ISSN 1746-8752

10

9 771746 875009
 Free Software Magazine is a magazine
by The Open Company Partners Inc,
EDITORIAL 90 Main St. Road Town, Tortola BVI

We can all finally install E DITOR IN C HIEF

Tony Mobily (t.mobily@)

verything seems to have long-term problems; my back, for example, has been hurting for quite a S ENIOR E DITOR

E while now and no real diagnosis or prognosis has showed up yet; my car makes a mysterious noise:
a “clunk” which is hard to describe and occurs randomly but consistently (except, of course, when
my mechanic is driving it). Linux has its problems too, especially when it’s being used by less
experienced users. (Yes, I know I am in dangerous waters here!)
People want to use Linux, and being able to install what they need is important. This is even more true for new
Dave Guard (d.guard@)

S ALES AND A DVERTISING

Bridget Kulakauskas (b.kulakauskas@)

users, who often want GNU/Linux to “match” what their old (Windows?) system used to do for them.
L AYOUT AND C OMPOSITION
I’ve seen a lot of new users—and even kids—using Linux comfortably. And everything goes fine—until they
Gianluca Pignalberi (g.pignalberi@)
decide to install new applications.
You see, in Mac people can install an application by simply downloading it, copying it wherever they like, and
double-clicking on it. In Windows, it’s a matter of running an ugly installer, answering a few questions, and G RAPHIC DESIGN
letting it copy a zillion files all over the place. Alan Sprecacenere (a.sprecacenere@)
In Linux. . . it depends.
Even though distributions such as Ubuntu do a terrific job in giving users an amazingly complete “base” system,
and a reasonably intuitive way of installing new software (thanks to the Synaptic Package Manager), users always
C ONTRIBUTORS
get lost when they try installing an “unplanned” piece of software (that is, one that was not pre-packaged and
Martin “MC” Brown, Tom Chance,
pre-compiled by the distribution’s maintainers).
Terry Hancock, Daniel James, John
Unfortunately, no matter how well distributions try to add every single package to the list of “supported” appli-
Locke, Edward Macnaghten, Tony Mo-
cations, such a list can never be 100% complete. The problem is even more relevant if somebody wants to install
bily, Arturo Fernández Montoro, David
a piece of non-free software, which will obviously never be supported by a distribution! Users also need to be
Sugar, Jeremy Turner, Maria Winslow.
“root” to install software, and that’s not always ideal. . .
I feel it’s only fair that I show my cards, and admit right away that I am a fan of “The Apple way of doing things”:
each application has its own directory, which contains everything needed by the program, and is seen by the user
as a fancy icon (chosen by the application’s developer). Uninstalling a program is as simple as dragging it into
the rubbish bin!
It sounds simple, and it is—for the user. Apple has been doing this for many years, and it definitely works.
However, things that makes life easier for the user often add much more complexity to the system. In the case
of Appdirs, there are several issues, and not all of them are purely technical. Appdirs need to be a joint effort:
distributions and desktop environments (see: Gnome and KDE primarily) need to agree what an “Appdir” is, how
it works, and so on; Appdirs go against the Unix philosophy of putting each file in the right place—this is possibly
why there is resistance to Appdirs in the Linux world; with Appdirs, the “automatic upgrade” process becomes
tricky to say the least (however, it’s definitely not impossible); the desktop environment must be able to “register”
each Appdir (probably at its first execution), and must be able to associate a file type with a particular Appdir;
finally, existing applications (and there are a lot of them!) all need to be repackaged.
Some of these problems might never be solved completely; this is possibly why Linux has taken some steps
toward Appdirs. However, no distribution or “desktop environment” has endorsed them fully. These problems
have never had an easy solution.
Until now.
Thanks to Klik (http://klik.atekon.de/), developed by Simon Peter, this situation has finally changed.
Klik deals with most of the issues I mentioned above (if you are curious, check the FAQs (http://klik.
T HIS PROJECT EXISTS THANKS TO
atekon.de/wiki/index.php/User\OT1\textquoterights FAQ)); very importantly, it works well
Donald E. Knuth, Leslie Lamport,
under Ubuntu, which happens to be the fastest growing distribution at the moment. Thanks to Klik, you can
People at TEX Users Group TUG
download your favourite applications, burn them onto a CD, and give them to your favourite and least experienced
(http://www.tug.org)
Linux user—even your grandmother!
Klik is another huge step towards Linux’s desktop domination. Right now, thanks to Klik, there’s one less excuse
for not using Linux. Are there any left now?
Every listed person is con-
Congratulations and thank you, Simon!
tactable by email. Please just add
Copyright information c 2006 Tony Mobily
freesoftwaremagazine.com to the person’s
Verbatim copying and distribution of this entire article is permitted in any medium without royalty provided this notice is preserved.
Tony Mobily is the Editor In Chief of Free Software Magazine username in parentheses.

For copyright information about the contents

of Free Software Magazine, please see the
section “Copyright information” at the end
of each article.
Free Software Magazine Issue 10, January/February 2006
Contents Issue 10, January/February 2006

EDITORIAL MIND SET

We can all finally install 3 Convincing management to approve
free software 39
by Maria Winslow
POWER UP Tips to better advocacy
What’s free about free software? 7
by John Locke
Computing and the American West
A techno-revolutionary trip on the
internet 43
by Tom Chance
USER SPACE
Reflections on the lessons from Dean for America
What is X? 10
by Edward Macnaghten
Discover the versatility and power of the X Window System

64 Studio 48
by Daniel James
Browsers for Mac OS X 22 Building a native 64-bit creative distribution
by Martin “MC” Brown
Comparing FOSS browsers for Mac OS X

Towards a free matter economy (Part

HACKER’S CODE 4) 52
GRUB tips and tricks 28 by Terry Hancock
by Jeremy Turner Tools of the trade
Spicing up a great utility for more IT fun

Jump to Debian GNU/Linux! 33 Free software liberates Venezuela 61

by Arturo Fernández Montoro by David Sugar
A guide to why the Debian distro is a good choice The free software revolution comes to Venezuela

4 Free Software Magazine Issue 10, January/February 2006

What’s free about free
software?

Computing and the American West

John Locke

omputer history has some interesting parallels ers in the US, none have had a monopoly on sales—we all

C with the history of the American West. After

the initial forays of Lewis and Clark and the
first set of explorers, early settlers crossed the
plains in covered wagons. But the West wasn’t accessible
to most Americans until the age of the railroads, when the
like different things in our cars, and make different choices
based on our likes and needs. And we all need to go through
some sort of drivers training before we’re safe on the roads,
yet hundreds of millions of us take this for granted.

Union Pacific Railroad put tracks across the continent and

started running a regular passenger service. Railroaded today

Microsoft is the Union Pacific of the computer world. Win-

Railroad history dows provides the tracks. In the computer world, we’re in
the very early days of the automobile, say in the 1920s—the
To ride the trains, you needed to pay a fare to the railroad railroads are still dominant, and go nearly everywhere you
companies. These companies became huge monopolies, be- might want to go. But you have to go on their schedule, and
cause they controlled the only way for the average person to
cross the country. For a time the railroads were it, and as
long as the tracks went where you wanted to go, the rails Driving the Golden Spike on the transcontinental railroad
were the cheapest and best way to get there. And after all,
the major cities all developed along rail routes, so where
would you want to go that you couldn’t get to by rail?
Of course, with the advent of the automobile, that all
changed, and today, while trains still exist and still go to
a lot of the same places, they are a shadow of their former
dominance of the transportation landscape, used by com-
muters in some cities, and by tourists. Most of us drive,
because we have the freedom to go wherever we want, on
our schedule. It still costs money to get there—we have to
pay for the upkeep of our cars and fill them with gas. We’ve
all paid for our roads through taxes, tolls, and other means.
While there have been three big, dominant auto manufactur-

Free Software Magazine Issue 10, January/February 2006 7

 POWER UP
you have to pay a fare to get on board. While there are dirt
roads all over the place, few are paved, the trains will get
you to your destination faster, and you pretty much have to
be a mechanic to keep that Model T running over any large
distance. But the car is clearly the future, and we’re starting
to build the freeways now.
---------------------------------------
In the computer world, we’re in the very
early days of the automobile, say in the
1920s. . .
---------------------------------------
Free software powers those cars. While there are still a cou-
ple of cities you can’t drive to yet, you can pretty much
go anywhere, including places you now can’t get to by
train. Free software will get you nearly everywhere you can
get with Windows. There are great free word processors,
spreadsheet packages, presentation programs, desktop pub-
lishing programs, astronomy tools, databases, everything
that you might imaginably need to get where you need to
go. But these programs aren’t from Microsoft, or Adobe, or
the other “standard” rail car manufacturers, which don’t run
on your average dirt road.
Once upon a time, only 15 years ago, there were several
different word processing programs considered “standard”.
Anybody remember Word Perfect? It had a much bigger
market share than Microsoft Word. Yet today, everybody
expects Word documents and little else. At that point, hav-
ing a different word processor was like having a rail car that
fit a different set of rails—you couldn’t just open a Word-
Perfect document in Word and expect it to look the same.
Freedom to drive
In the free software world, where programs need to commu-
nicate with each other, they use established, open standards.
A new document format has emerged, called the Open Doc-
ument format. Unlike Word Documents, you already have a
choice of several different programs that will read and write
these documents without issues: OpenOffice.org, StarOf-
fice, KWord, AbiWord to name a few. The state of Mas-
sachussetts recently mandated that all their office docu-
ments be stored in this format, to prevent being locked into a
single vendor. Naturally Microsoft is complaining—if they
can’t maintain their advantage of being the only vendor ca-
8 Free Software Magazine Issue 10, January/February 2006
pable of flawlessly working with their own office format,
how can they maintain their monopoly? They would in-
evitably get a whole slew of competitors—bad for them, but
good for the rest of us. Do you really want to be forced into
paying Microsoft time and time again to keep buying ac-
cess to your own documents? The state of Massachussetts
decided, at least for them, the answer was no.
You can free yourself from vendor lock-in, too, by go-
ing to OpenOffice.org (http://openoffice.org) and
downloading the full office suite, entirely free. Version 2.0
is out, and it has a database to rival Access—the previous
versions already provided excellent replacements for the
other Office products. OpenOffice.org Writer is better than
Word in several areas:
• Bullets and numbering
• Outline numbering
• Page templates
• Predictable page formatting
• Master and sub documents, for those book authors out
there
• Drawings with connectors that stick to objects as you
move them around, like Visio
Naturally, it does have some drawbacks, too: the Outline
view isn’t as nice, incompatible with any macros you’ve
developed (and Word Macro viruses), envelope printing is
confusing, merging to a catalog list doesn’t seem to exist.
Are these features worth the extra $300 per seat in your
business?
Land grants and land grabs
In the 19th century, the land grant railroad companies were
granted rights to pick and choose a swath of land across the
country to lay down their rails. Did they choose the best
route for their rails? Not necessarily. They picked the most
fertile land, the most valuable land, and planned their rail
route so that they could get the most dollars for the land
granted to them by the government. A great-great-great-
great grandfather of mine was on a crew that helped move
the rails to a more sensible route, after the land grants were
complete.
Today, it’s Microsoft doing the land grabs, trying to bundle
as much functionality into their operating system as pos-
sible. First it was the web browser. Lately it’s the media
player. There have been dozens of other small companies
unable to compete with the software giant, who have gone
out of business—and as a result, we have fewer choices as
a whole. When Microsoft asks, “Where do you want to go
today?” think customer survey: they’re trying to find out
where they should build their rails so they can extract more
fares from you in the future. Never mind the fact that you
can probably drive there today with free software.
Free software is about giving you the freedom to drive wher-
ever you want. It’s not free of cost—you have to buy your
PC (the car), you have to pay a mechanic periodically, you
need to learn how to drive, and there are taxes, toll roads,
and potholes all over the place. But when given the choice
between a railroad and a car, most of us choose to drive,
simply because we have the freedom to go where we want
and when, not just because it’s cheaper. In many cases, the
railroad is cheaper, especially in these days of escalating
fuel prices. But the more people who choose to drive free
software, the more roads will get paved. And if you need a
mechanic, a map, or a custom vehicle, that’s what Freelock
Computing, or dozens of other new service companies can
provide!

Copyright information

c 2005 John Locke

This article is made available under the Attribution-
Share-alike Creative Commons License 2.0 available
from http://creativecommons.org/licenses/by-sa/2.0/
(http://creativecommons.org/licenses/
by-sa/2.0/).

About the author

John Locke is the founder and manager of Freelock

Computing (http://freelock.com), a consulting
company that helps small businesses use technology to
solve real business problems. John is the author of
the book “Open Source Solutions for Small Business
Problems”, and also maintains his own blog (http:
//opensourcesmall.biz).
What is X?
Discover the versatility and power of the X Window System
Edward Macnaghten
To sell something to the public it needs to look good
veryone likes pretty pictures. The newsagent’s
E stand is now crowded with glossy magazines,
roadside advertisements glare out at you as you
drive along the freeway, you see a wondrous
mosaic as you look at all the packaging on supermarket
shelves. Television long ago replaced the radio as standard
home entertainment and the fact that you cannot judge a
book by its cover doesn’t prevent the vast majority of the
human population from doing so. The same applies to com-
puters now. The GUI (Graphical User Interface) or “win-
dows functionality” has become part of the machine that
everyone now takes for granted.
10 Free Software Magazine Issue 10, January/February 2006
Any home or client system software provider, free or oth-
erwise, that wants to be taken seriously has to provide a
graphical user interface (GUI) as their prime method of user
interaction. Microsoft include theirs as part of the main sys-
tem, or kernel, as do many other system providers. Most
GNU/Linux distributions include it in the default install.
However, with GNU/Linux and other POSIX operating sys-
tems, the GUI infrastructure is not part of the kernel but a
separate program with the mystical name of “The X Win-
dow System”. It calls itself “X11 Release 6” and everyone
else simply knows it as “X”. There’s a lot more to this X
than meets the eye, it has attributes and powers that are not
well known and can do things that other windowing systems
cannot. What is this X? What amazing super-GUI powers
does it have? This article attempts to tear off its mask and
reveal all.
What is X?
A more appropriate question to “What is X?” is “What
is X not?”. X is not actually a GUI. The GUIs in
GNU/Linux tend to be GNOME, or KDE or even MOTIF.
Most GNU/Linux distributions are now delivered with a
GUI configured as the default interface, and this can com-
pare with any other computer system’s GUI. And the end
user, who processes words in documents, calculates sheets
in spreads, mails e’s in readers, browses the surf in webs
and the such, need not know the technical ins and outs of
it. However, there is a lot more that can be done with a
 USER SPACE
The concept of a file server tends to be well understood
small amount of know-how and a bit of command manipu-
lation. . .
So then what exactly is X if not a GUI? X is an infrastructure
that a GUI uses to do its stuff. For example, a GUI handles
the buttons, text and combo boxes, windows etc., whereas X
handles the low level drawing of fonts, lines and pictures on
the screen and accepts keyboard and mouse inputs as well
as the inter-program communication of these. It can also
handle network distribution of users and remote sessions.
One of the most confusing aspects of X is the naming of the
parts, in that the terms “server” and “client” are the oppo-
site of what many would expect. An X server is the screen
and keyboard, what a lot of MS Windows people would au-
tomatically think of as the ultimate client. An X client is a
program that opens and uses windows, such as a browser,
email client, word-processor and so on. To go into why this
is so I will compare an X-server to a file server. . .
A file server is a machine on a network where files exist and
other machines, or clients, can connect to it to open, read,
write or manipulate files. Often, of course, the file server
and the client is the same machine, but sometimes it isn’t.
What a file server is serving is data in files.
An X-server is a machine on a network where a “window-
ing” program exists and other machines, or X-clients, can
connect to it and create windows, write or display text, pic-
tures and so on, into a window and can read any input the
Free Software Magazine Issue 10, January/February 2006
The concept of an X-server is really no different to that of a file
server
user makes to that window. The X-clients are often run on
the same machine, but sometimes they aren’t. What an X-
server is serving is windows and your input.
I will go through a detailed example of exactly what I mean.
However, before commencing it is worth pointing out what
follows is not how X is usually run. I am simply demon-
strating some individual components to show how the in-
frastructure hangs together. First though, to install X. . .
If you have a GNU/Linux system you almost certainly have
X already installed. To run through these below examples it
is best to switch the GUI off initially if it is on. To do this
go to the main virtual console by pressing CTRL-ALT-F1,
logging in as root then enter the command “init 3”, then log
out of root and log in as your user.
If you have a Windows system you can install the Cyg-
win version of X. Go to www.cygwin.com (http://www.
cygwin.com), click on the “install now” icon, follow the
defaults. When the list of packages come up ensure the
“X11” is marked as “Default” install. When installed dou-
ble click on the Cygwin icon (or navigate to it through the
“Start” menus) to get the “$” prompt.
The X server can be started manually at the command line;
this can be done if X is not already running by simply en-
tering:
11
 USER SPACE
An X-server running by itself—as such quite boring
X &
(X in capitals then an ampersand) at the dollar prompt. This
will start an X-server in the background. On GNU/Linux
this will be in a “virtual screen”, on Windows this will be in
its own MS-Window. A black or gray patterned screen will
appear with a plain graphical “X” in the middle that can
be moved around by the mouse. This as it stands is totally
useless—yup, a complete waste of time. Although you can
move that X about by playing with the mouse, pressing keys
and clicking on mouse buttons do nothing and it doesn’t
even look pretty. This is like a formatted file server but with
no files in it, interesting from a geeky point of view, has
great potential, but currently not much to see.
Note: on MS-Windows and Cygwin, if you would rather
use the entire screen to gain the “proper” X experience then
add the “-fullscreen” option to the “X” command. Pressing
the MS-Windows key (the one with the logo on it) will get
you back to MS-Windows.
After the X-server has been started go back to the con-
sole where you entered the command (CTRL-ALT-F1
on GNU/Linux, or select the Cygwin window on MS-
Windows), and enter the new commands:
xauth add :0 . ‘mcookie‘
Please note the backquotes, punctuation and spaces in the
command. This gives you permission to open windows on
12 Free Software Magazine Issue 10, January/February 2006
An X-server with an xterm X-client connected to it
the X-server and use it. I will go into this later. After that,
enter:
xterm -display :0 &
This opens a terminal shell on the X-server. To see this
look at the “X-server” screen (by pressing CTRL-ALT-F7
on GNU/Linux, or selecting the X-server window on MS-
Windows) and there should be a terminal program running
in the top left hand corner. By moving your mouse over it
you can enter commands (such as “ls”) and see the result.
In the above, the command “X” starts an X-server, which
simply sits their looking not-so-pretty waiting for a client
to connect to it. Xterm is an X-client that connects to the
server, opens a window and runs a shell.
There are command line options to open windows of dif-
ferent sizes and at different positions, but all who I have
talked to would agree that this is a very user unfriendly
way to do things. What is needed is a X-client called a
window manager. This provides a user-interactive way to
manipulate windows in a manner we have become accus-
tomed. Cygwin X usually comes with WindowMaker in-
stalled, and can be started by either entering wmaker &
in the xterm window, or by entering wmaker -display
:0 & in the Cygwin console. Some GNU/Linux distribu-
tions may have WindowMaker too, though it’s not normally
in the default installation; however “mwm” or “fvwm” of-
ten is, and can be started by entering mwm & (or fvwm &)
 USER SPACE

An X-server running the WindowMaker window manager and A schematic diagram of some X programs connecting to an X-
some other X-clients—things are beginning to get interesting server

in the xterm window or mwm -display :0 & (or fvwm

-display :0 &) on the console. Should none of these
work then there is the somewhat old and antiquated “twm”
(which is part of the base X package) is almost guaranteed to
be installed. Once a window manager is loaded then moving Cygwin under MS-Windows have the ability to run more
and resizing of windows is interactive, and launching new than one X-server.
applications is easier. Try entering xclock & in the xterm To demonstrate, make sure you are not running any X
window to find out the current time, or even xeyes & for servers on the machine then from the console (CTRL-ALT-
an impression of the boss watching you. F1 in GNU/Linux, or the Cygwin window in MS-Windows)
Although the look and feel presented in the above is not that enter. . . .
impressive it epitomises the philosophy of X—it provides a X :0 &
versatile infrastructure giving you the choice to run what-
ever program best suits your needs. Far better look-and-feel to start the first X-server. Then switch back to the console
suites of course can be, and are, run on X. However, instead when loaded, and enter
of forcing you to use a predefined window manager as in X :1 &
MS-Windows, you have a choice from the many available
in X systems. This will have started the second. Again, switch back to the
The easiest way to quit an X-server is to hit CTRL-ALT- console, then. . .
BACKSPACE.
xauth add :0 . ‘mcookie‘

if not already done, also. . .

