool I need to know how to
make the most of it. I am sure that I
am missing something obvious.
From the LXF forums
The general philosophy of
A Knoppix is to allow users as
little write access as
possible. For this reason,
existing partitions are either not
mounted or only mounted as ‘read
>>
11/7/05 2:42:47 pm

★ STAR QUESTION WINNER!
This issue's lucky winner is Goran – your new Neuros digital audiocomputer will be with you shortly!
Discover crontabs
Q
Back up, back up, back
up is the first rule of
the trade. Besides my
business, I run a small
home web server and tend to
back up my files on it as often as
possible. I use rsync on my
backup machine:
rsync -avz myserver:var/www /
backup.
What I would like to do is
schedule the process so that it
runs every four hours or so. Any
advice is welcome – my data
needs you!
From the LXF forums
Well, we know that
A scheduling tasks in Linux
is possible, because, as
most of us have found to
our cost, the /tmp directory
automagically deletes files with
tmpwatch at defined intervals –
often before their usefulness has
expired! So let’s explore how that is
implemented. First we’ll find the
tmpwatch configuration file:
[root@carve ~]# rpm -ql tmpwatch
/etc/cron.daily/tmpwatch
/usr/sbin/tmpwatch
/usr/share/man/man8/
tmpwatch.8.gz
It looks like the only file in /etc is
there under cron.daily. Let’s see
what’s in there:
[root@carve ~]# cat /etc/cron.daily/
tmpwatch
/usr/sbin/tmpwatch -x /tmp/.X11-
unix -x /tmp/.XIM-unix -x /tmp/
font-unix -x /tmp/.ICE-unix -x /
tmp/.Test-unix 240 /tmp
/usr/sbin/tmpwatch 720 /var/tmp
for d in /var/{cache/man,catman}/
{cat?,X11R6/cat?,local/cat?}; do
if [ -d “$d” ]; then
/usr/sbin/tmpwatch -f 720 $d
fi
done
If you checked the man pages for
tmpwatch, you’d find no switches
for setting a time or an interval for
running it, so obviously something
else is driving this program every so
often. Let’s query the rpm database
for the owner of the parent directory:
[root@carve ~]# rpm -qf /etc/cron.
daily/
crontabs-1.10-7
LXF70.answr 105
A quick search on the web tells us
that crontabs are actually a bunch of
files that allow us to run commands
at an hourly, daily or monthly interval.
Ah, we’re getting somewhere. On
closer inspection of the files provided
by the same package, we get:
[root@carve ~]# rpm -ql crontabs
/etc/cron.daily
/etc/cron.hourly
/etc/cron.monthly
/etc/cron.weekly
/etc/crontab
/usr/bin/run-parts
Looking at the code, /usr/bin/
run-parts is just a script that makes
those intervals work with crond just
by adding a script in the first four
directories in the listing above. run-
parts does not in any way make
scheduling possible, which leaves us
with /etc/crontab.
[root@carve ~]# cat /etc/crontab
SHELL=/bin/bash
PATH=/sbin:/bin:/usr/sbin:/usr/bin
MAILTO=root
HOME=/
# run-parts
01 * * * * root run-parts /etc/cron.
hourly
02 4 * * * root run-parts /etc/cron.
daily
22 4 * * 0 root run-parts /etc/cron.
weekly
42 4 1 * * root run-parts /etc/cron.
monthly
The listing now makes it obvious
that our answer is in those
encrypted-looking lines just after the
variable assignments. Let’s inspect
the first line: 01 * * * * root run-
parts /etc/cron.hourly. A quick
rummage in the manual pages
provides us with an explanation. The
first entry on the line is the hour at
which a command should be run,
the second entry is the minute, the
third is the day of the month, the
fourth is the month and the fifth is
the day of the week. Any interval can
be defined providing we use the *
character to tell cron (the running