What’s in a display?
xauth add :1 . ‘mcookie‘
You would have noticed the “-display :0” arguments of the
commands to open windows on the X-server. A “display” to grant yourself permissions to connect to the servers, Then
is the particular screen to connect to. By default, when X enter the commands. . .
is started it will create a display called “:0”, that is display xterm -display :0 -bg yellow
zero on the local machine. However, both GNU/Linux and xterm -display :1 -bg purple

Free Software Magazine Issue 10, January/February 2006 13

 USER SPACE

If you go to the first X-server (CTRL-ALT-F7 on An Xnest session emulating an X-server with two screens
GNU/Linux, select the appropriate window in MS-
Windows) you should see an xterm with a yellow back-
ground, and on the second X-server (CTRL-ALT-F8 on
GNU/Linux) you should see one with a purple background.
You will discover it is possible to run one window manager
on one X-server, and another on the other. It is also possible
to run the X-servers at different resolutions, and logged in
as different users—though doing that is not described here.
This is like having two sets of input/display devices (key-
board, mouse and monitor) connected to the machine.
However, that’s not all—there is a sequel!. The Display “:0”
is in fact short for “:0.0”. The “dot zero” is assumed unless
another number is defined. This is the screen number of
the display. To differentiate, a display in X consists of a
keyboard, a pointy device (like a mouse) and one or more
screens. Think multiple monitors here. The display “:0.0”
xterm -display :0 &
is the first monitor, “:0.1” is the second and so on. This
is rarely used as most PCs and workstations only have one
is the same as. . .
screen connected to it. However, X allows for more—the
multi-screen aspect of X can be demonstrated using the X DISPLAY=:0
program “Xnest”, which is both an X-client and an X-server. export DISPLAY
xterm &
It runs a virtual X-server in an X-window. To demonstrate
this, from an “xterm” window (on a session that has a win-
dow manager running) run. . . When an X session is fired up, or multiple X-clients for that
matter, it automatically places the display it is connected
Xnest :3 -scrns 2 &
to in the DISPLAY variable thus reducing the requirement
for the user to specify it within the X session itself. There
(Note the capital X.) This runs an “Xnest” client emulating
is a small caveat here of course: if you should change the
an X-server with two screens using display number 3. Each
DISPLAY variable in an xterm window it can mess up any
screen is represented as a different X-window. Then from
X commands you enter from there subsequently.
the “xterm” window. . .
Should no “-display” argument be defined and if the DIS-
xauth add :3 . ‘mcookie‘ PLAY variable is not set then the X-clients will use the dis-
play :0.0 by default.
to give yourself permissions. Then. . . .
xterm -display :3.0 -bg red
xterm -display :3.1 -bg blue X in the real world

for two xterms on these (virtual) screens. Should you run
“mwm” or “wmaker” on one of these, and “twm” on another
you can also see some of the versatility of X. To quit these
simply close the Xnest windows.
To save you entering the “-display” argument every time
you enter a command, X-clients will use the display spec-
ified in the DISPLAY environmental variable should no “-
display” argument be provided. Therefore. . .
No one runs X the way described in the above examples—it
would be far too cumbersome. The usual ways are with the
shellscript “startx”, or with xdm, which really needs to be
called by the POSIX init process (xdm is not available under
MS-Windows/Cygwin).
To start X from the console, simply enter “startx”, or “xinit”
if that does not work. This will start an X server on display
:0 and run an xterm in it (though that is configurable and

14 Free Software Magazine Issue 10, January/February 2006

 USER SPACE
A gdm (pretty cousin to xdm) graphical log in screen program, or
greeter
can do something different). To run an X-server on a dis-
play other than zero then the command startx -- :1,
or even xinit -- :1 will start an X-server on display 1.
However, most X-servers are started using xdm, or one of
its close cousins such as kdm and gdm, especially in the
GNU/Linux world. Most distributions are set up to start
xdm in run level 5, and not in run level 3. That means
that entering the command “init 5” as root in a console will
switch the X suite on, and entering “init 3” will switch it
off. Most GNU/Linux distributions automatically boot up
with X enabled, including RedHat and SuSE, so X is started
on bootup. Xdm, by the way, stands for “X Display Man-
ager” and kdm and gdm are, for all due intents and purposes,
functionally identical. They just look prettier.
Xdm (and kdm and gdm) present the user with an X-client
which is a graphical log in screen called a “greeter”. Once
you have entered your user name and password not only
a window manager, but an entire desktop environment is
usually loaded, such as GNOME or KDE, that can compare
favourably with the best of the best other systems such as
Microsoft Windows can offer.
Cygwin X can be started with a “-multiwindow” argument.
This is in fact the default when using “startx”. This inte-
grates MS-Windows’ own window manager and uses it for
all the X programs too. Thus integrating the X programs
into the MS-Windows display environment.
Free Software Magazine Issue 10, January/February 2006
Distributed X
At the beginning of the article, I skimmed over the network-
ing attributes of X. I’ll now go through them in more detail.
When an X-client connects to an X-server it does exactly
that. There is nothing virtual about it. It connects using its
own protocol designed for the purpose. When the X-server
is on the same machine as the X-client the means by which it
connects is through something called “Unix Sockets”. This
is one way that a program can communicate with another
on the same machine. However, the X protocol can work
over TCP/IP. In other words, you can run the X-server on
one machine and the connecting X-clients on another.
There are both advantages and disadvantages to running the
X-server on a seperate machine to the X-client. An advan-
tage is that the machine running the word-processor for ex-
ample, is not cluttered up with the bloat of the screen han-
dling process, and will run in a smaller footprint which can
make it faster. Also it is possible to have windows of pro-
grams running on several machines opened in front of you.
A disadvantage is that screen refreshes can be slower as
some network traffic is required for these.
It is difficult to show network features with one ma-
chine so to demonstrate this you really need two ma-
chines either with MS-Windows with Cygwin installed
or Linux or a mixture. If you are using MS-XP
with SP2 you need to place the path of the XWin.exe
(C:\Cygwin\usr\bin\X11R6\XWin.exe by default) into
the “exceptions” of the MS-Windows firewall. On the
GNU/Linux machines you need to let through TCP ports
6000 through 6009 and UDP port 177. You will also need
to know the IP address of the two machines.
For the sake of this demonstration I’ll assume one ma-
chine to have the IP address of 192.168.0.1, and the other
192.168.0.2.
On the first machine (192.168.0.1) start the X-server using
“startx”, then ensure a window manager is running. In the
“xterm” window enter the command:
xhost add 192.168.0.2
or whatever the IP address of the other machine is. This will
permit connections from that machine.
On the second machine, from a console, enter the command:
xterm -display 192.168.0.1:0 &
15
 USER SPACE
A schematic diagram of X programs running on a different machine
to the X-server
Then go back to the first machine which is running the X
server and you should see a new xterm window. Further
investigations in that should show that the shell in that xterm
is running on the second machine.
What’s happening here is an X-client on the second machine
is connecting to the X-server on the first opening a window
there.
X and security
Wherever you have a network protocol you have security
issues, and X is no exception here. The fact of the matter is
that the X protocol is not suitable for anything else except a
private local area network situated behind a firewall. Should
you wish to use X through the internet then you need to
tunnel it through a program like “ssh” or similar, which will
be described later.
LANs have security issues too. If ignored one person on a
LAN can open and close windows on another’s willy-nilly
without permission. That wouldn’t be good. Don’t worry
though, there are various ways you can protect machines
and only permit allowed connections to an X-server. The
first is using the “xhost” program as described above. How-
ever, this isn’t recommended as it tends to leave the X-server
16 Free Software Magazine Issue 10, January/February 2006
too open. I have only used it in the above example because it
was the easiest way to demonstrate the networking attributes
of X. A far better way is with the “xauth” program, which
can use a number of ways to authenticate clients, the main
one using “cookies”. In this, for an X-client to connect to
the X-server it needs to know a piece of data randomly gen-
erated, or accessed, by the server known as a cookie. It
works somewhat like a session password. If this cookie is
not transmitted by the X-client, the X-server will reject the
connection. I won’t go into the details of it here because
starting X using xdm automatically sorts this out, and is the
recommended way.
Over the network, the X server, by default, listens on port
6000 for display :0, 6001 for display :1 and so on.
Xdm in detail
The xdm program, and its gdm and kdm cousins, are small
daemons that run in the background usually started by the
“init” program at boot up, and run all the time. It does things
such as start appropriate X-server(s) where necessary (it is
not one itself), run the graphical login screen programs, sort
out the xauth cookies when people log in and start the appro-
priate programs for each desktop and so on. There is a com-
prehensive configuration file that controls what it does. The
first thing it usually does is to run the appropriate X-servers
on the local machine. Usually this is just one, but can be
more. It then runs the graphical log in a screen known as
a “greeter” on the X-servers that it has been configured to
do so. These are usually the X-servers that the xdm process
starts. However, it can start it on other X-servers, such as
workstations directly connected to the machine, or even not
start it on any, including its own X-servers.
Xdm can be configured to accept special XDMCP requests
through the network. These are special UDP packets that
an X-server can transmit (usually on port 177) to request a
client machine’s xdm to start a graphical login greeter pro-
cess for it. This would then permit remote logins. On most
distributions XDMCP can be enabled through a GUI from
the desktop. On Red Hat Fedora Core this is done from the
menu:
Desktop -> System Settings -> Login Screen: (root pass-
word)
Here on the XDMCP tab the check box titled “Enable
XDMCP” should be enabled. All other settings should
 USER SPACE

be unchanged. A similar configuration exercise would be Gdm’s chooser program: selecting a machine restarts the X-server
needed for other distributions. and displays the appropriate machine’s greeter
Xdm can be configured to accept special XDMCP requests
through the network allowing remote logins.
When an X server needs to “be connected” by another ma-
chine running xdm then the X-server needs to be started
with the “-query” or “-broadcast” option. The “-query” op-
tion tells the X-server to send an XDMCP request to a spe-
cific machine, whereas the “-broadcast” transmits one to all
machines on the network and will use the first machine to
reply.
For example. . .
X -query 192.168.0.2

or
X -broadcast

I use this in real life. When I am working on my windows

machine, and need to do things on my Linux box, I open a
Cygwin session and enter:
X -fullscreen -broadcast
This presents me with my graphical login greeter for my
GNU/Linux box.
Spoilt for choice
There are occasions when there are more than one machine
running xdm on the network. For this scenario a setting
can be tweaked so that xdm runs a “chooser” instead of the
usual login greeter. In this case a list of machines is pre-
sented to the user. Upon the selection of one, the X server
is restarted, connecting to the appropriate machine chosen
displaying that login greeter.
This is useful for maintaining a large number of GNU/Linux
or POSIX machines on a network. You only need one X-
server on your desk (which can be MS-Windows and Cyg-
win), and you can run xdm on all the GNU/Linux machines
you maintain. As xdm doesn’t need to run its own X-server
these can be headless, in so far as that they need no video
hardware or drivers. In short, you can have full graphical ad-
ministration functionality on all GNU/Linux machines that
you maintain on the network from a single workstation, with
a minimum of resources used on the GNU/Linux machines.
You only need one X-server on your desk to graphically ac-
cess a large number of machines.
X through the internet
As mentioned before, the X protocol is not really suitable
for the internet from a security point of view. Getting xdm
or the authentication to work correctly from afar would be
difficult. However, you can tunnel the protocol through ssh.
It is worth noting that for this to work the “ssh” server
sshd needs to be configured to permit X requests. This is
done by setting the “X11Forwarding” option to “yes” in the
/etc/ssh/sshd config file.
In the following examples it is assumed you are trying to
access a remote server called “centralbox.com”, say the ma-
chine at your office, from a local user machine perhaps from
your home.
To demonstrate running an X program using ssh start an
X-server using “startx” on the local user machine. Ensure a
window manager is running then in the Xterm connect using
ssh to the central machine using the command:
ssh -C -X myuser@centralbox.com
where “myuser” is your user code. When this connects
you can enter X-client commands on the ssh server and
the windows will open on the user’s machine. As the Cyg-
win package also includes ssh this can be done from a MS-
Windows/Cygwin box as well as a Linux one.
It is worth taking some time out to explain how ssh does this.
When an ssh client connects to the ssh server using the “-

Free Software Magazine Issue 10, January/February 2006 17

 USER SPACE
A schematic diagram of X programs running through the network
using SSH X forwarding
X” command, the ssh server on the central machine sets up a
“virtual” X-server on a higher display number (usually start-
ing at 10); it then creates its own authentication cookie for
it. When an X-client (on the ssh server) opens up a window
it connects to display 10, the pseudo ssh X-server validates
the X-client using its cookie and encrypts and transmits the
request to the ssh client on your user machine. This then
authenticates the request on the real X-server using the real
cookie and passes through the X requests to it for display-
ing.
The “-C” argument is in fact optional. It compresses the
data to transfer over the network decompressing it at the
other end. This can improve performance especially over a
slow internet connection.
This means that all transmissions through the network are
encrypted and compressed. Also, the ssh server is unaware
of the real X-server’s cookie, maintaining the X-server’s se-
curity integrity but at the same time allowing the remote
user’s machine to use the GUI in an easy and transparent
way.
Should you wish to run your desktop through ssh rather
than an individual program then you simply run “Xnest”
through the tunnel. Beforehand you would have had to en-
able XDMCP, then from an xterm on the user’s local ma-
chine run:
ssh -C -X myuser@centralbox.com
18 Free Software Magazine Issue 10, January/February 2006
to connect to the central ssh server forwarding X signals
and compressing data. After authentication and connection
enter:
Xnest :20 -query localhost
which will then start a “nested” X-server on display 20 of
the central machine, displaying the output of which on a
window of the local machine. The login greeter then should
appear.
If a different screen resolution is required the “-geometry”
argument can be used to specify one:
Xnest :20 -query localhost -geometry 1280x1000
To follow what is happening here may seem a bit compli-
cated, so watch out for this next bit. First of all, I need to
explain the Xnest program in detail. This is both an X-client
and X-server. The best way to imagine it is as an X-client
that creates a window on the session it was started from.
And, at the same time, an X-server that instead of display-
ing and accepting its goods on its own physical screen, key-
board and mouse, uses the window it created in the session
from which it started.
Now for the meat of it. On the ssh server you are running an
X-client Xnest, which is connecting to the X-server on the
ssh client through the network tunnel ssh provides. How-
ever, Xnest is also an X-server, which is running on the ssh
server providing an environment where you can run your
desktop remotely, and securely, through the network but as
though you were on the local machine. Confused? Well, I
did warn you. I hope the following diagram helps, but don’t
worry if you still don’t get it, this is not essential knowledge.
If two or more people are likely to connect at a time in this
way, then each needs to use a different display number. A
way to do this is to simply add a number to the display num-
ber the ssh server gives you:
NDISPLAY=‘echo $DISPLAY | cut -d: -f2 | \
cut -d. -f1‘ # Get the main display no.
NDISPLAY=‘expr $NDISPLAY + 30‘# Add 30 to it
Xnest :$NDISPLAY -query localhost
If you use this I recommend placing it in a shellscript. The
above should work if no more than thirty people are con-
nected to the machine at any one time.
You can also run X through a VPN such as IPsec or PPTP
and connect using XDMCP as described above.

A schematic diagram of an X desktop running through the network
using SSH X forwarding and Xnest
Dynamic duo: X and VNC
Using the X protocol, even through ssh, can have a perfor-
mance hit. The X protocol will transmit every line, area
and character whether required or not. There’s no cache-
comparisons or transmit-changes-only functionality. This
means that if an area of the screen is redrawn three times
between user inputs the X protocol will transmit that area
three times whereas it only really needs to transmit the last
of the three. This can make running X over a slow network
like a dial-up painful. VNC’s RFB protocol helps to take
care of this problem, and X interfaces have been written for
it.
VNC is a cross-platform facility to access desktops re-
motely using a more conventional client/server mechanism.
The VNC server is located on the machine where you would
have the GUI and listens for connections from a VNC client.
When a client connects, VNC transfers the entire desktop to
the client where a user can work on it.
The program used for using VNC in a remote X-session en-
vironment is “Xvnc”. This is both an X-server and a VNC
server. Xvnc, when started, listens on a VNC TCP port
number (5900 + display number) for any VNC clients to
connect to it. It also listens on the display socket and TCP
port like any other X-server, but instead of displaying the
requests on a monitor or screen it caches it in memory to
transmit it to a VNC client that is connected to it.
VNC’s X-server—Xvnc—is included with
Free Software Magazine Issue 10, January/February 2006
USER SPACE
GNU/Linux distributions. If it isn’t installed, it can
normally be done easily through the distribution’s package
manager. The VNC client, on GNU/Linux, is called
“vncviewer”. There’s also a native MS-Windows version of
this, so it’s possible to log into GNU/Linux desktops from
remote Windows machine without installing the consider-
ably large Cygwin and X-Windows packages. Should you
need a VNC client for MS-Windows one can be downloaded
from www.realvnc.com (http://www.realvnc.com).
And if the machine you are connecting from doesn’t have
a VNC client installed all is not lost, as there’s a Java one
that comes with the server.
There are several ways the VNC server can be set up on
a GNU/Linux box; I’ll deal with two of them here. Both
require the VNC server to be fired up manually.
The first is using a wrapper called “vncserver”. To pre-
pare for this you need to set up a VNC password. This is
achieved by logging into the GNU/Linux server and enter-
ing the command “vncpasswd” at the $ prompt. You will
then be prompted to enter a password, which needs not be
the same as your log in password (in fact it’s better if it’s
not) but needs to be secure, i.e a mixture of upper and lower
case letters, numbers and symbols.
The way to connect remotely this way is to log in using ssh,
but without the “-X” option:
ssh myuser@centralbox.com
then when connected, enter the command on the central ma-
chine. . .
vncserver :1
This will start an Xvnc process up and fire off a few X-
clients running inside it (such as window manager and so
on). It will accept X-clients on display “:1” and VNC client
connections on TCP port “5901”. Should “:2” be used in-
stead it will accept X-clients on display “:2” and connec-
tions on port number “5902” and so on. If you leave the
display number from the command, then an unused one is
assigned and reported as the output of the command. You
can control the size of the screen by using the “-geometry”
argument. This works the same way as described in the
“Xnest” program.
After the server has fired up, from the user’s machine start
most up the VNC client. When asked for the VNC server enter
19
 USER SPACE
A schematic diagram of an X desktop running through the network
using VNC
“centralbox.com:1”. You should then be prompted for the
VNC password; upon successful entry you will be presented
with a desktop. If you don’t have a VNC client installed
then you can use the Java one. To do this point a browser to
“http://centralbox.com:5801” (5802 if display :2 was used
and so on) and a java client will be fired up.
The precise X-client programs started here will depend on
the shellscript “.vnc/xstartup” on the central server in the
specific user’s home directory. This can be tuned accord-
ingly.
When finished, in the original ssh screen enter. . .
vncserver -kill :1
which will stop the server.
The second method is to use the XDMCP features with
VNC. For this to work XDMCP needs to have been enabled
on the central server as described above. To connect access
the central server from the user’s local machine with ssh in
the same way as above. . .
ssh myuser@centralbox.com
Then, when connected, enter the command. . .
Xvnc :1 -query localhost -securitytypes none -once
This will start Xvnc the same way as above except it
will simply display the greeter in the X session. The “-
geometry” argument works the same way here too, as does
20 Free Software Magazine Issue 10, January/February 2006
alternative display numbers. Connecting to this using VNC
(or the Java) clients are no different with the exception that
no password is asked for.
Dynamic trio: X, VNC and SSH
Although the above works, there are security issues. Ports
590x and 580x need to be opened up if there is a fire-
wall. Everything with the exception of the VNC password
is transmitted over the internet unencrypted. A far better
method is to tunnel the connections through our old friend
ssh.
To connect using this method, from the user local ma-
chine. . .
ssh -C -L5901:localhost:5901 \
-L5801:localhost:5801 \
myuser@centralbox.com
Please note the “-X” argument is not used here. Then, when
connected, on the central machine enter:
vncserver :1 -localhost
or
Xvnc :1 -query localhost \
-securitytypes none -once -localhost
depending on how you want the X server to be. Back at the
local machine the VNC server to connect to is “localhost:1”
and the URL to point the browser at for the Java client is
“http://localhost:5801”.
After finishing, don’t forget to enter the command
vncserver -kill :1 if the “vncserver” command
was used.
The above works for display :2 if 5802 and 5902 ports are
used instead as before.
VNC through ssh provides a secure, efficient, fast and ver-
satile means of accessing the desktop remotely
I will now go through go through the mechanics of this, but
again be warned, it’s complicated too!
The ssh command creates two tunnels, one for TCP port
5901 and the other for 5801. The VNC clients (and possi-
bly browser clients) then connect through these, encrypting
the data, to form the connection. The attributes of the tun-
nel mean that the user clients need only connect to ports
 USER SPACE
A schematic diagram of an X desktop running through the network
using VNC and SSH
listening on the user’s local machine that are created by the
ssh client for this purpose. The data is then encrypted and
transferred to the central server where they are unencrypted
and regenerated by the ssh server process. A connection is
established from that to the VNC server which is now local
to it.
What? That, complicated? Naahh. . .
This has the best of a number of worlds. The data transfer
is encrypted and compressed by ssh and the GUI windows
are properly cached by VNC, creating a reasonably fast and
very secure open platform remote desktop facility. How-
ever, should more than one user need to connect this way
to the central server simultaneously, then each user should
be given a unique display number which they would always
use.
For completeness I need to point out that there are other
methods which can be used to fire up Xvnc, including start-
ing it as a service. These are documented as part of the VNC
distribution and I won’t go into it here.
Conclusion
A popular misconception exists that X is only available for
GNU/Linux and some UNIXes. The fact is that X was de-
signed to be platform independent and is available for many
other systems as well. This article mentions many times the
free X suite that runs under Microsoft’s Windows with the
Free Software Magazine Issue 10, January/February 2006
use of Cygwin. On top of that, this article demonstrates how
a MS-Windows machine, as well as GNU/Linux ones, can
be used as a terminal for a central GNU/Linux system both
on a LAN and through the internet.
One project that makes good use of the versatility of X is
the Linux Terminal Server project (http://www.ltsp.
org). This uses X features to create scenarios where many
users are connected to a central GNU/Linux machine using
thin clients producing massive running and installation cost
savings.
This article has only scratched the surface of the X infras-
tructure. Hopefully though, it has demonstrated X’s con-
cepts, proven its versatility and given a taste of its power.
While the Microsoft Windows GUI is functional and intu-
itive, I’m finding it now no more so than GNOME or KDE
on X running on GNU/Linux. And when I need to do some
remote accessing or use other machines as terminals, I find
the Microsoft Windows infrastructure to be encumbered,
restrictive and clunky compared to the slickness of X. In
short, if given a choice between Microsoft’s own window-
ing environment and that of the free software community,
GNU/Linux and POSIX, then you can give me an “X” ev-
ery time.
Copyright information
Copyright Edward Macnaghten (C) 2005
Permission is granted to copy, distribute and/or modify this
document under the terms of the GNU Free Documentation
License, Version 1.2 or any later version published by the
Free Software Foundation; with no Invariant Sections, no
Front-Cover Texts, and no Back-Cover Texts. A copy of the
license is available at http://www.gnu.org/copyleft/fdl.html
About the author
Edward Macnaghten has been a professional pro-
grammer, analyst and consultant for in excess of 20
years. His experiences include manufacturing com-
mercially based software for a number of industries
in a variety of different technical environments in Eu-
rope, Asia and the USA. He is currently running an
IT consultancy specialising in free software solutions
based in Cambridge UK. He maintains a web site at
http://eddy.edlsystems.com.
21
Browsers for Mac OS X
Comparing FOSS browsers for Mac OS X
Martin “MC” Brown
hen Apple migrated the Mac operating
W system platform to Mac OS X, one of
the key components was an underpinning
based on the FreeBSD operating system.
The use of an open source operating system as the core has
in turn led to an increase in the use and availability of free
and open source software (FOSS). It is now much easier to
develop software for the OS X platform (development soft-
ware is included, instead of being an expensive addition)
and this makes it both easier for people to get involved and
more likely to take part in open source community projects.
Despite the ease, there are still some areas of software de-
velopment where the complexities of the application are too
great. Web browsers are a classic example of this—although
the principles of web browsing are quite simple, making all
of the components of a typical web page (JavaScript, dif-
ferent image formats, plug-ins, CSS etc.) work effectively
is quite difficult. Hence the need for community-based
projects where many people can pool their experience.
As a knock on effect, web browsing in the free software
space is now based around two main camps, those based
on the Mozilla codebase (Camino, Firefox, Mozilla) and
those based on the KDE sourced KHTML rendering engine
(Safari and OmniWeb). Projects based on the former are
completely free software, with both the rendering engine
and interface being a free software project. The latter are
not strictly free software, but the core rendering engine on
which they rely is. It is only the interface which is not free
software.
22 Free Software Magazine Issue 10, January/February 2006
For the review, I’m going to be taking a closer look at
the features and functionality of the three free software
browsers—Mozilla, Firefox and Camino, and what differ-
entiates the functionality of these three browsers from each
other. I’ll primarily be comparing speed and functionality,
but I’ll also use the opportunity to cover some of the issues
that affect the browsers on individual sites. I’ll then take a
very brief look at the quasi-free software browsers; Safari
and OmniWeb.
---------------------------------------
Web browsing in the free software space
is now based around two main camps,
those based on the Mozilla codebase
(Camino, Firefox, Mozilla) and those
based on the KDE sourced KHTML
rendering engine (Safari and OmniWeb)
---------------------------------------
Mozilla Suite
The Mozilla project was borne out of the Netscape Naviga-
tor project; Mozilla being a combination of the term Mo-
saic killer (which is what they hoped Netscape would be)
and a reference to Godzilla. Mozilla was also the name of
the Netscape mascot—appropriately enough usually repre-
sented by a humanoid-lizard, like Godzilla. The Mozilla Or-
ganization was created in 1998 to create the next generation
of Netscape and was registered as a not-for-profit organiza-
tion (the Mozilla Foundation) in 2003. Today, Mozilla is the
 USER SPACE
MCslp.com in Mozilla
name used to refer to the range of browsers produced by the
Mozilla group, notably the main Mozilla Suite and the forks
of Firefox and Camino.
The Mozilla Suite is the name given to the software pack-
age that incorporates the functionality of a browser, email,
newsgroup reader and HTML editor. The core of the
Mozilla suite is the Gecko rendering engine which is also
shared by Firefox and Camino. The Gecko engine is based
on the HTML/XHTML, DHTML and other standards that
make up the web environment and is specifically designed
to render these in a consistent fashion across platforms. This
means that you browse a page on Mac OS X and Windows
within a Mozilla browser and get a consistent view. It also
means that, in theory, different browsers that use the engine
will have a consistent view. This isn’t quite the case for rea-
sons that will become clearer as I look at the other browsers
from the Mozilla stable.
For example, let’s look at the homepage for MCslp.com
(http://www.mcslp.com), my website, which is
based on the WordPress blogging engine with a theme
which is XHTML compliant. You can see how this looks
in Figure 1.
As the larger “internet suite”, the Mozilla browser is more
geared toward users who want a consistent interface and en-
vironment for all of the browsing and internet needs. The
interface, as shown in the figure, is based around the orig-
inal Netscape browser. Shown is the Classic theme but a
standard installation also includes the Modern theme which
Free Software Magazine Issue 10, January/February 2006
is a little more aesthetically pleasing. This retains the “tool-
bar” style of the original—including “handles” for moving
the different toolbars around and changing their order and
layout. You can also click on these tabs to “collapse” the
tool bar so that it takes up less space. What you don’t get
in the standard format is a convenient search bar for Google
or other sources; instead you get a button that takes you to a
predetermined Search button and the location field doubles
up as the source for search criteria. What you cannot do is
customize the toolbar you can enable and disable buttons,
but not actually modify or add to the suite of buttons.
Like Firefox, Mozilla supports both themes—which change
the appearance of buttons and other aspects of the inter-
face—and extensions, which provide additional function-
ality. For example there are extensions that provide Ad
filtering, download management or additional information
during browsing. For example, the LiveHTTPHeaders ex-
tensions can display the header information sent during an
HTTP transaction—invaluable during development. Unlike
Firefox there is no convenient extensions manager, and in
fact Mozilla almost hides the existence of extension support
from the application.
In terms of simple internet browsing Mozilla is more than
capable. It does sometimes feel slower than the alternatives
and there is often a discernable pause just before the page
is displayed that can become a little frustrating. For simple
text sites it isn’t a problem, but those with heavy graphics,
or complex tabular pages do demonstrate the problem.
In terms of the quality of the display and parsing, I experi-
ence very few problems with Mozilla when laying out dif-
ferent pages and sites. The only times you experience any
specific difficulty is either when the site is not standards
compliant, in which case some interesting “default” choices
are made about how to layout components, or where the site
has deliberately used Internet Explorer capabilities to pro-
vide functionality. For example, some menus and items that
rely on JavaScript or ActiveX simply don’t work. I’ve also
experienced some occasional issues when logging into sites
through a simple password system. For example, my blog
at Computerworld never loads properly after login. Again, I
suspect this is more of an issue of the site code, rather than
Mozilla, but it can be a frustrating experience.
Being a suite of internet applications, rather than just a
bare browser, Mozilla has the benefit of including most
of the tools you will need. I also love the way that you
23
 USER SPACE
can read/reply to email directly from within Mozilla. This
comes in particularly handy if you are frequently using sites
that use email links; there is none of the delay sometimes
experience while the OS works out what to do with the link.
The embedded IRC client is also very handy. However,
once you start using all of the components at the same time,
you can sometimes experience some performance problems.
Also, as a single monolithic application the memory usage
can be prohibitive. Although memory management in OS
X is efficient enough that you don’t often experience prob-
lems, the memory issue is almost certainly related to some
of the performance issues I experienced, especially while
monitoring the load and memory usage of Mozilla while
browsing various pages with all of the options enabled.
I liked Mozilla, but I have to admit to preferring other ap-
plications for my Email and chat requirements, and this is
where Mozilla, as the “do all” application, starts to look like
overkill if all you want to do is use the browser. A good ex-
ample here of the limitations of the built in system is the
address book. Within Mozilla you have your own special
address book section, but this doesn’t integrate with the OS
X address book and that means you can’t share instant mes-
saging details with iChat and there is no provision for SMS
messaging, an option built into the standard OS X Address
Book application.
Where Mozilla is most likely to be used is within an organi-
zation where they want to standardize on a single platform
for browsing and email access, and are probably supporting
an LDAP server for address book functionality. In fact, I
deployed Netscape Communicator, which provides similar
functionality, in this way several years ago.
Firefox
If you removed the web browsing component from Mozilla
and turned it into a separate application, you would es-
sentially get the Firefox application. It is a bare-knuckle,
browser only, application that combines the Gecko display
engine on top of the Mozilla application base to produce
a light-weight internet browser. Because it’s built on the
Mozilla core you can add extensions (just as with Mozilla,
although with a convenient extensions manager) and adjust
the theme and display of the browser.
Because Firefox is just a browser it is a direct competitor
to the standard Safari browser supplied with OS X and as
24 Free Software Magazine Issue 10, January/February 2006
Mozilla Suite
Name Mozilla Suite
Maintainer(s) Mozilla
License Mozilla Public License
(OSI Approved)
Platforms Linux, Unix, Mac OS X,
Windows
MARKS (out of 10)
Installation 10
Vitality 10
Stability 9
Usability 9
Features 10
a cross platform solution has obviously gained some pop-
ularity as a more secure alternative to Internet Explorer on
Windows. Since Microsoft have dropped support for Inter-
net Explorer support under OS X, it is often the main choice
if the user is unhappy with Safari.
Fortunately, Firefox has a lot of additional benefits that
make it a popular alternative. The support for extensions
enables the use of ad blocking and enhanced security func-
tionality that Safari does not support. Many also find Fire-
fox faster than Safari, although I have to admit to finding so
little difference between the two that this wouldn’t be my
reason for choosing Firefox. The ad blocking extensions
though are invaluable, and a combination of AdBlock and
the filter updater (which automatically reloads new filter sets
for you) can eliminate nearly all ads to the point where only
interstitials remain. As with Mozilla, the rendering of Fire-
fox pages is generally excellent, all because of the Gecko
engine. The same, occasional, site-specific problems occur.
Unlike Mozilla, Firefox is more flexible in its interface.
Toolbars are customizable, and, as can be seen in Figure 2,
you get a convenient search field that can be used to search
a variety of locations simply by selecting an alternate en-
gine from the embedded popup. Tabs are also slightly bet-
ter supported with switching between tabs noticeably faster
(for me at least), and with convenient hot keys (Command-
1, Command-2, etc.) for flipping between multiple tabs in a
window.
Firefox is undoubtedly a very fast browser and is very quick
 USER SPACE
MCslp.com in Firefox
and easy to use. With the extensions, particularly ad block-
ing, Firefox becomes an even faster alternative to the full
Mozilla Suite product. You may not get the embedded
mail/news reader, address book and IRC, but you get the
flexibility to use other clients that may contain richer fea-
tures (I find Apple’s Mail best for Email), and the reduced
memory footprint is certainly an improvement if you have
512MB of RAM or less. If free software mail/news tools
are important then Thunderbird, also from Mozilla, is their
standalone email application.
Camino
The Camino project is based on the same rendering engine
as both Mozilla and Firefox, but unlike these two packages,
the Camino interface is designed to be more Mac OS X like,
rather than the architecture neutral interface offered by its
brethren. The effect provides all of the power and speed of
the Firefox browser—Camino is just a browser—but with
the OS X look and feel.
The architecture neutral format of both Mozilla and Fire-
fox is obviously a benefit in a cross platform environment,
but some prefer the look and feel of OS X. Most of the ba-
sic look and feel remains the same; key differences when
browsing are the use of the OS X style popups, fields and
buttons in the display, as you can see here in Figure 3. The
most significant differences are in the dialog boxes. For ex-
Free Software Magazine Issue 10, January/February 2006
Firefox
Name Firefox
Maintainer(s) Mozilla
License Mozilla Public License
(OSI Approved)
Platforms Linux, Unix, Mac OS X,
Windows
MARKS (out of 10)
Installation 10
Vitality 10
Stability 9
Usability 10
Features 10
ample, the preference panel looks and works almost identi-
cally to the main OS X System Preferences application.
The other main aim of the Camino project is to produce
a very fast browser. Based on the same Gecko engine the
Camino project includes some improvements and uses the
functionality of the OS X display engine to help improve
the overall speed of displaying web pages. The result is
quick—pages display very quickly, even the complex ones.
There are some problems, well documented on the site, that
can cause some interesting effects, but as a product still con-
sidered young and in pre-release form (the latest at time of
writing was 0.8.4), Camino is still impressive.
Although Camino includes a lot of integration with OS
X—for example the bookmarks integrate with Rendezvous
and the standard Address Book to include sites described in
these repositories—Camino is not a typical Mozilla project.
The extensions and themes supported by both Mozilla and
Firefox are not supported. Although the themes limitation
makes sense (since Camino is essentially permanently in the
“OS X” theme) the lack of extension support means that
we cannot block ads. Ironically, this often has the effect of
severely slowing down the browsing experience. Most of
the extensions are not supported because they rely on the
XUL interface for configuration, which is not included in
the Camino project.
Although Camino is very quick and is the most OS X-
friendly of the Mozilla applications, the occasional display
problems can be a pain. Also, the lack of extension support,
25
 USER SPACE
MCslp.com in Camino
which eliminates the ability to filter adverts, is a feature that
Camino could certainly take advantage of. If you browse
an intranet or other controlled sites where adverts are not a
problem then Camino is probably the best of the bunch, but
if you need ad blocking you will need to consider Firefox.
Safari and OmniWeb
When Apple launched Safari the OS X community was
rightly amazed. It included a number of features that in to-
days browsers we take for granted. Easy access to Google,
for example, was functionality we didn’t have in Internet
Explorer, and Firefox was not available at the time. Snap-
back functionality (which takes you back to a Google, or
indeed any other, page after clicking through other choices)
makes it easy to browse sites without losing your root.
Finally the integration with the operating system is abso-
lute—in the latest versions, you can even synchronize your
bookmarks between Macs using the Apple .Mac service.
Most interesting of all, Safari uses the KHTML rendering
engine. KHTML is part of the Konqueror browser and the
KDE Project, a desktop environment commonly used un-
der Linux. The KHTML engine is incorporated into what
Apple call the WebKit core, which provides HTML render-
ing for a variety of systems within OS X, including Safari,
Sherlock and the built-in help system. The KHTML engine
is a mature product and the rendering of HTML pages is
based on the same open standards for HTML that drive the
26 Free Software Magazine Issue 10, January/February 2006
Camino
Name Camino
Maintainer(s) Mozilla
License Mozilla Public License
(OSI Approved)
Platforms Linux, Unix, Mac OS X,
Windows
MARKS (out of 10)
Installation 10
Vitality 10
Stability 9
Usability 10
Features 8
Mozilla project. There are some circumstances, just as with
Mozilla, where you get odd rendering issues but they are
few and far between and again very site specific. The use of
free software is perhaps not that surprising—Mac OS X is
itself built on top of the free software Darwin project which
is in turn based on the FreeBSD operating system.
Unfortunately, although KHTML is a free software project,
Safari is not and therefore cannot strictly be included in this
roundup review. In terms of functionality, Safari is probably
closest to the Camino project, it supports browsing tabs and
integrates well with the rest of OS X, but it does not do ad
blocking as standard. There are however some products that
support this (PithHelmet for example).
Also based on the KHTML engine is the OmniWeb browser
from OmniGroup Software. OmniWeb is a commercial
product and is one of the richest of the browsers available
for the platform. It supports a number of features designed
to make it easier for people who do a lot of browsing to
organize their environment. For example, as well as sup-
porting text tabs, OmniWeb also supports icon tabs, which
show a live thumbnail of the tabbed page. You can also cre-
ate multiple workspaces—these are groups of windows (in-
cluding all the tabs in each window) and the information in a
workspace is retained, even across application restarts. This
means that you can create individual workspaces for differ-
ent projects or tasks; I have separate workspaces for shop-
ping, business work, recreation and a “transient” workspace
that I use to browse the pages that I’ll probably only ever
 USER SPACE
MCslp.com in Safari
view once.
OmniWeb also includes a built-in ad filter and as an added
security measure you can configure security settings on a
site by site basis. For example, you can enable JavaScript
on your financial websites, while having a default setting
where JavaScript is disabled. All of this additional security
and functionality is built right into the browser, making it
more convenient than Firefox or Mozilla which support this
functionality only through additional extensions. As a com-
mercial product though, these features do, quite literally,
come with a price. Still, OmniWeb remains my favourite
c 2005 Martin C Brown
browser.
Conclusion
If what you want is a plain and simple web browser that
is free software based then really your choice is limited to
Firefox or Camino. Mozilla, while a very capable browser,
is technically a complete internet suite and you will only
get the best out of Mozilla if you are also happy to use the
included email, news and IRC clients.
Camino offers the best all round environment, from an OS
X perspective, simply because it offers the best integration
with the system and retains the look and feel of the operating
system. However, Camino does not support the entire range
of Mozilla extensions and this means that if you want to
avoid adverts or improve the security of your browser you
may want to consider the Firefox project instead.
Free Software Magazine Issue 10, January/February 2006
MCslp sites in OmniWeb
Firefox is fast, configurable and supports all of the exten-
sions and themes you would want to use. It lacks integration
with the operating system, storing authentication informa-
tion in its own database instead of the keychain for exam-
ple, and there is no syncing of bookmarks through .Mac or
iSync. However, it remains the strongest choice in the free
software market.
Copyright information
This article is made available under the “Attribution-
NonCommercial-NoDerivs” Creative Commons License
2.0 available from http://creativecommons.org/licenses/by-
nc-nd/2.0/.
About the author
Martin “MC” Brown is a freelance writer and consul-
tant. He works with Microsoft as an SME, is a fea-
tured blogger for ComputerWorld, a founding mem-
ber of AnswerSquad.com, Technical Director of Food-
ware.net and, and has written books on topics as di-
verse as Microsoft Certification, iMacs, and free soft-
ware programming.
27
GRUB tips and tricks

Spicing up a great utility for more IT fun

Jeremy Turner

he GRand Unified Boot loader, or GRUB, has

T
Figure 1: you can run GRUB on a USB flash memory key!
all but replaced the default boot loader on many
GNU/Linux distributions. It includes some
conveniences over LILO, the LInux LOader.
One advantage is not having to remember to run /sbin/lilo
every time you make a configuration change. It also can
function as a boot loader for removable media such as flop-
pies, CD-R/W and USB flash memory keys. It is short-
sighted to view GRUB only as a boot loader to be installed
on a hard drive of a GNU/Linux system. Combined with
a few other utilities, GRUB can be a powerful and good-
looking tool for your home, organization or workplace.