daemon that’s configured via /etc/
crontab) that it should run the
command we want at every iteration
of the field type – in other words,
every second or every day.
The sixth entry (root) is the user
under which the command should
www.linuxformat.co.uk
ANSWERS
be run. The seventh and last field is accepts not only defined values,
the command, which in the example but also ranges and lists of values.
above is run-parts /etc/cron.hourly. Ranges are defined as <start>–
Let’s go through the rest of the <end>: lists are defined as a group
crontab entries to further of comma-separated values, such
understand the format of /etc/ as ‘value1, value2, value3...’. So
crontab. In the line 02 4 * * * root the command you want to run can
run-parts /etc/cron.daily, it tells be implemented with the following
crond that it should run run-parts / lines of code:
etc/cron.daily every 02 minutes * 0,4,8,12,16,20 * * * root rsync -avz
when the hour is 4, no matter what myserver:var/www /backup
day of the month, month or day of Alternatively, you may use a step-
the week it is, as the user root. So type hour definition:
run-parts /etc/cron.daily runs daily * 0-20/4 * * * root rsync -avz
at 4:02. myserver:var/www /backup
Moving on to the next line, 22 4 * This is an instruction to go
* 0 root run-parts /etc/cron. through the integers from 0 to 20,
weekly, we can tell already that run- incrementing the counter by four
parts /etc/cron.weekly will run as
the user root at 4:22 no matter what
day of the month or month it is, at
day 0 of the week. Or, translated,
run-parts /etc/cron.weekly will
each time, which would equate to
exactly the hour definition above.
Alternatively, you could use an
‘under-privileged’ or non-root
crontab, which can be done with
run at 4:22 on Sunday. the userspace tool crontab. HH
A description of the values you
can use in each field is available in
section five of the crontab manual
pages. You should read it, as some
fields start with 0, others with 1.
Now, let’s set up a
cron job for you. You
said you want the job to
be run every four
hours no matter
what day or
month it is.
Fortunately,
crontab’s
hour field
LXF70 SEPTEMBER 2005 105
11/7/05 2:42:50 pm
 ANSWERS
>> only’. If you click with your
right mouse button on an
icon, the ‘read only’
attribute under item
‘device’ can be un-checked.
After this, the partition can
be mounted ‘read write’
(for already mounted
partitions, first click on
Unmount).
CAUTION: writing to
NTFS partitions can lead to
data loss, since Linux does
not really support this
filesystem! However, DOS
and FAT32 filesystems are
safe for write access. In the
shell, the command mount
PHOTO COURTESY LACIE
-o remount,rw /mnt/
<partition name> can
allow already-mounted file
systems to be made
writeable. DC
Writer’s block
Q
I am using a USB2 160GB
Lacie external hard drive,
formatted as a primary
FAT32 partition.
Whenever the disk is connected, an
unmounted icon is placed on the
desktop, which mounts just fine
when I click on it. Unfortunately, I
cannot write to it. I checked the
relevant permissions and as far as I
can see there’s no reason why I
shouldn’t be able to write to it –
but it still denies me write access.
Ideally I would like to just put the
entry for this disk into my /etc/fstab
file. I tried to do that with
/dev/sdb1 /mnt/LACIE vfat 0 0
I have even tried it with
fmask=775 and mask=775 but it
still won’t let me write. Also when I
have mounted sdb1 to /mnt/LACIE
the automated USB icon gets
confused because it wants to mount