Introduction

First, what exactly is GRUB? GRUB is a boot loader, which

means it passes control of the boot process from the Power-
On Self Test (POST) to the kernel of your GNU/Linux dis-
tribution. GRUB works in a modular, layered fashion so that
any unneeded modules are not loaded. Not only does this
reduce execution time, but it saves valuable resources when
running from removable media. GRUB optionally loads its
configuration file at run/boot time, so you don’t have to type
in commands manually each time. However, the command-
line option is still available in case there is an error in your
configuration file. So why use GRUB when there are other
options out there? The beauty of free software is that you
have choices. Alternatives to GRUB include LILO, syslinux
and isolinux. The benefit of GRUB is that it will work in
many different types of boot devices, but you only need to
learn one set of menu commands. In addition, GRUB can
work on other forms of bootable storage, such as CD-R/W,
USB flash memory keys, floppy disks, and even via a TFTP
server with PXE ROM booting.
Installing GRUB on a USB flash memory key
I got the inspiration for this article after trying DSLinux (or
DSL), which is a fully graphical Linux distribution weigh-
ing in around 50 MB. After seeing an advertisement on their
website for a USB flash memory drive with DSL installed,

28 Free Software Magazine Issue 10, January/February 2006

 HACKER’S CODE

I figured I could probably learn how to set DSL up myself Figure 2: this is the file structure on my USB Memory Key
on my Lexar 256 MB JumpDrive. The DSL documenta-
tion pointed towards installing via syslinux and reconfigur-
ing the cylinder/head/sector information of my JumpDrive,
but I didn’t have any luck trying to get my USB flash mem-
ory key to boot successfully. Finally, I tried using GRUB
and I was up and running with DSL in no time!
---------------------------------------
GRUB is a boot loader, which means it
passes control of the boot process from
the Power-On Self Test (POST) to the
kernel of your GNU/Linux distribution
---------------------------------------

First, I recommend creating a directory structure to orga-

nize your boot-related files, and keep them separate from
any other files you’d like to keep on the USB flash mem-
ory key. You could create two partitions, but I couldn’t get
both partitions to load correctly when I inserted the key back
into Windows. On my USB flash memory key, I created a
root folder named boot to hold all the data necessary for
USB booting (see figure 2). Under the boot folder, I cre-
ated a directory named grub for GRUB-related files, images
which are initial ramdisk (initrd), floppy, or disk images,
and finally kernels to hold all the kernels. You may want
to organize your boot folder differently, but make sure that
you change the corresponding paths and directory names in
GRUB’s menu.lst file. The menu.lst file that I use can be
found in Sidebar 1.
Next, you’ll need to copy some of GRUB’s stage files,
including stage1, stage2, and fat stage1 5, and put them
into the boot/grub directory on the USB flash memory key. grub> find (hd2,0)/boot/im<TAB>
These will allow GRUB to boot into GNU/Linux and other grub> find (hd2,0)/boot/images/
operating systems. After the files are copied over, it’s time
to install GRUB to the Master Boot Record (MBR) of the Since the /boot directory on /dev/hda doesn’t have an im-
USB flash memory key. ages directory, I know that (hd2) is the hard drive that I want
Luckily, it’s the same process as installing to a hard drive: to install GRUB on:

# grub grub> root (hd2,0)

grub> find /boot/grub/stage1
(hd0,1)
(hd2,0)
On my system, hd0 is /dev/hda and hd2 happens to be
/dev/sda. Just to make sure, we can use a bash-like tab com-
pletion to look through a filesystem:
Filesystem is type fat, partition type 0xb
grub> setup (hd2)
Checking if "/boot/grub/stage1" exists... yes
Checking if "/boot/grub/stage2" exists... yes
Checking if "/boot/grub/fat_stage1_5" exists... yes
Running "embed /boot/grub/fat_stage1_5 (hd2)"...
15 sectors are embedded.
succeeded
Running "install /boot/grub/stage1 (hd2)

Free Software Magazine Issue 10, January/February 2006 29

 HACKER’S CODE

(hd2)1+15 p (hd2,0)/boot/grub/stage2
Figure 3: DSLinux is a 50 MB fully-graphical live GNU/Linux dis-
/boot/grub/menu.lst"... succeeded
tribution
Done.
grub> quit

Great! Now we have GRUB in the USB flash memory key’s

MBR. Now, we have to put some files on the memory key
to boot into and create a menu.lst file!
---------------------------------------
When you boot from the USB flash
memory key, the key itself becomes hd0,
even before the primary master hard
drive
---------------------------------------

GRUB with disk images

One cool trick is to use GRUB and memdisk to boot

floppy disk images. Using the memdisk kernel from the
syslinux package (http://syslinux.zytor.com/
memdisk.php), you can load disk images and execute
Sidebar 1: Contents of menu.lst
default=0
timeout=10
root=(hd0,0)
splashimage=/boot/grub/debsplash.xpm.gz
title DSL 1.2 (2.4.26) 1024x768 (save to RAM)
kernel /boot/kernels/dsl-linux24 ramdisk size=100000
init=/etc/init lang=us apm=power-off vga=791 toram
nomce noapic quiet knoppix dir=images knop-
pix name=dsl
initrd=/boot/images/dsl-minirt24.gz
title Debian Sarge Installer
kernel /boot/kernels/di-vmlinuz initrd=/boot/images/di-
initrd.gz ramdisk size=10240 root=/dev/rd/0 de-
vfs=mount,dall rw
initrd /boot/images/di-initrd.gz
title HP nx5000 F0.d BIOS Upgrade
kernel /boot/kernels/memdisk
initrd /boot/images/hpnx5000f0d.img
title Memtest86+ (1.60)
kernel /boot/kernels/memdisk
initrd /boot/images/memtestp.bin
them in a non-emulated environment. How might this
be useful? Let’s say you have an organization with sev-
eral different models of desktops and laptops. You could
create a CD-R/W or a bootable USB flash memory key
with all of the different BIOS upgrades or hardware tests.
Rather than carry around a book of floppies, you can sim-
ply copy the floppy image and boot from the CD-R/W or
USB flash memory key. Using this method, you can also
add Memtest86+’s floppy image to your bootable CD-R/W
or USB flash memory key and have it at your disposal. Here
is an example of a menu.lst snippit using memdisk to boot
into Memtest86+:
title MemTest86+ Ver 1.60
kernel /boot/kernels/memdisk
initrd /boot/images/memtestp.bin
There is nothing special about the filenames. The only im-
portant thing is that the path and name referenced matches
with the actual files. Check out Sidebar 1 for more examples
of disk images.
GRUB with DSLinux
So how can you boot a full GNU/Linux desktop off a USB
flash memory key with GRUB? First, download the DSL
ISO9660 image, and either burn it to a CD, or mount it via
loopback:
# mkdir dsl-test
# mount -t iso9660 -o loop dsl-image.iso dsl-test

30 Free Software Magazine Issue 10, January/February 2006

 HACKER’S CODE
Next, copy the KNOPPIX file, kernel, and initial ramdisk:
# cp dsl-test/KNOPPIX/KNOPPIX \
/media/usb/boot/images/dsl
# cp dsl-test/boot/isolinux/linux24 \
/media/usb/boot/kernels/dsl-linux24
# cp dsl-test/boot/isolinux/minirt24.gz \
/media/usb/boot/images/dsl-minirt24.gz
# sync
# umount dsl-test && rmdir dsl-test
This assumes that your USB flash memory key is mounted
at /media/usb. Next, edit the /media/usb/boot/grub/menu.lst
file and make sure it looks like the entry in Sidebar 1.
You might have noticed that the root line at the top of the
menu.lst file says (hd0,0) even though we used (hd2,0) ear-
lier. When you boot from the USB flash memory key, the
key itself becomes hd0, even before the primary master hard
drive. Once you have the menu.lst file edited, go ahead
and reboot. Make sure that your BIOS is set to USB-HDD,
USB-ZIP, or USB-FLOPPY. You might need to experiment
to see which one works. Once you get to the menu, select
the option for DSL. If you get a GRUB error and are unable
to successfully boot into a kernel, press the ‘c’ key to open
a GRUB prompt. You can try commands like find to help
locate files to boot from.
GRUB splash images
Another cool function with GRUB is putting a splash im-
age on the boot menu screen. By default, GRUB will make
the menu screen a plain black-and-white menu. There are
menu options to change the black and white colors, but why
stop there? Grab your favorite picture, or head to one of the
URLs listed below which have splash images created for
you. If you are creating your own, it will need to be in XPM
format, a maximum color palate of 14 colors, and 640x480
resolution size. The GIMP can help transform your graphic
to these specifications. As an alternative, you can use the
ImageMagick suite of programs. The application convert
can help with this conversion process. You can run it as
follows:
$ convert -resize 640x480 -colors 14 \
mycoolpicture.jpg mybootsplash.xpm
$ gzip mybootsplash.xpm
In this case, the file mycoolpicture.jpg will be resized to
640x480, reduced to 14 colors, and saved in the XPM
Free Software Magazine Issue 10, January/February 2006
graphical format. The second step compresses the XPM file
using the gzip compression method. GRUB can display the
gzipped-xpm splash images well.
If you don’t want to create your own splash image, check
out some of the following web sites which have them avail-
able for download:
• GNU GRUB Public Splashimage Archive
(http://ruslug.rutgers.edu/ mcgrof/ ∼
grub-images/images)
• GRUB Splash (http://vision.featia.net/
linux/grubsplash)
• Once you have the XPM or gzipped-xpm file, you need
to add a line at the top of GRUB’s menu.lst to instruct
GRUB to load the specified boot splash image. The
line should read as follows:
splashimage=/boot/grub/debsplash.xpm.gz
In my case, I’m using the Debian boot splash image as my
background.
---------------------------------------
Another cool function with GRUB is
putting a splash image on the boot menu
screen
---------------------------------------
GRUB on CD-R/W media
I’ve also mentioned several times about installing GRUB to
use for a bootable CD-R/W disk. When generating the ISO
image, you will use some special settings of mkisofs. The
command to build a bootable GRUB CD-R/W looks like:
$ mkdir -p iso/boot/grub
$ cp stage2_eltorito iso/boot/grub
$ mkisofs -R -b boot/grub/stage2_eltorito \
-no-emul-boot -boot-load-size 4 \
-boot-info-table -o grub.iso iso
The key is to copy the stage2 eltorito file into the /boot/grub
directory of the CD image tree, and run mkisofs with the
options specified above. As mentioned earlier, you can also
burn a menu.lst file along with kernels and disk images and
put them all on the CD. In the menu.lst, you will need to use
(cd) as the device, rather than (hd0). Splash images work
well, too.
31
 HACKER’S CODE

Conclusion (http://ruslug.rutgers.edu/∼mcgrof/
grub-images/images)
Congratulations! Now you have a bootable USB flash mem- • GRUB Splash (http://vision.featia.net/
ory key, with DSL and memtest86+, and even a nice boot linux/grubsplash)
splash image. You even have the knowledge to add extra
images, like your favorite BIOS update disk image, and oth-
Copyright information
ers. For more information, check out some of the links in
the resources section below.
c 2005 Jeremy Turner
This article is made available under the “Attribution-
NonCommercial-NoDerivs” Creative Commons License
Resources 2.0 available from http://creativecommons.org/licenses/by-
nc-nd/2.0/ (http://creativecommons.org/
• Grub Homepage (http://www.gnu.org/
licenses/by-nc-nd/2.0/).
software/grub)
• Grub Manual (http://www.gnu.org/
software/grub/manual/grub.html) About the author
• Memdisk (http://syslinux.zytor.com/
Jeremy Turner enjoys freelance writing when given the
memdisk.php)
opportunity. He often plays system administrator, hard-
• Syslinux (http://syslinux.zytor.com)
ware technician, programmer, web designer, and all-
• Memtest86+ (http://www.memtest.org/
around nice guy. You contact him by visiting his web
#downiso)
site (http://linuxwebguy.com).
• GNU GRUB Public Splashimage Archive
Jump to Debian GNU/Linux!
A guide to why the Debian distro is a good choice
Arturo Fernández Montoro
here are hundreds of GNU/Linux distributions
T around, each with its strengths and weaknesses.
One that stands out from the masses is De-
bian. It is the only major distribution not de-
veloped (or even backed) by commercial vendors, but by
a group of volunteers around the world. Its main fea-
tures are robustness, great software package management,
a huge software collection consisting of more than 15,000
pre-compiled packages ready to install and run, and a trans-
parent and always helpful support system based on mailing
lists and a bug tracking system.
But, there is something else that makes Debian special: No
other distribution has seen as many offspring distributions
as Debian has. Among them you’ll find customized Linux
distributions for regional markets like LinEx (a government-
driven project in Spain), or the shooting star Ubuntu (devel-
oped by a commercial vendor). The reason for this popular-
ity is obvious: The Debian distro is not only free, but boosts
flexibility and transparency.
If you use another Linux distribution and you are interested
in changing, after you read this article you should install
Debian for sure.
Introduction
Debian/GNU Linux is a universal operating system. You
can install and run it not only on Intel- and AMD-based 32
and 64 bit PC systems, but also on different computer archi-
tectures like Compaq’s and Digital’s Alpha systems, ARM,
Free Software Magazine Issue 10, January/February 2006
Motorola 680x0 processors (m68k), SGI’s big-endian MIPS
systems and Digital’s DECstations, Sun’s SPARC and Ul-
traSPARC systems, the PowerPC (using IBM and Motorola
processors), IBM S/390 mainframe systems and Hewlett
Packard’s PA-RISC machines (hppa).
The Debian project doesn’t only produce a free (as in free-
dom, not only of charge) distribution, but is itself a strong
supporter of free software. While many people spend
hours discussing the differences between “free software”
and “open source software”, Debian explicitely explains its
position in two documents: “The Debian Free Software
Guidelines” (DFSG), which defines what constitutes free
software according to the Debian project; and the “Social
Contract” with the free software community which posi-
tions the project itself and defines its links to the outside
world:
• Debian will remain 100% free
• The project will give back to the free software commu-
nity
• The project will not hide problems
• It prioritizes its users and free software
• It describes how to deal with work that does not meet
its free software standards
You can read this document at Debian Social Contract
(http://www.debian.org/social contract).
Differing from other Linux distributions, you’ll find not just
one Debian version at a time, but three different releases:
“stable”, “testing” and “unstable”. These are three different
33
 HACKER’S CODE

distributions. Each one ships with its own software pack- /etc/apt/sources.list configuration
ages which may or may not stem from the same original
source-code version. For example: Debian “stable” at the
time of writing shipped with a gimp package tagged 2.2.6,
while “testing” at the same time included version 2.2.7. The
“unstable” release uses the most recent software versions.
There is, however, only one official release: “stable”. De-
bian recommends it for production environments. The “test-
ing” distribution contains packages that haven’t been ac-
cepted for the “stable” release yet, but after extensive testing
will eventually move over. The “unstable” tree is the De-
bian developers’ working ground. At times this distribution
can show problems like broken dependencies. Neverthe-
less, this distribution is usually completely functional since
quality assurance (QA) is a task the Debian project takes • Non-Free: consists of software with a DFSG-
serious. Packages in “unstable”, however, simply have not incompatible license like “doom-wad-shareware”, a
been tested in depth. If you wish to run a system equipped package that includes shareware game files for the 3D
with the latest software versions “testing” is a good bet, but game DOOM
if you need a robust server you should choose “stable”.
---------------------------------------
---------------------------------------
The codename of the most recent
The reason for this popularity is obvious:
“stable” distribution is “Sarge” also
The Debian distro is not only free, but
known as Debian GNU/Linux 3.1
boosts flexibility and transparency
---------------------------------------
---------------------------------------

Each of these distributions has a codename which (apart
from “unstable”) changes with every release. The code-
name of the most recent “stable” distribution is “Sarge”,
also known as Debian GNU/Linux 3.1. It was released June,
6th, 2005. The current “testing” distribution is nicknamed
“Etch”, and “unstable” always remains “Sid”. All code-
names are taken from Pixar’s movie “Toy Story” since a
Debian project leader worked for this company. Obsolete
releases are “Woody” (3.0), “Potato” (2.2), “Slink” (2.1)
and “Hamm” (2.0).
Each distribution groups its packages by their software li-
censes:
• Main: includes all software compatible with the
DFSG, for example the GNOME web browser
“epiphany”
• Contrib: you’ll find free software that depends on non-
free (according to Debian) software to run. “ant” (a
Java development tool) is an example for this
Advantages and disadvantages on the
technical level
Debian’s different approach becomes visible not only for-
mally but also in technical details. It ships with a unique
and robust package management system, centering around
the APT tools and the “dpkg” utility, that Debian develop-
ers and users are especially proud of. It is the best way to
install software quickly and easily on your machine—even
as a newbie you’ll appreciate and love it. The package man-
agement system uses “dependencies” between packages to
ensure correct software installation. Pre-compiled packages
are distributed in a specific archive format with “.deb” file
extension. While utilities like “dpkg” and “apt-get” are pure
command-line tools, Debian also provides a set of package
management front-ends to choose from like “dselect”, “ap-
titude” and “synaptic”. Most tools access software reposi-
tories via FTP or HTTP, provided the user wishes so. You
type a single command, and Debian will download, install
and configure the software for you. You don’t need to worry

34 Free Software Magazine Issue 10, January/February 2006

 HACKER’S CODE
about where the software resides—as long as the configura-
tion file /etc/apt/sources.list contains the repository’s proper
URL. Each repository entry looks like the following line:
deb http://www.debian.org/debian \
sarge main contrib non-free
This means that the tools will download and install software
belonging to the “Sarge” distribution from the main Debian
web site. To install the GNU Image Manipulation Program
(GIMP) the “root” user will now type e.g. the following
command line:
#apt-get install gimp
The “apt-get” program will download the GIMP package
and all packages it depends on (i.e. all software needed to
run the GIMP).
Debian does not hide errors and bugs. Users can to send
bugs using the bug tracking system and the Debian devel-
opers can quickly access them by web or e-mail. Bugs
are accessible to everybody because of the importance De-
bian places on Quality Assurance. “The Debian Policy” is a
specification for the standards of quality using by Debian.
On the other hand, Debian also has some disadvantages
when compared with other distributions: its hardware auto
detection is still way behind the appropriate functionality
in SuSE or Mandriva, and the installation process doesn’t
make it easy for beginners.
Installing Debian/GNU Linux
To install Debian you need at least a little patience. To give
you a headstart, a short summary of the Debian “Sarge” in-
stallation process for a desktop machine follows. The offi-
cial Debian installation page at
Debian installation page (http://www.debian.org/
releases/stable/installmanual.en.html)
provides more detailed information.
First you have to choose an install method:
• Network: you can install Debian via the internet or us-
ing a partition that one of the other machines in your
LAN provides by means of NFS (Network File Sys-
tem)
Free Software Magazine Issue 10, January/February 2006
• CDs or DVDs: books, magazines and independent
software vendors will provide you with Debian instal-
lation media. You can also download the appropriate
images from the Debian web-site and burn them your-
self. To boot from the installation DVD is the easiest
way for newbies. Check your BIOS to ensure your sys-
tem will do so!
Before you start make sure you have as much information
about your hardware as possible, because during the install
process the installation program will request the informa-
tion.
“Sarge” allows you to use the 2.4.17 or the 2.6.8 kernel.
To choose the latest Linux version, type “linux26” at the
“lilo” prompt when booting from the install media and press
“Enter”.
The Debian installer program asks you about your language,
country and keyboard layout. If possible it configures your
network and loads appropriate kernel modules for the hard-
ware it auto-detects.
---------------------------------------
“Sarge” allows you to use the 2.4.17 or
the 2.6.8 kernel
---------------------------------------
Now it’s time to partition your disk. Be careful. After you
have selected the installation partition the next step is to in-
stall the base system. To be able to boot your system you
need to install a boot loader like GRUB or LILO. When
done, reboot your machine. Now Debian starts some post-
boot configuration routines: you’ll be asked to:
• Configure your time zone
• Choose a name for your machine
• Set up users and passwords. During this stage you must
choose the password for the “root” user
• Create an ordinary user for your day work
• Set up PPP or PPPOE for dial-up connections with the
internet
• Configure the APT system in charge of the package
management
• Install some software packages. To do this you first
have to choose the relevant software. On a desktop
machine you’ll probably want the office-suite Open-
Office, the Mozilla Firefox web browser, the e-mail
35
 HACKER’S CODE

Debian official web site Conclusion

Debian GNU/Linux as a distro is different due to its tech-

client Kmail, the instant messenger client Gaim, the
image manipulation program GIMP, Totem (a video
player) and a desktop environment like GNOME or
KDE
• Configure your Mail Transport Agent. This is optional
but useful since internal system notification depends
on e-mail
nical details and for its philosophy. Freedom is a very
important concept for the Debian Project and it is a non-
commercial distro.
Due to its robust, flexible and highly configurable nature,
some governments have chosen Debian upon which to base
the development their own Linux distro and the project has
been a success.
If you use other Linux distributions, I recommend you give
Debian a try. Debian is not only for geeks—everybody can
install and use it. Don’t doubt it, jump to Debian!
---------------------------------------
Due to its robust, flexible and highly
configurable nature, some governments
have chosen Debian upon which to base
the development their own Linux distro
and the project has been a success
---------------------------------------
Acknowledgements

After these steps you’ll be presented with the login prompt. The author would like to thank Patricia Jung for her com-
Your system is now ready to use. ments and grammatical revision.

International example
Linex web site
Back in 1998, the government of Extremadura, a small re-
gion of Spain, launched an educational project for tech-
nological alphabetization. The success of this educational
project depended totally on the chosen software, and access
to source code was a very important issue. The govern-
ment developed a new Linux distribution based on Debian
GNU/Linux 2.0 called “Linex” (Linux + Extremadura).
Currently, Linex has an installation base of over 10,000
computers, mainly in government offices and schools.
Another region of Spain (Andalucı́a) follows the exam-
ple and makes their own distribution called “GuadaLinex”
based on Linex.
Countries like China, Italy and Brazil have been studying
the success of the project.

36 Free Software Magazine Issue 10, January/February 2006

 GuadaLinex web site

Bibliography

The Debian web site (http://www.debian.org)

The Linex distro web site (http://www.linex.org)

The GuadaLinex distro web site (http://www.

guadalinex.org)

Copyright information

c 2005 Arturo Fernández Montoro

Permission is granted to copy, distribute and/or modify this
document under the terms of the GNU Free Documentation
License, Version 1.2 or any later version published by the
Free Software Foundation; with no Invariant Sections, no
Front-Cover Texts, and no Back-Cover Texts. A copy of the
license is available at http://www.gnu.org/copyleft/fdl.html

About the author

Arturo is a software engineer specialist in web devel-

opment and a freelance author for various Linux maga-
zines. He works with Debian GNU/Linux since 2000.
Convincing management to
approve free software
Tips to better advocacy
Maria Winslow
he grassroots efforts of system administrators
T have brought Linux and other free software into
the mainstream. To be an effective advocate
for free software at work, you need to speak
the language of management and convince them from their
point of view. This article discusses how to present your
case, why your audience makes all the difference, how to
hook them with proof of cost savings, and reveals two se-
cret weapons for your quest to promote free software.
This article explains why bashing Microsoft won’t help you
in your case, which migration recommendations will seem
the most practical and feasible to management, and how to
present those recommendations in terms that management
will respond to. It talks about how your goals are differ-
ent from those of management, and how to adjust your ap-
proach accordingly. Finally, it demystifies return on invest-
ment (ROI) and shows how you can put together simple cal-
culations to back up your case.
The golden rule of advocacy
The most important thing to remember when you are mak-
ing a case for free software is to be sensitive to your au-
dience. Who are you talking to? How technical are they?
What is most important to them? You must address their
concerns. Most people consider it a courtesy if you tell them
that you’re researching different free software solutions and
you’d like to take their concerns into account as you inves-
tigate the options.
Free Software Magazine Issue 10, January/February 2006
Tip 1: don’t bash Microsoft
Criticizing Microsoft can be counter-productive in your ad-
vocacy for free software. There are three important reasons
for this.
The first and best reason is that you really don’t have to.
Jupiter Research conducted a survey in 2003 of IT profes-
sionals in the small- to mid-size business market, asking
what free software they were using and why. The number
one reason for migrating to Linux wasn’t cost savings, secu-
rity, or reliability. Perhaps surprisingly, the primary reason
stated for migrating to free software was to get away from
Microsoft. Therefore, you don’t have to say anything nega-
tive; your audience may be thinking it anyway.
The second reason involves a little pop psychology. Con-
sider this: when you denigrate Microsoft, you insult people
who only have experience working with Microsoft. Either
your audience or at least a percentage of their reports may
be in this camp, so be careful of insulting the base of their
professional knowledge.
The third reason is that you will seem more credible and
professional if you stick to the positive and resist the urge to
use a derogatory nickname.
Tip 2: be practical
Focus on recommendations that will seem the most practical
and feasible to your audience. Your campaign should begin
on the edges of the enterprise, with systems that are less
39
 MIND SET
critical. It’s best to start with migration recommendations
for systems that are less likely to cause unacceptable dis-
ruption. Remember to consider management’s perception
of disruption as you consider potential candidates. For ex-
ample, e-mail systems can be a difficult place to start unless
you have a compelling reason to promote a migration in this
area. Management will naturally shy away from migrations
perceived to have a high risk of disruption.
Basic internet infrastructure is usually the best place to start.
Basic web servers, DNS servers, web content, spam/virus
filters, and similar services are commonly trusted to Linux
and free software. Sometimes it’s possible to get approval
for a Linux deployment on reused hardware. So don’t over-
look this possibility, especially for file servers. It’s easier for
management to say yes when the capital outlay is so low.
In your advocacy efforts, be sure to stress to management
that free software adoption doesn’t have to be radical; just a
few systems in non-critical roles can add up to savings now
and increased staff experience for future deployments.
Tip 3: lose the tech-speak
Sometimes it might seem that management doesn’t take
your technical advice seriously. Understand that your goals
are different from management’s goals. You want to work
with the best technology, advance your career by keeping
up with new technologies as they develop, and do your job
efficiently using tools that will make your job easier. You’re
technology-centric. Management wants to hold down the
fort, make sure nothing goes terribly wrong, try to keep peo-
ple from complaining, and do it all within the budget. They
are not technology-centric. Note that the actual technology
doesn’t make it into the list of most important goals—it’s
just a means.
The significance is that when you focus on the technology,
management thinks you don’t understand the big picture. To
management, it’s about making things work well enough to
keep everyone reasonably happy within the constraints of a
budget. If your pitch to management focuses primarily on
the superiority of the technology, you’re only giving them a
small part of the story from their point of view. Even worse,
if you argue about the technology with a colleague in front
of management, you’re likely to get even more pushback.
For example, you may recommend dumping the mail server
for Sendmail on Linux. Your colleague may advocate Post-
40 Free Software Magazine Issue 10, January/February 2006
fix because it’s more secure. When you enter a discussion
on the technical merits of each solution, your boss may tune
out the tech-speak while perceiving that either solution may
pose a risk by being the wrong choice. By focusing on the
technical, and especially by debating technical points, you
and your colleague introduced more risk (or at least percep-
tion of risk) into the equation. So lay off the tech-speak and
focus on your two secret weapons instead.
Secret weapon 1: case studies
Management loves case studies. They are anecdotal, easy
and quick to read, and easy to identify with. Most impor-
tantly, case studies feel like proof to management. Case
studies usually reflect their peers—IT directors, mainly, and
an endorsement from a peer goes a long way. So start
googling.
As you search for case studies to help you in your recom-
mendations, follow these two basic rules:
• Avoid case studies provided by vendors, as they usu-
ally sound too much like marketing materials to be
truly credible. Your best source is always going to be a
magazine because the analysis will be independent.
• Choose case studies that are as similar to your scenario
as possible to help management recognize how your
free software recommendation would work in your sit-
uation as well as it did in the case study. Consider
whether the case study is reflective of your industry
segment, or if the basic IT landscape is familiar.
To get started, look at (http://www.
windows-linux.com/articles) a few of the case
studies I presented in my book The Practical Manager’s
Guide to Open Source (http://windows-linux.
com/practicalOpenSource).
Secret weapon 2: show them the money
The single most effective way to influence any decision to
deploy free software is to prepare a cost justification. When
you hand your boss a spreadsheet showing the cost savings
gained, they will consider your migration recommendation
more seriously. Fortunately, this doesn’t have to be compli-
cated.
 MIND SET
The Practical Manager’s Guide to Open Source
Return on investment (ROI)
Return on investment (ROI), also called the cost-benefit ra-
tio, is the benefits gained divided by the costs to acquire the
investment. You can think of ROI as “for every dollar spend,
I get back X percent”.
Benefit = the total amount of savings (proprietary
price—free software price)
ROI = benefit / cost
To calculate the ROI of a particular migration, you need a
few cost numbers. Whenever possible, use historical bud-
getary data for the best accuracy. Not everyone pays the
same for software, so if you use the actual value your orga-
nization pays, your calculation will be more credible. You
need to find out two figures: the cost of the status quo re-
placement system and the cost of deploying your free soft-
ware alternative. Calculate the cost of the status quo sys-
tem as everything your organization will pay if you don’t
go with free software. The cost of the free software deploy-
ment should not be $0. It’s neither believable nor true. Be
sure to include support contracts from a third party and out-
sourcing costs for deployment or staff training, even if you
pay nothing for the actual software.
Why ROI and not TCO?
TCO stands for Total Cost of Ownership, which measures
the cost of the system over its entire life. So why should you
calculate ROI and not TCO? There is one simple reason to
use ROI—it is easy to calculate now. ROI involves the costs
and benefits of the acquisition of a system, which are known
at the time of deployment. By its very nature, TCO involves
guessing, simply because you can’t know the total cost of
the system until the end of the cycle. TCO also doesn’t
Free Software Magazine Issue 10, January/February 2006
show the benefit gained by the organization, just the total
cost over the life of the system. The TCO of paper and
pencil is lower than desktop computers, but clearly most
organizations employ desktop computers.
An ROI example
Take as an example a Windows to Linux file server mi-
gration. You’re making the recommendation because the
server, accessed by 100 users, is now five years old, and
you’re coming up on a scheduled hardware/software up-
grade. This is the perfect time to make a migration rec-
ommendation.
To calculate the ROI of this recommendation, you need to
know:
1. What is the cost to upgrade the hardware and software,
if we remain status quo?
2. What is the cost to migrate to Linux and Samba?
Because eWeek Labs found Linux/Samba to be two-and-a-
half times faster at file serving than Windows 2003 Server,
you’re also recommending that you reuse the hardware.
Once you have the pricing, you can find the benefit to the
organization of migrating (status quo cost - free software
cost). Then divide the benefit by the cost to arrive at the
ROI.
Status quo upgrade:
• Windows 2003 Server, Enterprise Edition, includes 25
client access licenses ($3999)
• 75 additional CALs ($2448.75)
• New server hardware ($2000)
Total cost of a status quo upgrade: $8447.75
Linux replacement:
• Debian GNU/Linux with Samba ($0)
• Initial installation support contract with local system
integrator ($1500)
• Reuse server hardware ($0)
Total cost of a Linux/Samba replacement: $1500
Total benefit: 8447.75 - 1500 = $6947.75
ROI = total benefit (6947.75) / cost of free software deploy-
ment (1500) = 463%
41
 MIND SET

This example shows that a simple migration can bring a
very healthy ROI, which will help convince management
that free software is a good idea for your IT environment.
Use this example as a base for your calculations, and be
sure to get the most accurate cost information possible.
Copyright information
This article is made available under the “Attribution-
Share-alike” Creative Commons License 2.0 available from
http://creativecommons.org/licenses/by-sa/2.0/.

Making your case

About the author
Remember, the golden rule of advocacy is to know your
As an Open Source Practice Leader with Virtuas,
audience and be prepared to address their concerns. Tech-
Winslow assists clients in understanding the techni-
nology matters less to management than it does to you, so
cal and budgetary impact free software will have on
make an effort to tone down the bombardment of informa-
their computing environments. Her recent book, “The
tion and give them just the highlights. Recommend migra-
Practical Manager’s Guide to Open Source”, guides IT
tions that will seem the most feasible and least risky to man-
directors and system administrators through the pro-
agement. Be professional and positive in your recommen-
cess of finding practical uses for free software that
dations. Search for case studies that are relevant to your
will integrate seamlessly into existing infrastructures,
situation—they are one of your secret weapons. Addition-
as well as understanding the costs and savings. She
ally, it is worth taking the time to put together a simple ROI
is a frequent speaker and author on the topic of free
calculation. Using the example in this article as a base, you
software. She can be reached via the Practical Open
can impress management with your well-rounded view of
Source website (http://windows-linux.com/
free software in the enterprise and meet with better success
about.html).
in your internal advocacy.
A techno-revolutionary trip on
the internet
Reflections on the lessons from Dean for America
Tom Chance
hen I think about American presidential
W elections, three things come to mind:
money, corporate power and disenfran-
chisement. One of the big political sto-
ries of our time is the decline of party politics, especially
for the young. But another story is that of the internet revi-
talising democracy, empowering and connecting citizens in
a new, vibrant space. Often Utopian, theoretical and roman-
ticised, this vision of the future was made real in the race for
the Democratic presidential candidacy recently in America
by Howard Dean. With a campaign team numbering in the
hundreds of thousands mobilising over the internet, Dean
went from being a no-hoper to pole position in a matter of
months.
The campaign manager credited with much of this success,
Joe Trippi, wrote a book recounting the experience and his
lessons for America, under the provocative title: The rev-
olution will not be televised: democracy, the internet and
the overthrow of everything. Trippi’s big idea is that the in-
ternet is going to totally change American politics, in part
returning it to a golden era of localised engagement while
also propelling it into a bright new future of decentralised
participation. In this article, I reflect on some of the wider
issues he tackles and try to understand the implications for
representative democracies and their established organisa-
tions. Because of the enormity of the subject area, I’m sim-
ply going to step past the many activities happening outside
the mainstream.
First, I should recount some of the facts of his campaign.
Free Software Magazine Issue 10, January/February 2006
They started with very little money, a tiny campaign team
and a candidate with no real chance of winning. When hired
in as a new campaign manager because of his decades of ex-
perience, Trippi immediately started to make some changes.
His first was to leverage a web site called Meetup, a so-
cial networking site where people registered their location
and an interest in meeting about some subject—in this case
Howard Dean. The web site would suggest a time and loca-
tion, and so people began to meet and discuss ways to help
the Dean campaign. The more Trippi’s team ceded control
to volunteers on the internet, the more the campaign became
successful, to the point where he was the front runner for a
short period of time. He was eventually brought down, iron-
ically, by the mainstream media.
Breaking down the broadcast media
To understand Trippi’s book, and many other arguments
to do with the internet, you need to understand the differ-
ence between broadcast and multicast media. Television,
radio, newspapers and other “traditional” media broadcast
information at the consumer. Any opportunity to feed your
ideas back to the media, and to have others hear your views,
tend to be trivial. Examples include a “points of view” pro-
gramme, a “letters to the editor” page or live comments sent
in from mobile phones. The internet, Trippi suggests, is fun-
damentally different because it is multicast, which is to say
that information comes from many outlets and you are free
to add your own. Blogs covering politics, for example, will
43
 MIND SET
Are we in a dictatorial democracy? “America”, by ?
LoVe
c
?
c other aspects of politics, the power to do this depends upon
Released under the Creative Commons Attribution license
necessarily provide a far wider range of views and generally
allow you to comment directly on articles, start debates or
even start up your own web site or blog to counter opinions
you disagree with.
Trippi suggests that the broadcast media have subverted
democracy and that the internet will “overthrow every-
thing”. That’s a little hyperbolic. Rather, as the centralised
corporate broadcast media have taken an increasingly cen-
tral role in democratic discourse—telling us what the po-
litical parties think, deciding what constitutes news, and so
on—that discourse has been subverted. We may still dis-
cuss politics around the dinner table, at the pub or at the
workplace, but our relationship with politicians is marked
by our consuming their ideas and the media’s interpretation.
We are in a sense no longer citizens unless we get involved
with certain mainstream political organisations, which often
provide little scope for genuine participation and grassroots
influence.
Worse still, because politicians and their parties tend to only
get short media spots with which to communicate their mes-
sage, they become simplified. In the US elections political
adverts tend to be no longer than thirty seconds. How can
a candidate possibly convey their complex programme of
policies and opinions in that time frame? The answer is that
they can’t, and so instead the choose to emote, to whip up
fear or anger, and to vilify the opposition. As with many
44 Free Software Magazine Issue 10, January/February 2006
money; good adverts take a lot of funding; “opposition re-
search folders” needed for attack ads take additional time
and expertise. So parties and campaigns become ever more
geared towards winning the allegiance of rich donors, leav-
ing the people behind.
Trippi appeals to this dystopian vision to emphasise the
power of his “peoples’ campaign”, but it is (perhaps inten-
tionally) misleading. People simply aren’t this apathetic and
apolitical, their political interests just tend to manifest them-
selves outside of the mainstream, where the media lens fails
to reach. The broadcasted message is that local politics,
such as a campaign to save a well-loved building, is fun-
damentally different to national politics, and that both are
declining.
---------------------------------------
Broadcast media turn citizens into
consumers, while the internet can help
reassert their role in the democratic
process
---------------------------------------
Into this valley of death strode Trippi and his cohort of in-
ternet entrepreneurs. While his writing is a little dramatic,
Trippi’s team was truly innovative. Though the Dean for
America campaign still ran the traditional media ads, they
saw an opportunity to use the internet to break this lock,
both for the good of democracy and their campaign. The
big idea is that web sites are cheap to set-up and can be
opened up in any number of ways to encourage participa-
tion. Trippi’s team blogged about their work, read feedback
and responded to it; they encouraged activists to set-up lo-
cal meetings via the innovative social networking web site,
Meetup; they showcased others’ work, and promoted efforts
that went far beyond their own designs. They turned the
logic of politics on its head, from a one-to-many relation-
ship in which a politician leads the masses, into a many-to-
many network of citizens engaging in issues and choosing a
politician to represent and lead this discourse.
Of course, the internet is still open to the abuse of the broad-
cast media because it can replicate the broadcast method-
ology. Consider that when Xerox first designed graphical
user interfaces, they made them resemble a secretary’s of-
fice, with folders, documents, a waste bin and so on. Rather
than working on an innovative interface that would change
 MIND SET
the way we work, they chose to adapt new technology to
old methodologies. Similarly, most politicians use web sites
to simply broadcast their message, replicating the method-
ologies of television campaigning on the internet. Indeed,
Dean’s campaign team was reluctant even to link to the
Meetup web site from their own until Trippi convinced them
otherwise; when they did the numbers on Meetup shot up.
The obvious point here is that it is not the technology alone,
the internet and web sites, that will revitalise democracy.
Their decision introduced a new dynamic to their campaign,
one that made it ever more democratic and open. As they
became increasingly dependent upon participation for mo-
mentum, so they were forced to spend more time and ef-
fort working in that area. From the moment the Meetup
link went live on the official web site, Trippi says, the peo-
ple took over the campaign. First thousands, then tens of
thousands, then hundreds of thousands of people were reg-
istering to meet up; over 180 campus groups were started;
millions of dollars were raised not through pandering to the
interests of the rich but by appealing to the civic spirit of
Dean’s supporters. At its height “the discussion was the
campaign”, and the hundreds of thousands of volunteers
were dwarfing the brainpower and efforts of the thirty-odd
official campaign team.
Transforming the internet and the dinosaurs
One of Dean’s slogans was: “you have the power”, reflect-
ing his belief in a new kind of politics. Like many dis-
illusioned with the bureaucratic modern political parties,
the Dean campaign wanted to get away from transactional
politics, where politicians negotiate deals, to transforma-
tional politics, where politicians empower citizens to work
together towards a common cause. The internet provided
them with the tools to mobilise on a national scale, where
previously grassroots politics had generally happened on a
far more local scale. The status of a presidential campaign,
and the fact that Dean’s message was so different to the
standard Democratic party line, saw people flocking to his
cause.
Young people were a big surprise. Called “deanie babies”,
they pushed the grassroots campaign forward more than
any other demographic, with some even travelling vast dis-
tances across the country to sleep on the office floor and
work 17 hour days. Trippi hadn’t seen anything like it since
Free Software Magazine Issue 10, January/February 2006
Bobby Kennedy in the 1960s. While some like to roman-
ticise the internet as an underground that is undermining
the dinosaurs of politics, what was significant about Dean’s
campaign is that it was relevant to the kids. They tapped
a mainstream organisation into an apparently apathetic de-
mographic and found that the kids wanted to be engaged,
but that they were alienated by middle aged baby boomers
who had obviously forgotten their youth. So obsessed with
their “apathetic youth” frame were these politicians and me-
dia moguls that they began to ridicule the deanie babies
as extremist left-wing vegans—an unfair characterisation
by Trippi’s account—when they should have been praising
them for their involvement.
Civic involvement spread beyond the confines of Dean’s bid
for the Democrat nomination. Bloggers and Meetup afi-
cionados organised “Dean Corps”, which Trippi describes
as “a sort of low-impact, weekend Peace Corps. . . they got
together in neighbourhoods to clean up riverbanks, to read
to children, and to collect food for homeless people”. It’s
not as though Trippi’s team used the internet to invent the
notion of civic community participation; they didn’t even
come up with the idea of combining the two. Given the tools
to organise them, and the profile to attract enough attention,
people took it upon themselves to organise and participate in
all kinds of community activities that would have the likes
of Margaret “there’s no such thing as society” Thatcher run-
ning for cover.
Dean’s campaign was, in American campaigning jargon,
“insurgent” because he started with apparently no hope of
success. His team took his party and his opponents, lum-
bering dinosaurs in comparison, by surprise. The campaign
may well be the first ripple in a series of waves that wash
away levels of bureaucracy and corruption in representative
democracies. At least, that’s what Trippi would have us be-
lieve. At the very least we can place this series of events in
a wider context to see how change is happening. For this we
can look on the other side of the Atlantic to Britain.
The British Broadcasting Corporation (BBC), one of the
largest and most respected in its field, and also one of
the oldest organisations in broadcasting, is embracing open
technology and content like no other comparable organisa-
tion. They have been opening up their web sites with RSS,
XML and APIs; they are starting to release their archives
under a Creative Commons-based license; they have even
been developing their own video compression and transmis-
45
 MIND SET
sion technology in a series of new free software projects,
rather than doing it all in-house and “protecting” their work
with patents and proprietary software licenses. Their strat-
egy is not only opening the corporation up technologically
but also socially, encouraging civic participation in the pub-
lic institution whether through their developers’ network,
their web-based collaborative projects or the imperative to
remix implicit in their Creative Archive.
But you would be wrong to assume that some top executives
in the BBC have decided and decreed that the BBC must
embrace the digital era. Rather, small groups of visionaries
are working towards that vision in their department, and the
cumulative effect is that the lumbering dinosaur of public
service broadcasting is in fact evolving faster than its com-
mercial competitors. Those visionaries made some ripples
that have turned into a wave, just as Dean’s team may have
done in politics. The core message of Trippi’s book could
almost read like a guide book to corporations wanting to
adopt the free software community’s approach. Free soft-
ware hackers used free software as the basis for the Dean
campaign web site. Slowly these interests are coming to-
gether—free & open technology, participatory politics and
the mainstream—to change our world.
Will demonstrations become a thing of the past thanks to the inter-
net? Photo by dustpuppy. Released under the Creative Commons
Attribution license
46 Free Software Magazine Issue 10, January/February 2006
Trippi’s seven rules for internet activism
1. Be first
2. Keep it moving
3. Use an authentic voice
4. Tell the truth
5. Build a community
6. Cede control
7. Believe again
Are we heading towards an anarchistic
utopia?
If you’re like me, you would have put down Trippi’s book
having finished it thinking: “the internet is the first technol-
ogy that truly gives people full access to knowledge. . . we
can accomplish anything . . . [I] have the power!” That’s a
selection of the claims Trippi makes in the last page. Part
of the problem with his book, and that of many writers in
the field, is that he is an unabashed technophile. Because
his purpose in the book is to sell his Big Idea and to enthuse
us with his success, he doesn’t spend much time looking
at where the internet might be bad for democracy, nor the
extent to which it is the internet specifically, and not other
factors, that led to his success. Of course he’s not so stupid
as to suggest that it is the internet alone—his emphasis on
his team’s strategy makes that clear—but he is still uncriti-
cal in his appraisal.
This is not new. According to Douglas Kellner:
“. . . film, for instance, was celebrated by some
of its early theorists as providing new docu-
mentary depiction of reality, even redemption of
reality, generating a challenging art form and
novel modes of mass education and entertain-
ment. But film was also demonized from the be-
ginning for promoting sexual promiscuity, juve-
nile delinquency and crime, violence, and copi-
ous other forms of immorality” (Douglas Kellner,
New Technologies and Alienation: Some Criti-
cal Reflections (http://www.gseis.ucla.
edu/faculty/kellner))
Trippi is rightly critical of the mainstream broadcast media
for the mediocrity of much of its content and for the very
 MIND SET
Hopefully internet campaigning will make balloons unnecessary.
Photo by Tom Chance. Released under the Creative Commons At-
tribution license
nature of its transmission. But while television can be un-
critical, biased and unengaging, so can the internet. Many
blogs are every bit as bad as the worst of the broadcast me-
dia, and an active minority can gain disproportionate power
on the internet just as easily as through the broadcast me-
dia, given the right infrastructure. Almost 2500 years ago,
Plato warned of the powerful speaker, the dictatorship of
orators, who can whip up a frenzy amongst communities.
Would Trippi have been so rapt if an ultra-conservative can-
didate had done the same, bringing hundreds of thousands
of people forward to campaign against everything he thinks
is important, or would he have engaged more critically with
the issues he has raised?
So let’s summarise some of these issues quickly. The inter-
net can foster a sense of community, of civic participation
and belonging to a mainstream political party. It provides a
significant source of fundraising, though generally only for
c 2005 Tom Chance
candidates in the main political parties. But television still
reaches more voters, it broke Dean’s candidacy and those
who benefit from the power imbalances it creates are un-
likely to simply give up their position when the internet be-
comes the main mode of communication. Of course the in-
ternet, both in terms of the technology and the social and
political practices associated with it, are evolving.
What will this future look like? We can find some an-
swers by looking at the cutting edge in mainstream poli-
tics today. Already, some politicians in the UK use blogs
that allow comments, and actually respond to comments
posted. The people behind the Dean campaign software
have been working on CivicSpace, “an integrated and ex-
Free Software Magazine Issue 10, January/February 2006
tensible platform of online organising tools for all man-
ner and size of organisations”. Though no political party
in any election that I am aware of has replicated Dean’s
successful use of the internet, it must surely be a matter
of time. A promising development in Europe is Greensnet
(http://www.greensnet.org), which will try to link
together all European green organisations, parties and indi-
viduals in a grassroots activism network using similar soft-
ware to CivicSpace.
To answer my own question, the future is very unlikely to be
an anarchistic Utopia any time soon. So long as mainstream
organisations adapt to the internet, the political spaces on
the internet will probably adapt to their presence; if the par-
ties can really change the way they interact with people on-
line, then people previously lost to the world of mainstream
politics will reintegrate the parties into their lives. Chal-
lenges from outside these norms, from organisations and
collectives using the internet for radical politics, will con-
tinue to shape those political spaces but so long as the di-
nosaurs evolve they won’t replace them. The challenge for
all of us who are both technically minded and politically
aware is to ensure that the internet continues to be a tool
for empowerment rather than a tool for control, to push the
mainstream in the same direction that Trippi took Dean’s
wing of the Democratic party. We won’t see a technorev-
olution, but we might just form a revolt that will change
politics forever.
Copyright information
This article is made available under the “Attribution-
Share-alike” Creative Commons License 2.0 available from
http://creativecommons.org/licenses/by-sa/2.0/.
About the author
Tom Chance is a philosophy student, free software ad-
vocate and writer. He is the Project Lead of Remix
Reading (http://www.remixreading.org), the
UK’s first localised Creative Commons project. You
can contact him via his web site (http://tom.
acrewoods.net).
47
64 Studio