Iptables can be used to configure either IP packet filtering on a router or
a bridge to secure a LAN or public network.
106 LXF70 SEPTEMBER 2005
LXF70.answr 106
the device sdb1 to /media/sdb1.
I am using Kubuntu with KDE 3.4
Mike Smith
You can try passing the
A User option to fstab, which
will allow the disk to be
mounted by a non-root user
so that the disk can be written to and
read by the user who mounted it.
Alternatively, you can pass uid= and
gid= statements to set the disk’s
default UID/GID, although with certain
filesystem types, such as ext3 or VFAT,
this won’t make any difference. DC
Severed connection
Q
I’m configuring a firewall
that’s got an eth0 link to
the internet and an eth1
link to an internal subnet
(172.16.2.0). I’ve put in the
following rule to stop all SSH access
www.linuxformat.co.uk
to a PC (172.16.2.120) on
the subnet via the
firewall, thus:
iptables -A FORWARD -p
tcp -s 0/0 -d 172.16.2.120
--dport 22 -j DROP
However, this rule is
still allowing other PCs on
the subnet to connect to
the PC. I’ve also tried the
following rules, and even
gone to the point of
specifying an individual
source PC on the subnet,
dropping all SSH traffic to
the destination PC and
changing the FORWARD policy
to DROP.
iptables -A FORWARD -p tcp -s
172.16.2.0/24 -d 172.16.2.120 --
dport 22 -j DROP
iptables -A FORWARD -p tcp -
s172.16.2.220 -d 172.16.2.120 --
dport 22 -j DROP
iptables -A FORWARD -p tcp -d
172.16.2.120 --dport 22
-j DROP
iptables -P FORWARD DROP
Yet I can still contact the
destination PC from another PC on
the subnet. I’ve read and read and
read till I’m blue in the face, and
can’t for the life of me figure out
why this isn’t working.
Missy
As you are SSH-ing
A between two systems on a
local network, you won’t
route across your firewall for
this access. Thus, the packets will
never be inspected by the firewall. If
you want to block SSH access, you will
have to set a firewall up on the SSH
server to block traffic itself.
Another option if you have a spare
NIC is to split the network into two
sections and bridge the two using the
bridge-utils package in Linux. You will
then be able to perform packet
filtering on the firewall for traffic that
goes between the two LAN segments,
even though the packets are not
actually routed. Lots of information on
this configuration can be found at
http://bridge.sf.net. DC
Again and again and…
Q
I have a two computer
network at home, one
running Windows XP and
the other Mandrake 10.1.
The Mandrake box acts as the server
for the windows machine and shares
the internet connection.
My trouble is that every time the
internet is restarted I have to enter
the following three commands as su:
iptables -t nat -A POSTROUTING -o
ppp0 -j MASQUERADE
echo 1 > /proc/sys/net/ipv4/ip_
forward
route add default ppp0
What files do I need to edit to
put those commands into so I don’t
have to manually do this each time?
Once you add the iptables
A rule, you can do /etc/init.d/
iptables save, which will
save your iptables
configuration for the next reboot.
IP Forwarding is enabled through /etc/
sysctl.conf, where you can add a line
‘net.ipv4.ip_forward = 1’ to set up IP
Forwarding next time you reboot.
For the default route through ppp0,
you should be able to reconfigure your
dialer to automatically add a default
route when the modem connects. It’s
difficult to stop it setting a default route,
so checking the logs to establish why
one is not being added may be a good
place to begin. DC 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.
11/7/05 2:42:55 pm
 LXF BOUNTY