Building a native 64-bit creative distribution

Daniel James

reative computer applications are a niche,

C
Not much choice here. . .
and a relatively small one at that. Even
brand-leading proprietary software companies
like Steinberg, the developers of the long-
established Cubase music sequencer, have been recently
bought out. Consolidation in the creative application mar-
ket has seen Adobe buy Syntrillium, who created Cool Edit,
Avid buy Digidesign and Apple buy Logic—and there are
plenty of other examples. What this means is that a hand-
ful of multinational companies could now effectively mo-
nopolise the gateway to creative expression, at least as far
as computers are concerned. This might not be an issue if
it were not for the wide proliferation of powerful, general
purpose computer hardware in the first world.
In addition, the internet, and by extension the personal com- will take. What is clear is that there is less real choice in
puter, are now the principal channel for distribution of cre- “industry standard” creative software than ever.
ative works in many fields. Proprietary tools on the creative The retail cost of just the software listed here for a single
desktop mean proprietary formats will dominate the inter- computer runs into thousands of euros. The high cost of
net, now that it is no longer a purely textual medium. The entry to this field helps maintain the artificial distinction be-
landscape today for “industry standard” creative software tween consumers and producers. Now of course the soft-
on the proprietary platforms looks a bit like that in the fol- ware companies who make these programs would say that
lowing figure. they represent great value for money, and that educational
In case you’ve heard Apple described as the creative alter- institutions can benefit from generous discounts. On the first
native or competitor to Microsoft, we should not forget that point, it would be fair to mention that as mass access to per-
back in 1997, before the launch of OS X, Microsoft actually sonal computers has brought the cost of hardware down, the
invested $150 million dollars in Apple. That’s not some- creative applications haven’t followed suit. Quark Xpress
thing a company usually does with a competitor. More re- was really expensive in 1995 and it still costs a lot today.
cently, Adobe and Macromedia have announced a partner- To answer the second point, I believe proprietary software
ship, although it’s not yet clear what form that partnership in education, even with discounts, represents a hidden sub-

48 Free Software Magazine Issue 10, January/February 2006

 MIND SET

sidy from the state to the manufacturers of this software. If
we look at one Adobe educational pack on offer in the UK
for example, it appears very reasonably priced at £9.99 per
seat. But a school or college needs to buy 200 seats to get
that price, i.e. pay nearly two thousand pounds, in order to
run feature-limited “Elements” versions of just two applica-
tions. What this means is that taxpayers around the world
are sending millions of dollars every year to fund propri-
etary software development, and yet many schools and col-
leges still can’t afford to put a full set of creative tools in
front of each student.
Even if the software was practically free, consider the value
to a company like Adobe that every graduate of the creative
disciplines around the world leaves college with training in
their software—which the company doesn’t contribute to di-
rectly at all. Even the potential funding benefits of corporate
taxation for education are offset by the fact that most, if not
all multinational software companies use offshore account-
ing.
It’s certainly the case that artists in a number of disciplines
are using GNU/Linux and other free software—free in both
the economic sense and the political sense—to realise their
ambitions. It would also be fair to say that this software is
only just starting to penetrate the consciousness of the main-
stream. But how can we make it practical for the average
computer user?
The 64-bit question
Since any software project takes a while to get to a mature
stage, when I launched a start-up earlier this year, I decided
to concentrate on the kind of desktop systems which I be-
lieve will be common among creative users in the future.
We’ve had native 64-bit Linux on the Alpha and the Itanium
for years, but these architectures never reached the main-
stream desktop—and I don’t think they ever will. SGI now
has an Itanium2 based GNU/Linux desktop product aimed
at the creative market, but it costs US$20,000 per machine.
Compared to Windows or any other operating system,
GNU/Linux clearly has a head start on x86 64, and you can
choose from a number of natively compiled desktop distri-
butions for the platform. Unfortunately for the creative user,
all of these are aimed at the general purpose audience. It’s
impossible to be all things to all people, and what’s good for
the so-called consumer is rarely right for the content creator.
The 64 Studio default desktop is a slimmed-down Gnome install
For example, typical distributions use Arts or ESD
to share the sound card between applications, while
many GNU/Linux musicians would want to use
JACK—admittedly more complex, but far more pow-
erful. (I was asked recently what was so difficult about
JACK that means it isn’t found as the primary sound server
in any mainstream GNU/Linux distribution. I don’t think
it is difficult to use, but for the time being it still requires a
patched kernel, and some knowledge of sample rates and
buffers. Many users just want to be able to throw audio at
any sample rate to the soundcard, and could care less about
real-time priority.)
In addition, the creative user’s default selection of applica-
tions would be very different to—for example—a sysadmin.
Even gigantic distributions like Debian don’t package all of
the specialist tools needed for media creation, and the in-
tegration between packages is often less than perfect. So
the goal of 64 Studio Ltd. is to create a native x86 64 dis-
tribution with a selected set of creative tools and as much
integration between them as possible.
Why Debian?
Most of the packages in 64 Studio come from the unofficial
Pure 64 port of Debian testing, with some from Ubuntu,
some from DeMuDi and some custom built. A more obvi-
ous choice might be Red Hat, given that many of the high
end (which is to say expensive) proprietary tools used in

Free Software Magazine Issue 10, January/February 2006 49

 MIND SET

Hollywood studios and elsewhere are sold as binary-only JACK running with realtime priority on a native x86 64 desktop
Red Hat packages. However, the split between Red Hat En-
terprise and Fedora Core presents serious problems for any
derived distribution. You could rebuild Red Hat Enterprise
from source as long as you removed all Red Hat trademarks,
but that’s a lot of extra work—and you’d have to follow Red
Hat’s agenda for its distribution, which you couldn’t have
any input to.
---------------------------------------
GNU/Linux clearly has a head start on
x86 64
---------------------------------------

On the other hand, you could build a distribution on top

of Fedora Core. It’s broadly Red Hat compatible, and
there are the beginnings of a community process taking
place—although it’s still far more centrally controlled than
genuine grass-roots distributions. The key problem with this
approach is that Fedora Core is not designed or built to ac-
tually be used. I can say this with some confidence because
I was able to ask Michael Tiemann, former Red Hat CTO
and now vice president of open source, this question my-
self. Fedora Core remains a technology preview for Red
Hat Enterprise, and the Fedora Project has absolutely no
commitment to stability or usability. If Red Hat wants to
try a major update to see what breaks, it can.
Debian does have a commitment to stability, and a bona-fide
community process. There are other reasons for favouring
Debian over Red Hat: apt-get is just better than rpm when
it comes to upgrades, and on the creative desktop we’ll be
upgrading continuously. The work of the Debian Pure 64
port team is of a very high quality, not to mention that of all
the many Debian package maintainers.
I recognise that whatever packages we put into 64 Studio,
users will want some of the packages that we haven’t in-
cluded—so being able to use thousands of binaries straight
from the Pure 64 port without modification would be a ma-
jor advantage. Because we’re sticking very closely to De-
bian with the 64 Studio design, it’s our intention that users
will be able to install any application from Pure 64 simply
by enabling an additional apt source. This will include most
of the well-known applications with the exception of Open-
Office.org, which just won’t build natively on x86 64 yet.
In fact, 64 Studio is not so much a distribution based on
Debian as a Debian remix. 64 Studio maintainer Free
Ekanayaka is a Debian Developer, so we hope to contribute
our improvements back directly—where they are Debian
Free Software Guidelines compliant. However, we do bene-
fit from the flexibility of not being an official part of Debian.
For example, the Debian project has decided that they do not
want to package binary audio interface firmware, which is
required to be loaded by the driver for the interface to work.
That’s fair enough, and I understand the reasons for their de-
cision, but it’s a major pain if you own that kind of interface,
because it won’t work out of the box.
The alpha releases
There are a number of challenges we still have to face. The
first is following the rapid pace of kernel development. We
are currently using Linux 2.6.13 with Ingo Molnar’s real-
time preemption code and a few other patches. Not so long
ago, these patches didn’t build on x86 64 at all, and as far
as I know, we are the only native 64-bit distribution using
them. The first indications are that this combination works
really well for audio with full preemption enabled, the most
aggressive setting. For the time being we are using the
realtime-lsm framework to give real-time priorities to non-
root users, because we know it works. We may switch to
rlimits in the future, as this has been merged into the main-
line kernel now.
Another challenge is the issue of support for proprietary for-
mats within free software. At the level of encoding and de-

50 Free Software Magazine Issue 10, January/February 2006

coding, we think the best solution we’ve seen is the Fluendo
plugin collection for GStreamer, which as far as we can tell
meets the requirements of free software licences regarding
linking, and also the legal requirements of the patent hold-
ers. It’s simply not sustainable to expect users to locate and
download libraries of dubious legal status, and install these
by themselves. Apart from any ethical problems, it’s impos-
sible to support users properly in that situation. Download-
ing these libraries is certainly out of the question for any
institutional user, such as a college.
At the level of project interchange, for example moving a
complex project from Ardour to ProTools, there does seem
to be a move among proprietary applications towards sup-
port for AAF, the Advanced Authoring Format. Free soft-
ware must support this kind of high-level project compati-
bility format, otherwise it doesn’t stand a chance of gaining
a significant user base. When I talk to people in the music
industry, it’s almost a mantra that “everyone mixes in Pro-
Tools”. I’m not aware of any free software audio application
that supports ProTools format import or export, but at least
with AAF we have the chance of finding a middle way.
64 Studio version 0.3.0 alpha is currently available for
download (http://www.64studio.com/) as an .iso
image. Changes from stock Debian include X.org instead
of XFree86, the custom kernel package, and a base selec-
tion of packages including the Gimp, Inkscape, Blender,
Ardour, Jamin and Kino. Version 0.4.0 came out at the
end of September with more packages and enhancements,
and the distribution is seamlessly upgradeable from a 0.3.0
install with apt-get of course. We’d be more than pleased
to hear your test reports and suggestions for the distribu-
tion—you can help us make free software the creative desk-
top of choice.

Copyright information

c Daniel James 2005. Verbatim copying permitted

About the author

Daniel James was one of the founders of LinuxUser &

Developer magazine, and the original director of the lin-
uxaudio.org consortium
Towards a free matter economy
(Part 4)
Tools of the trade
Terry Hancock
A good scientist is a person with original ideas.
A good engineer is a person who makes a de-
sign that works with as few original ideas as pos-
sible. There are no prima donnas in engineer-
ing.—Freeman Dyson
magine where free software would be today if it
I weren’t for the GNU C Compiler! Just as free soft-
ware depends heavily on free compilers, so does free
design rely on having free computer aided design
and authoring tools.[1]
Before the gcc was created, when free software had to be
written on proprietary compilers, the software development
community was limited to the very small number of peo-
ple who could afford to purchase such tools—either because
they were professional programmers or very dedicated am-
ateurs.
The idea of “bazaar style” development hadn’t yet been con-
ceived, but that was just as well, since such small bazaar
sizes would lead to a breakdown of the bazaar development
strategy[2]. No doubt some centrally controlled “cathedral”
projects (such as the GNU project) would’ve continued, but
the overall effect would be an extreme chill compared to the
hotbed of innovation that free software currently represents.
Lacking the kind of professional-quality design authoring
tools that are found in the commercial engineering work-
place, the free design community is in just that situation
today.
Of course, it can be (and has been) argued that engineering
is a specialized discipline and that therefore only a small
52 Free Software Magazine Issue 10, January/February 2006
technocratic elite can participate in the process—hence
bazaar size might always be too small to be effective. But
twenty years ago, this was the conventional wisdom about
software development, too!
By users for users
A compiler is a program to write programs, so the users of
the tool are also those qualified to create it. With design
tools, we’re not so lucky: it’s a fairly rare engineer who
has the programming expertise to develop proper engineer-
ing design software. This is a problem because bazaar de-
velopment works best when applications are developed by
the people who need them. In order to get a CAD/CAM
system started, it will probably be necessary to start with
a centrally organized solution—a cathedral development
model—working as quickly as possible towards a solution
that relies heavily on scripting that can be done by the typi-
cal end user of engineering software.
This is the sort of approach that has proven itself functional
in many projects in the multimedia and desktop application
sphere: customization and scripting facilities have been the
free software solution for desktop environments like KDE
[3], vector graphics applications like Skencil [4], 3D anima-
tion and modelling applications like Blender [5], and partic-
ularly in game engines. All of these are situations where the
same problem applies: the typical user is not a particularly
skilled programmer, so a simplified programming environ-
 MIND SET
Figure 1: QCAD GUI. QCAD has an excellent and intuitive
palette-based GUI that allows quick access to drawing constraint
features, and is easy to learn
ment is needed to make a more graded slope from user to
developer.
This effectively increases the bazaar size for the parts of
the program that can be scripted. User interfaces matter
the most to the serious user and are the hardest for the pro-
grammer (who is not particularly skilled in the application
domain) to predict. A user interface with a scripting engine
gives a lot of leverage for the overall program, and the avail-
ability of easily-embeddable high-level interpreters such as
Python make such a facility easy to provide. Thus, the “by
users for users” philosophy of free software design can be
stretched to suit real world applications, where skillsets vary
between users and developers.
Computer aided design, simulation, and
manufacturing
Although there are many other areas of design software
which are useful and available, and many others for which
no free solution yet exists, the most pressing and obvi-
ous need is for a general purpose 2D/3D mechanical Com-
puter Aided Design and Computer Aided Manufacturing
(CAD/CAM) system.
Specialized cases of CAD have already been covered by free
software offerings—such as xcircuit [6] for drawing and
capturing electronic schematics and pcb [7] for designing
printed circuit boards, and general purpose 2D CAD draw-
Free Software Magazine Issue 10, January/February 2006
ing applications such as QCAD [8] are moderately well-
developed (although they still fall short of proprietary com-
petitors). The GNU EDA project[9] is making progress in
the direction of integrated circuit design, and it should not
be surprising that these “highly ephemeralized” technolo-
gies are among the first addressed.
---------------------------------------
Specialized cases of CAD have already
been covered by free software offerings
---------------------------------------
Getting into the harder technologies, such as aerospace or
mechanical engineering or robotics, requires much more so-
phisticated 3D CAD tools than those that are currently avail-
able. The modern manufacturing industry relies extensively
on these tools to produce the kinds of complex technologies
that we are accustomed to, and we won’t have much chance
of being competitive in the free design world, until we’re
able to use tools that are at least as good as these.
Friendlier interfaces
The interfaces in modern versions of proprietary CAD sys-
tems such as AutoCAD [10] have hardly changed at all in 20
years. They have traditionally relied on the high expressive
power at low development cost of linguistically-oriented
command line interfaces, leaving their visual-tactile graphi-
cal interfaces to stagnate. End user oriented free software
artistic 3D tools like Blender, however, have seen much
greater innovation, and they have demonstrated that a well-
designed graphical interface for power users can greatly in-
crease productivity, even for very complex tasks.
Collaborative design
In order to maintain a sufficient size for a free-design bazaar
to establish itself and grow, CAD systems which are ad-
equate to the task must be created. They must be free-
licensed, so that all comers to the community can partici-
pate. They must use free, standardized drawing storage for-
mats, and they must be friendly to version control and any
other necessities of online collaboration.
Furthermore, they must permit a form of markup that allows
communications about drawings to occur electronically. To
my knowledge, this has never even been fully achieved even
53
 MIND SET
Figure 2: Blender GUI. Blender’s interface is cursed by new users,
but greatly loved by power users. It does seem a bit daunting at first,
but I quickly became attached to it. It compresses an enormous
amount of options into a small space with an intuitive and highly
internally-consistent design, using color-coded widgets, icons, and
words as needed. If not actually used in a 3D CAD system, I think
it should at least be emulated
with proprietary software, although I am aware of an aban-
doned attempt at the Jet Propulsion Laboratory called Su-
pernova, based on “Lambda MOO” technology[11].
However, these network effects are precisely the sort of
problems that free software is better positioned to solve!
CVS, Subversion, and most forms of internet interpersonal
communication are the types of software that started out
free.
Integration with Narya Bazaar
In the Narya Bazaar system (see parts 1 (http://www.
freesoftwaremagazine.com/free issues/
issue 07/free matter economy), 2 (http:
//www.freesoftwaremagazine.com/free
issues/issue 08/free matter economy 2) and
3 (http://www.freesoftwaremagazine.com/
free issues/issue 09/free matter economy
3) of this series), there’s a vital role played by specification
objects which must define work to be done adequately to
form a contract between the donors, projects, and vendors.
Whatever CAD system we create must be up to the task
of defining these specification objects adequately for such
agreements. That will require a commercial-quality CAD
system, and it will need to support a range of meta-data
54 Free Software Magazine Issue 10, January/February 2006
from tolerances and materials to the relationship between
components in large assemblies. Furthermore, as the bazaar
community grows, parts from many different vendors will
have to be integrated.
This has long been a problem in commercial design appli-
cations (Consider what happens when space station com-
ponents from RK Energia in Russia need to be mated with
components from Lockheed Martin in the USA), and good
CAD systems constitute the solution. In the bazaar environ-
ment, virtually all components will be “outsourced”, so it
will be necessary to use good engineering communications
processes to ensure that components interact properly.
Breaking the proprietary grip on the CAD
market
There is a kind of circular logic that keeps people locked
into the systems they have—they are expensive and difficult
to use because they have a very small user base. But the
user base is small, precisely because they are expensive and
difficult to use!
The providers of CAD software have effective monopoly
power of a very limited market, defined by the restriction
to the commercial industry who can pay the high licens-
ing costs (typically $3000 per seat! [12]). Since the pro-
prietary industry mostly develops proprietary technology,
there hasn’t been that much progress in making CAD sys-
tems fully collaborative, nor in making them integrate with
communications channels to allow the virtualization of such
standard engineering rituals as design-reviews and mark
ups.
These systems should be “low hanging fruit”, easy to pick
off, but for one very unpleasant reality: 3D CAD systems
are intrinsically very complex pieces of software that take
a lot of work to write. Ultimately, however, they exhibit
all of the criteria for a free software success story: they are
complex, have rich network effects, and are just the type of
high buy-in software that users expect to pay maintenance
and support contracts for.
Unlike the desktop computing market, the manufacturing
market is very much aware of the importance of standard-
ization and the dangers of lock-in. Lock-in to single-source
suppliers is a long-standing trap in the manufacturing indus-
try, and managers already know the importance of avoid-
ing it, which is why the value proposition of free-licensed
 MIND SET
Figure 3: AutoCAD GUI. Although AutoCAD certainly has a
GUI, many users find themselves relying on the command run line
to accomplish most 3D modelling tasks, perhaps reflecting Auto-
CAD’s origin as a 2D CAD system (Image credit Wikipedia)
software will be easier to present to them. On the other
hand, there is still a credibility gap in proving that better
(or even adequate) software can be maintained using free
software development models—since conventional wisdom
in the engineering community understandably favors engi-
neered solutions (but there is a recent “rapid prototyping”
trend, which could be comparable).
Pieces of the puzzle
We have to consider the individual components that are
available to piece together a CAD system out of major el-
ements that already exist in the free software community.
To do this, I’m going to assume a “Model-View-Controller”
(MVC) model of development [13], in which steps are taken
to keep these major components separate from each other,
although there are some reasons why we might want to relax
this constraint.
Here I’m going to assume Python as an integration lan-
guage, to simplify some of the choices, and also because
Python is a language that is pretty popular and fairly easy to
use for engineers as opposed to career programmers. For a
project to succeed in the free software world, it’s important
that the implementation language be accessible to a suffi-
cient quantity of the program’s users to form a healthy de-
velopment bazaar. [14]
Free Software Magazine Issue 10, January/February 2006
Model (standard representation of CAD data and
file formats)
CAD drawings are complex structured data, which is intrin-
sically object-oriented. Therefore it makes sense that some
form of object database is used to hold drawing data, but the
design of the schema is pretty complex.
There isn’t one obvious way to represent a 3D object math-
ematically; there are many different systems, optimized for
different applications. Fortunately only a relatively small
number of the nearly infinite possibilities are actually in
use, and so the representation of CAD data remains solv-
able. But it’s a big enough problem to be recognized by the
commercial manufacturing industry, and to call for some
form of standardization.
Luckily for us, this has happened in the form of the Standard
for the Exchange of Product model data (STEP) standard
(also known as ISO 10303 and ANSI PDES) [15].
The most important contribution of STEP is not the data
representation, but the classification and standardization of
the object-oriented data models for representing CAD data.
There are so many representations in use, segmented across
so many industrial classifications, that collecting all of that
into one standard is a monumental task, and the STEP stan-
dard schemas are indeed a monster as a result. However, this
means we can regard the STEP standard as a fairly complete
assay of what the manufacturing users need in a CAD sys-
tem, which provides a blueprint not just for the file formats
that we will use, but the internal data representations, and
even the GUI controls that must be provided to manipulate
that data. Even though STEP is intended as an exchange
standard, it also provides a plan for developing the software
itself.
STEP is not as open as we would like it to be. There don’t
appear to be any patent encumbrances, which is important,
but the “source” documents from ISO are mostly copy-
righted and available only for a fee [16], which you usually
must pay to your national standards organization (ANSI in
the USA, for example), and at least at present, the fees are
too high for individual developers to pay. Since a free soft-
ware development project relies on the cooperation of unaf-
filiated parties to operate, there’s no easy way to spread the
cost of these documents over a development team as there
would be in a single company, so this practice is effectively
protectionist to proprietary software vendors. This is not
55
 MIND SET

Special domain design tools

Although I believe 3D mechanical CAD is the most important design authoring challenge to be met, there are a lot of
specialized domains for which there are existing free software design applications.
For software and systems design, there is GNU’s dia program which makes several kinds of specialized diagramatic
drawings easy to create. What can’t be drawn with dia can be represented with more general vector graphic programs
such as Inkscape or Skencil, and presented with KPresenter, or other free software slide presentation packages.
Electronic schematics can be drawn easily with more than one program, including for example, xcircuit and oregano.
The GNU EDA project primarily addresses integrated circuit design, and there is pcb for designing printed circuit
boards. The biggest complaint I have about these applications is that they have very inconsistent interfaces. I can
imagine a project to refactor these programs to a common GUI standard, but they are each well-established in a small
design community of their own, so it isn’t clear how popular such a project would be.
Shop drawings for mechanical projects can be created in 2D using programs like QCAD or PythonCAD, and indeed,
this is how things were once done in the manufacturing industry, although it’s easy to see that relying on this leaves us
decades behind the commercial proprietary design world.

particularly popular with the manufacturing industry (who compatibility, but it’s a definite start at liberating the STEP
are the customers of those vendors of course, and would standard.
stand to benefit considerably from a free-licensed CAD so-
lution), nor apparently with the ISO standards developers
with whom I corresponded in preparing this article. [17] Figure 4: Liberating STEP. The STEP standard could be “liber-
ated” using the mapping tools from the Step Modularization project
So, if you look closely enough, you will find that a free- to convert Express schemas to an XML format. Once converted
licensed open-source standard is there, in the form of the to XML, there’s a wide range of object-mapping tools for different
Express schema listings for STEP which are available freely languages, including Python. It won’t be a unique mapping, as it
(they are explicitly copyright disclaimed by the ISO, which will also depend on the translation tools used, so it will probably
be desireable to standardize on a toolchain after testing the alterna-
means they are essentially subject to public domain rules tives, and provide free documentation for the XML-based derived
[18]). If you regard the ISO “source” documents as docu- standard
mentation, which is really what they are, and the Express
listings as the true source, then STEP can at least unoffi-
cially be used as a free format.
There are already two important sites dedicated to this map-
ping of the STEP standard into a free format, including
the NIST’s Step Modularization project [19], now hosted
at Sourceforge, which provides an Express-to-XML trans-
lator, and the Engineering Exchange for Free (EXFF) site
[20], which contains a number of useful resources for STEP.
Combined with the free Express listings defining STEP,
it’s therefore possible to map the standard first from Ex-
press to XML, and then from XML to a variety of object-
oriented programming languages, by using existing XML-
to-object mapping libraries, available for popular free soft-
ware programming languages, such as C, C++, and Python.
Clearly, the wide range of alternatives in this mapping pro-
cess means that there will be some issues with assuring

56 Free Software Magazine Issue 10, January/February 2006

 MIND SET
There’s also a library called OpenCascade [21] based on the
SDAI C++ implementation of STEP, but despite the name,
this package is “non-free”. Free software developers who
want to base their work on an SDAI implementation would
do better to go back to the original NIST C++ implementa-
tion [22], which is “copyright free” (produced by the US
government).
Less expensive electronic forms of the complete ISO
“source documents” may become available in the future,
but there is no plan to make them freely distributable, as
far as I have been able to ascertain. Working from the freely
available documents to create a “liberated” CAD standard
is probably the best strategy.
View (3D rendering)
At the most basic level, of course, are fast 3D rendering
libraries and hardware, such as Open GL, but these are pri-
marily for interfacing with (or emulating) hardware, and so
are really more primitive than we want. Fortunately, many
3D rendering libraries have been built on top of these stan-
dards, primarily driven by game and computer animation
applications.
Some of the more prominent free-licensed 3D rendering li-
braries that exist include: Crystalspace [23] (written in C++,
mainly to support massively-multiplayer online games),
Soya 3D [24] (written in Python and Pyrex, also mainly for
games), and Blender [25] (written in C, mainly for commer-
cial computer animation applications). All are under suit-
able free licenses (GPL).
Crystalspace and Blender are both primarily frameworks,
which embed a Python interpreter; while Soya 3D is a
python module library. The latter is better from the point
of view of embedding the viewer as a major component, es-
pecially if the integration language will be Python. Blender
also has embeddable builds: a game engine and a browser
plugin, that might be investigated as options for a more pure
“view” component.
In fact, though, the “impurity” of Blender as both “view”
and “controller” may be an advantage, since there is already
excellent control for viewing or browsing a 3D model at var-
ious detail levels. This is a large part of what one would
hope to gain by using Blender, so it’s worth considering
abandoning the MVC model for this more complete system.
Free Software Magazine Issue 10, January/February 2006
Controller (graphical user interfaces)
Given that the concept here is for a fully internet-connected
collaborative CAD system appropriate for free-licensed de-
sign projects, a very attractive idea is to integrate it with
cross-platform browser technology. This would mean using
Mozilla’s XPFE system, and therefore XUL for the GUI en-
vironment. The biggest drawback to this has hitherto been
the limitation to Javascript as a GUI scripting language,
which is less than ideal for a project of this type. By the
time you read this, however, XUL will likely support Python
bindings, so this problem is being resolved [26]. This would
have the further advantage of easing integration with other
out of channel communications such as email, web forums,
and chat systems, in addition to the formal markup and re-
view process that must be intrinsic to a collaborative CAD
system.
The design of the GUI, both as a skin style and as a guide
to organization would do very well to follow Blender’s lead.
Although Blender’s non-standard GUI is often criticized by
newcomers, it is much loved by the people who use it daily.
It is highly optimized for efficient use in 3D modelling ap-
plications by people who use it a lot. That makes it an excel-
lent choice for a 3D CAD GUI as well. Although using the
Blender interface itself (rather than emulating it with XUL,
for example) is an attractive option, it has some technical
obstacles: Blender isn’t really factored into an MVC model.
And, although there’s some developer interest in this, it isn’t
a priority for Blender developers, who are focused on the
core tasks of artistic 3D modelling and animation.
Other GUI options include the usual suspects, such as
PyGTK, Qt, and wx.
Design concepts
There are obviously many, many different ways in which
these components could be assembled to create a function-
ing, network collaborative 3D CAD/CAM system. Using
Python as an integration language (as I have chosen to do for
the server side on the Narya project), one could use ZODB
[27] as an object model, deriving the schema from the ex-
press schemas for STEP via the Step Mod package. Then
one could, for example create a Mozilla-based design using
the new Python bindings for XUL and integrate with a cus-
57
 MIND SET
Figure 5: Ideal MVC design. Ideal MVC designed CAD sys-
tem elements. The purist approach to the CAD system would be
to create separate model, view, and controller components (in a re-
ally ideal case, each would be pluggable, allowing for a variety of
separate but compatible tools to coexist, but I’ll be happy with just
one of each). This approach is the soundest for development from
scratch (or even nearly from scratch)
tom build of Blender based on the game engine or plugin
versions.
Or one could abandon pure MVC design, and build the sys-
tem entirely within the Blender Python API. Since Blender’s
internal model lacks features of the STEP model, and there
is not likely to be a round-trip guaranteed set of transfor-
mations between the two, it is likely that it would be nec-
essary to maintain two parallel models—a “representation
model” within Blender, used for visualization, and a “de-
sign model” within the CAD database, which is the true
“original”. A transformation would be provided to render
the design to representation in order to display it. Tweaks to
the Blender source code (in C) would probably be needed to
provide CAD objects with callbacks that could be handled
by Python callbacks acting on the design model (otherwise,
the representation model could drift away from the design
model whenever non-CAD operations were performed on
it, causing a synchronization problem and a lot of user con-
fusion).
58 Free Software Magazine Issue 10, January/February 2006
Sometimes you need a cathedral after all
Sophisticated 3D graphics and design modelling software
is no picnic to write. This is one of those big, complex,
highly-interdependent problems at which “cathedral” engi-
neering seems to excel, and on which “bazaar” development
frequently stumbles. The many half-finished projects in this
area attests to this point. It will be necessary to create a suc-
cessful prototype of the core of such a system before it can
be built on and extended by a bazaar community. That pro-
totype will require a lot of “up front design” contrary to the
new conventional wisdom of “extreme programming”.
Machines in general are complex, tightly-coupled systems,
so it shouldn’t come as a surprise that the mathematical
models of mechanical systems are also complex and tightly-
coupled, nor that the CAD/CAM software systems that ma-
nipulate those models need to be.
Fortunately, there is strong reason to believe a free software
CAD system could be developed commercially. CAD sys-
tems are high-buy-in software, not unlike web servers, soft-
ware compiler and library toolkits, or data reduction and
analysis systems. They are primarily used within organiza-
tions which rely heavily on them working without a hitch.
This means there is a substantial motivation for those or-
Too many cooks
In researching this article, I found nearly a dozen dif-
ferent initiatives to create free software CAD/CAM sys-
tems, started by different people, and ranging widely in
success. None of them was on the scale of the need ex-
pressed in this article, but they make a clear case that
such software is wanted, and there are people willing to
do work on it. Also, few projects referenced the others,
which suggests a lack of communication and organiza-
tion.
I know that I don’t really have the resources to pur-
sue a CAD programming project myself (unless through
some form of commercial collaboration), but what I cer-
tainly can do is provide a resource Wiki to keep track of
other people’s projects. Please have a look (http://
client.narya.net/Wiki/CadSystems), and
help me update the site with any additional projects that
you know about. Maybe if the information is collected
in one place, a solution will present itself.
 MIND SET

ganizations to pay for maintenance and support contracts, Figure 6: CAD visualization. Model of hardware for the NASA
and therefore a Cygnus Solutions [28] or Zope Corporation “return to the moon” initiative. Visions of space frontier technology
[29] business model (free-license the software, but charge have relied extensively on CAD and 3D models, to the point that
for support services) would seem to be viable. people expect to see such images before they take a technological
idea seriously. It is far too much of a disadvantage for free de-
Free software tools and free data exchange formats are in signs to either reject 3D CAD or use proprietary 3D CAD systems.
Blender can already create scenes like this, but without the technical
the interest of the majority of the manufacturing industry,
data model to back it up (Image credit NASA/John Frassanito and
as demonstrated by initiatives such as EXFF. Thus we can Associates)
expect to see cooperation and support from the end users
of CAD once a serious effort is made to implement it, even
though we’ll also see competition from existing vendors of
proprietary CAD systems, such as AutoDesk, and free soft-
ware will still have to prove itself to a new set of users.
The real challenge is getting started. Cygnus started out
supporting GNU software that already existed and was de-
veloped for quite different reasons, while Zope Corporation
(then called Digital Creations) originally developed their
product on a proprietary basis. It was only after these soft-
ware products already existed and were relatively mature
that these businesses made the decision to make support
contracts their primary source of income. I’m not aware
of an example where such a complex piece of software was
developed by a company planning to release it for free and
support it commercially.
---------------------------------------
This is an area where a free software
CAD system could not only match the
Planning for the future
proprietary competition, but leave it in
the dust
Although free design is already beginning to be done in lim- ---------------------------------------
ited domains now, the availability of better design authoring
tools to a wider range of potential developers is needed to This is an area where a free software CAD system could
spur an explosion of interest in bazaar model development not only match the proprietary competition, but leave it in
of hardware. Nowhere is this gap more severe than in the the dust—simply by leveraging the many years of experi-
area of 3D mechanical CAD/CAM, which is extremely im- ence in building collaborative, internet-based systems that
portant for the kinds of aerospace and mechanical engineer- are the legacy of free software developers worldwide. A
ing that will be called for in the development of hardware true free software, general purpose, collaborative 3D me-
that will be required by space pioneers as they attempt to chanical CAD/CAM system won’t be simply as good as the
settle and build using the materials they find in their new en- proprietary alternatives, it will likely be better than anything
vironment. If we want to follow a free development model else on the market.
with all the advantages it implies, we’ll need this kind of
software to be created, and there is reason to believe that it Bibliography
can be profitable as well as widely useful to develop it, al-
though it will require a fair amount of entrepreneurial vision [1] Richard Stallman. The GNU Project (http://www.
to make that happen. gnu.org/gnu/thegnuproject.html).