Solve this puzzle and

News of a new challenge for you, and there’s more to it than meets the eye.

4 3
S 1 6
udoku is the craze
that’s sweeping the

7
world – as we write

6 1
there are batteries
of statisticians
being employed to

8
work out just how many

7 5
man-hours are being lost
compiling statistics about people
playing Sudoku when they
should be working.
Usually Linux Format is the
first to turn and run at the sight
of a new trend, but this one’s got

8 2 4
5 3 6
numbers in it and looks slightly
geeky, so we’re hooked. Welcome,

1 9 7
then, to a special LXF Hyper
Mega Sudoku!

8 9
On the right is a diverting
puzzle for you but, this being
Linux Format, we’re not going to
just give you cash for writing a few
numbers in some squares. That’s a 3
mug’s game – why go to all that
effort when you could write a
complex algorithm that could not
only solve problems, but generate 9 7 5
9 8
them too?

2
That’s right: we don’t want you to
solve the puzzle, we want you to write

7 9
some software to solve the puzzle.

The challenge
There are already a small number of
Sudoku solvers available, but quite
frankly we aren’t greatly impressed.
Some are better than others, but we
demand that open source does better
and believe that LXF readers can
deliver a classier solution.
What do we want? We want you to
write a nice, flashy, friendly GUI-driven
app that can do the following:
■ Display Sudoku puzzles of varying
formats – 3x3, 4x4 and 5x5 minimum
(see What On Earth Is Sudoku? box,
above right).
1 3
■ Create a standard file format,
preferably XML- or text-based, for
MORE TO COME
storing and exchanging puzzles. This is the first of our programming maybe, just maybe, something that
■ Solve puzzles. challenges. We want to help make Linux should be a lot better documented?
■ Generate new puzzles. better, and we need you to help us do We may need to have different rules
■ Make sure solutions are unique. that! Is there some feature that your and schemes for different projects, but
favourite text editor should have but it’s all open to debate. Why not join the
Your solution must also be
doesn’t? Some file format that should discussion on our website at
released under an open source
be supported by your audio player, or www.linuxformat.co.uk/bounty?
licence, preferably under the GPL, but

112 LXF70 SEPTEMBER 2005 www.linuxformat.co.uk

LXF70.sudoku 112 11/7/05 1:46:15 pm

 LXF BOUNTY

win a share of £500!

WHAT ON EARTH IS SUDOKU?
Originating from the US, a Sudoku is just
a number puzzle made up of a standard
9x9 standard grid, which is made up in
turn of nine smaller grids of 3x3. US
readers might know it as Number Place.
The rules are simple – you must add a
number from 1 to 9 to each empty
square until the grid is full. No number
can be repeated in any one line, any
column or any small 3x3 grid.
Given a handful of numbers to start
with, it should be possible to complete
the grid, and if the puzzle has been
compiled correctly, there will be only
one solution.
For larger puzzles, the range of
numbers and dimensions of the grid
need to increase. For example, a
16x16 grid might need to use 0-9 and
then A-F to give you 16 characters.
There is plenty more information
on Suduko available. Just pick up any
daily newspaper in the UK, for a
start. Alternatively try visiting a few
Sudoku resources, such as
http://en.wikipedia.org/wiki/sudoku
and www.sudoku.org.uk.
You could also ask for clues and
advice at the LXF bounty forum,
www.linuxformat.co.uk/bounty.

HOW TO ENTER
Send your program to lxf.competitions@futurenet.co.uk by September 30
2005. Entrants will be assumed to have read and accepted the following
rules and conditions.

THE RULES
1. The challenge is open to anyone of any age and nationality, other than employees of
Future Publishing Ltd and their families.
2. You may submit as many solutions as you like.
3. Proof of email sending is not proof of delivery. Incomplete entries may be treated as invalid.
4. Winning entries will be chosen by judges after September 30 2005. A list of winners will be
published on www.linuxformat.co.uk/bounty. The list may also be obtained by written request
from: The Editor, Linux Format, 30 Monmouth Street, Bath BA1 2BW.
5. The judges’ decision is final and no correspondence may be entered into.
6. No purchase is necessary.
7. Entrants must be able to claim the work as their own.
8. LXF intends to publish the winning entries in the magazine, on the coverdisc and online.

any licence that complies with the
Open Source Initiative will do. We’ll
award bonus points for solutions that:
■ Look good.
■ Are cross-platform.
■ Solve puzzles quickly.
■ Solve puzzles elegantly (not just
brute force).
■ Can grade puzzles (such as easy,
medium and hard).
■ Are easy to install and build.
The prize
The admiration you’ll win from the
Linux community for completing the
Sudoku challenge is of course
priceless, but if you need a little more
incentive there is also a prize pool of
£500, to be divided into £300 for the
winning entry, £150 for second place
and £50 for third.
This is the first of our programming
challenges: we hope to introduce
more LXF bounties in the months to
come. The deadline for entries to this
challenge is September 30 – all code
must be submitted to us by then.
Good luck!
Oh, and by the way, Nick
completed the puzzle on this page in
around ten minutes. If you can beat
this, let the forum know! LXF

www.linuxformat.co.uk LXF70 SEPTEMBER 2005 113

LXF70.sudoku 113 11/7/05 1:46:23 pm