Free Software Magazine Issue 10, January/February 2006 59

 MIND SET
[2] Forrest J. Cavalier, III. Some Implications of Bazaar Size
(http://www.mibsoftware.com/bazdev/), 1998.
[3] KDE Documentation (http://docs.kde.org/).
[4] Skencil Development Guide (http://www.
nongnu.org/skencil/Doc/devguide.html).
[5] Blender Documentation (http://www.blender.
org/modules/documentation/).
[6] xcircuit (http://opencircuitdesign.com/
xcircuit/).
[7] pcb (http://pcb.sourceforge.net/).
[8] QCAD (http://www.ribbonsoft.com/qcad.
html).
[9] GNU EDA (http://www.geda.seul.org/).
[10] AutoCAD (http://en.wikipedia.org/
wiki/AutoCAD).
[11] Michael Brundage, Network Places (http://www.
qbrundage.com/np/index.html).
[12] Price estimated from CDW Catalog (http://www.
cdw.com), August 2005.
[13] Model View Controller (http://en.wikipedia.
org/wiki/Model view controller) Definition.
[14] Terry Hancock. Praise for Python (http:
//blog.freesoftwaremagazine.com/users/
t.hancock/2005/11/11/praise for python),
2005.
c 2005 Terry Hancock
[15] STEP / ISO 10303 (http://www.tc184-sc4.
org/SC4%5FOpen/SC4%5FWork%5FProducts%
5FDocuments/STEP (10303)).
[16] ISO Restricted Documents Notice (http:
//www.tc184-sc4.org/Notices/Authorized%
5FUsers%2DPadlocked%5FDocuments/).
[17] Howard Mason, Chair, ISO TC 184/SC4. Private com-
munication.
[18] Express Listings (http://www.tc184-sc4.
org/SC4%5FOpen/SC4%5FWork%5FProducts%
5FDocuments/STEP%5F%2810303%29/) (Search
each block for “schema”).
60 Free Software Magazine Issue 10, January/February 2006
[19] STEP Modularization Project (http://stepmod.
sourceforge.net/).
[20] EXFF (http://exff.org/).
[21] OpenCascade (http://www.opencascade.
org/) (non-free).
[22] NIST SDAI C++ Library (http://www.mel.
nist.gov/msidstaff/sauder/SCL.htm) (free-
licensed).
[23] Crystalspace (http://www.crystalspace3d.
org/).
[24] Soya 3D (http://gna.org/projects/soya).
[25] Blender (http://www.blender.org).
[26] Brendan Eich Python for XUL scripting
(http://weblogs.mozillazine.org/roadmap/
archives/008865.html), 2005.
[27] Using ZODB Standalone. (http://www.zope.
org/Wikis/ZODB/FrontPage).
[28] Cygnus Solutions (http://en.wikipedia.
org/wiki/Cygnus Solutions).
[29] Steve Litt. Zope Corporation/Digital Creations
(http://www.troubleshooters.com/tpromag/
199906/ digcreate.htm), 1999.
Copyright information
This article is made available under the “Attribution-
Share-alike” Creative Commons License 2.0 available
from http://creativecommons.org/licenses/by-sa/2.0/
(http://creativecommons.org/licenses/
by-sa/2.0/).
About the author
Terry Hancock is co-owner and technical of-
ficer of Anansi Spaceworks (http://www.
anansispaceworks.com), dedicated to the
application of free software methods to the develop-
ment of space.
Free software liberates
Venezuela

The free software revolution comes to Venezuela

David Sugar

he third International Forum on Free Knowl- los Gentile who personally asked me to attend. And finally,

T edge brought together many groups and indi-

viduals interested in the development of free
software worldwide to the city of Maracaibo.
One reason Venezuela choose to host this event is because
starting in January (2006), their new free software law, di-
rective 3.390, comes into effect, which mandates all govern-
ment agencies to migrate to free software over a two year
period. I was invited to speak about Telephonia Libre: the
use of free software in telecommunications.
---------------------------------------
Directive 3.390 mandates all
government agencies to migrate to free
software over a two year period
---------------------------------------
I have always been immensely curious about Venezuela.
While there, I had the extremely lucky chance to speak with
directors in many of the organizations charged with carrying
out Chavez’s vision of a “Bolivarian Revolution”.
While my travel had been planned a number of weeks in ad-
vance, as with all travel I have experienced in Latin Amer-
ica, this turned out to be on a different concept of time. I
didn’t hear back at all from Venezuela until the weekend be-
fore departure, but this is actually not that remarkable. By
Monday the 21st of November, I knew I’d arrive in Mara-
Map of Venezuela

While I am invited to speak at many events and conferences

worldwide, most often I reject them immediately because
they’re not open to the general public. This is one of the
reasons I rarely speak in the U.S.—virtually all U.S. scien-
tific conferences in my field are for profit and are organized
by groups who charge fees so high that it discourages the
general public from participating, or they are organized for
the benefit of commercial vendors who are trying to market
themselves to potential customers. Science Fiction conven-
tions actually would be closer to my choice of U.S. venue,
although I don’t seem to get invited to those.
I accepted this invitation for several reasons; first, it was
open and free to the general public. Second, it was Juan Car-

Free Software Magazine Issue 10, January/February 2006 61

 MIND SET
caibo the next day, and return on the 29th. That much was
confirmed to me by Ambar Rodriguez, who works for Cona-
tel, which is their state telephone regulatory agency. I had a
chance to speak with Ambar over the weekend, but I didn’t
know which airport I would departing from, or even what
airlines I’d be flying, until Monday morning.
To understand the blissful attitude I had taken, you have to
understand this: I recall one time I was staying with a fam-
ily in San Paulo, where we were scheduled to take a flight
to Porto Alegre. The airport was across town, and our de-
parture time was about half an hour away when we finally
wandered out to the car. We didn’t even travel in much of a
hurry. Yet, somehow, in the twisted and bizarre time warp
that is Brazil, we arrived on time for our flight anyway, and
I never quite figured that out either. Time often has a very
different meaning in Latin America.
---------------------------------------
Samán proposes that the mission of
SAPI should instead become that of
promoting “Intellectual Prosperity” by
creating laws and services that promote
the ability to share knowledge as the
common heritage of all mankind
---------------------------------------
Many of the events and presentations at the event were,
much like mine, of a rather technical nature. My presen-
tation caused some difficulty for the translator I was given,
who had no experience or understanding of the specialized
technical terms I was using. This was only corrected near
the end when a different person came forward to translate
my speech. Some presentations were from groups who were
using free software in some social setting. The event was
heavily attended by many people, and particular technical
directors from many key parts of the Venezuelan govern-
ment, because of their migration plans for 2006.
I eventually meet up with Jeff Zucker from Perl Mongers,
who traveled by bus from Caracas and the well known in-
ternational free software activist, Juan Carlos Gentile, who
drove all the way from Caracas along the same roads with
Marko, who is also from Italy. While it is said to take
ten hours to drive from Caracas to Maracaibo, as he and
Marko are Italian, naturally I expected he would arrive in
only five. These three, and Ana Isabel Delgato from the De-
62 Free Software Magazine Issue 10, January/February 2006
bian Venezuela group, were my primary “translation team”
whenever I spoke with others who didn’t speak English.
The People’s Ministry of Economics
Venezuela is blessed with not one, but two economic min-
istries. There is the old ministry of economics, which deals
with the traditional capitalist economy. It is worth noting
that capitalism continues in Venezuela and will likely con-
tinue to do so for some time. While lands are at times re-
distributed to landless laborers, for the most part existing
industries and businesses are left alone, and left to the old
ministry of economics. Instead, they have a different idea
of how to transform society here, and this brings me to the
second ministry.
The Ministerio Para La Economia Popular, or roughly, the
People’s Economic Ministry (and for simplicity, to be re-
ferred to simply as Minep) is tasked with transforming
Venezuela with a new economy. While the ministry does
a number of important tasks, I believe their most interesting
is to train and educate ordinary Venezuelans, who volun-
teer on how to run a worker co-operative. This is done by
providing co-ops the tools, financing, and practical training
they need to operate their new enterprises.
My interest in this aspect of Minep came in part from their
interest in providing VOIP services along with the comput-
ers they are offering to their worker managed co-ops. This
was a rather specific technical issue, and one they were very
interested in discussing with me.
Many of these worker co-ops are composed of very small
startups that typically have 10 people or less. Minep offers
training and support, as well as financing, to allow co-ops to
purchase computing systems for their business needs. These
systems are now offered entirely with free software, starting
with the Debian GNU/Linux operating system, along with
Open Office for general business use, and web hosting under
Apache. Co-ops that go through the Minep program also
have the ability to host web sites with their own content,
and these usually feature the products or services a given
co-op wishes to offer. Co-ops are also trained in the use of
the free software they receive and in how to maintain their
own IT infrastructure.
The Minep co-op training program was piloted in 2004,
with some 3000 such worker managed co-ops formed. Dur-
ing this year (2005) they have formed over 45,000 such co-
 MIND SET
ops nationwide, and are expecting to train over 700,000
Venezuelans in how to form and be part of a new econ-
omy. This suggests to me that perhaps 40% of those that
go through the Minep program eventually do form a com-
mercial enterprise.
---------------------------------------
The Venezuelan revolution is perhaps
the first revolution in history saved by
computer hackers
---------------------------------------
The use of free software and offering of computer systems
for business use as part of the co-op program is actually rel-
atively new. I believe, if I understood correctly, the full ver-
sion of free software training program is a 6 month course,
and so is rather comprehensive. This year (2005), they’ve
only trained people from at most a few thousand of the co-
ops on the use of free software through the initial pilot pro-
gram. In 2006, however, that program, and free software
training should be available to all interested.
The Ministry of Intellectual Prosperity
SAPI, the Independent Service ministry of Propiedad In-
tellectual, is the ministry that used to define Venezuela’s
so called “Intellectual Property” laws. I understand SAPI
also at one time concerned itself with the issue of what was
called “Piracy”. I would have thought, however, that con-
trolling murderous gangs of anarco-capitalist “gentlemen of
fortune” who raid ships, would be the job of the navy, or
perhaps the interior ministry.
I had the good fortune to meet the current director general
of SAPI, Eduardo Samán, while I was in Maracaibo. He
has very different ideas for the purpose of SAPI. He is a
well known internationalist, and had been a key person in
establishing the program for promoting a developing na-
tions agenda within WIPO. Rather than creating new in-
tellectual restrictions, Samán proposes that the mission of
SAPI should instead become that of promoting “Intellectual
Prosperity” by creating laws and services that promote the
ability to share knowledge as the common heritage of all
mankind.
Assuming that private corporate interests in the developed
world today do succeed in the great program of owning what
people are allowed to think, it is very possible that places
Free Software Magazine Issue 10, January/February 2006
“Intellectual Property”
The term intellectual property itself is of course a new-
speak propaganda word that didn’t even exist 20 years
ago. First, the topic it covers varies from Copyrights,
Patents, Trade Secrets and Trademarks, to a variety of
other things, all of which are in reality all very different
and unrelated. Second, it’s based on the premise that
you can give something intangible to someone else, and
yet control it and decide what other people do with it,
as if it or they (and even the ideas they may have about
it) were your physical property. Intellectual property
amounts in part to thought control through legal fiction.
Some even say it amounts to Intellectual Slavery.
The consequence of treating ideas and thoughts as if
they are tangible property is the very destruction of sci-
ence and education and the elimination of individual
rights and freedoms. Science is in part built upon the
idea that new knowledge is created by incrementally
improving ideas. Education is based on the idea that
one can learn from existing things and then use that
knowledge to create new works. The idea behind “Intel-
lectual Property” interferes with both. It is barbarism,
and could well lead to a new “dark ages”, where only a
privileged few are allowed to learn, under the exclusive
control of greedy intellectual monopolies.
Since “Intellectual Property” involves exclusive licens-
ing, when public universities do this and then let oth-
ers license their discoveries, the public is made to fund
research that only benefits a small number of people.
Even worse, those companies which receive such fund-
ing can then use this exclusive grant to sell back to so-
ciety the fruits of what society already paid for. This
can be thought of as paying for something twice. This
could also be thought of as public welfare for private
corporations, or more simply: exploitation.
like Venezuela will become the new leading nations in sci-
ence and technology.
How oil fuels the Bolivarian Revolution
Maracaibo is also the heartland of the oil industry, and the
state run oil company, PDVSA. Oil companies are also tra-
63
 MIND SET
ditionally conservative in nature. However, PDVSA also is
a contrast, as both the primary wealth producing institution
in the country, and the strongest source of support for Pres-
ident Hugo Chavez’s revolutionary changes.
I met a number of PDVSA oil workers, who seemed well
represented among the ranks of PDVSA management. I also
had the chance to talk over lunch with one of their directors,
Socorro Hernendez, as well as Jose Luis Rey, whose renoun
is both as a skilled hacker and financial genius who was
involved in helping rebuild the financial trading systems that
were sabotaged in 2003.
Today, the state-run oil company is a major backer of the
free software movement (software libre) in Venezuela and
is a major sponsor of the 3rd International Forum on Free
Knowledge, which is what brought me to Maracaibo. Every
question related to the use of free software in Venezuela,
and to how the Bolivarian revolution started, seems to come
back to PDVSA and the worker oil lockout in 2002.
A little history. . .
Before the worker lockout, the administration of the state
oil company was strongly connected to the wealthy elite of
Venezuela. Many of the wealthiest people in Venezuela had
been getting much richer thanks to the oil company, in part
through contracts and corruption, not unlike what has been
happening here in the U.S. with politically connected com-
panies like Halliburton.
President Hugo Chavez was originally elected on a platform
to use the oil wealth to help pay for the poor of the coun-
try through education and health programs, rather than sim-
Hugo Chavez
64 Free Software Magazine Issue 10, January/February 2006
ply making the country’s wealthy even wealthier. Many of
Venezuela’s wealthier citizens, used to having money from
the state oil company, would not tolerate this, and so they
decided President Hugo Chavez had to go at any cost, even
if it meant sabotaging their own nation to do it.
So they tried to close the oil company in December of 2002,
by locking out the workers, holding the oil resources of the
nation as a whole hostage, and by having the entire IT in-
frastructure under their control. If the data and systems
present then had been destroyed, it would have been years
before another drop of oil could have been produced.
Out of 4800 managers, about 200 chose to stay behind, and
together, with the help of many by then retired former man-
agers who were less corrupt than the ones who left, the
workers tried to save the oil company. But the biggest chal-
lenge was the computer infrastructure.
Management of IT was at the time contracted to SAIC,
(Science Applications International Corp), which has well
known political and business connections to Cheney’s of-
fice, to the U.S. DOD, and the CIA. At first, when the
Venezuelan army was called out to secure the oil facili-
ties during the lockout, the SAIC staff created videos of
the troops securing the facilities in an attempt to claim they
were under attack and tried to persuade the U.S. congress
to give Bush war powers to seize the oil fields. When
this scheme failed, the SAIC workers fled the country, but
changed all the passwords and kept remote control of all
of the computer servers of PDVSA. They choose not to de-
stroy the data on them because they thought they’d be back
in a few months once the government of President Chavez
finally capitulated.
Much of the infrastructure of PDVSA was under Microsoft
Windows-based servers, and used proprietary database soft-
ware such as Microsoft SQL. The IT managers didn’t expect
a bunch of oil workers to be capable of thwarting their plans.
Those same oil workers, working together with local com-
puter hackers, were able to secure control of vital computer
servers, and in doing so saved the oil infrastructure.
The Venezuelan revolution is perhaps the first revolution in
history saved by computer hackers and this is one of the rea-
sons the government is so very strong on promoting the use
of free software, particularly in public administration. The
Venezuelan government wishes never again to have vital in-
frastructure held hostage or sabotaged by agents of foreign
nations. This cannot be accomplished by source secret pro-
 MIND SET
Conatel Telecenter
prietary software, such as Microsoft Windows, with its infa-
mous backdoor NSA key. Even proprietary software from a
trustworthy source has to be suspect for possible tampering,
and so must be rejected, not just by Venezuela, but by any
nation that wishes to protect and maintain its sovereignty
against sabotage.
Back to the present. . .
Everyone I had met from PDVSA appears completely
committed at all levels to the basic idea of converting
Venezuela’s oil resources into long-term and self-sustaining
wealth for the nation as a whole. This is done in part through
the development of a new economy, as planned for through
Minep.
Capturing this wealth is viewed as an urgent matter because,
even though Venezuela posses one of the largest known re-
serves of oil, they expect world oil production to begin de-
clining and see this wealth as very temporary. Socorro Her-
nendez said PDVSA believes that nobody will “burn” oil
(as for example in automobiles) in as little as 20 years. He
also said they believe that, while oil will remain important
to the multitude of other industries in which it is used, the
price will settle to $5 a barrel, so now is not only the best,
but also the last, chance to create something useful from this
wealth.
Conatel and Conclusions
I flew from Maracaibo to Caracas on November 26th. Even
in Venezuela’s revolutionary republic, custom officials are
still custom officials, and airports are still like airports ev-
erywhere. Given the lack of revolutionary posters, pictures
Free Software Magazine Issue 10, January/February 2006
of Chavez, or military checkpoints promised by the state
department, what is worth noting is the rather ordinary way
society and most institutions operate in Venezuela.
One interesting program is run by Conatel, Venezuela’s tele-
com regulatory agency, which now runs a program to deploy
telecenters into communities around the country. Conatel is
a regulatory agency for the telephone and broadcast services
in a manner akin to the FCC in the United States. However,
in this instance, Contatel also runs a community telecenter
project in order to bring computing and telephone resources
directly to communities across the nation. There are other
similar programs running in various Latin American nations
today.
I actually saw the model Venezuelian telecenter at the Cona-
tel building while I was in Caracas. A typical community
telecenter comes with up to a dozen PC workstations, and
a server. Connectivity is offered through a telecom carrier
for both internet data and for voice. These systems entirely
use free software, and each telecenter includes a staff of two
people.
One of the people is trained to manage and teach how to use
the computers and resources of the telecenter, and charged
with maintaining the equipment. The second person is
someone trained in the social needs of a given community.
For example, for a telecenter that is deployed in an agri-
cultural town, the second person would likely be someone
who was educated in agriculture. In a mining town, it would
likely be a miner.
Each telecenter desktop PC runs Debian GNU/Linux, and
includes software for internet browsing, for performing rou-
tine work such as Open Office, and includes a camera along
with GNOME Meeting for voice and video conferencing.
The telecenters also have VOIP telephones that are made in
China and that load an embedded Linux.
Many carriers in Venezuela offer direct H.323 connectiv-
ity for VOIP, and presumably, like Deutsch Telecom, more
than likely use GNU GateKeeper to form their mesh net-
work. The client workstations use GNOME Meeting, which
is an H.323 client, and even the telephone instruments use
H.323. No doubt, it would bring a tear to the eyes of Craig
Southern, who heads the OpenH323 project, to know that
there is a complete end-to-end national H.323 network in
Venezuela, running the OpenH323 project stack, from the
national carrier down to the individual telephone instru-
ments.
65
 MIND SET

I believe telecenters are or will be the public libraries of Copyright information

the new millennium. Unfortunately, most existing libraries
elsewhere in the world today, while often they include com- c 2005 David Sugar.
puters, don’t understand how they should be used. For ex- Verbatim copying and distribution of this entire article is
ample, many libraries in the U.S. have computers, but they permitted in any medium without royalty provided this no-
are really only used for web browsing, and come “attached” tice is preserved.
with nutty politicians more deeply concerned about library
patrons potentially reading about sex rather than about the
laws that require library content to be filtered for this reason.
All these things began with the oil worker lockout. Rather
than bringing down the government of Hugo Chavez, by
working together with foreign interests to directly sabo-
tage the country’s most vital industry, the wealthy elite of
About the author
Venezuela radicalized the oil workers in a way that no other
action could. The workers of PDVSA are now fully com- David Sugar is an active maintainer for a number of
mitted to creating the new economy, and will remain so re- packages that are part of the GNU project, including
gardless of who is in power. When the rich of Venezuela GNU Bayonne. He has served as the voluntary chair-
ponder who it was that made Venezuela become a revolu- man of the FSF’s DotGNU (http://www.dotgnu.
tionary nation, they shouldn’t look at President Chavez, who org) steering committee, as a founder and CTO for
may not have even been thinking of this at the time, and cer- Open Source Telecomm Corporation, and currently
tainly had no means to accomplish it if he had; instead, they owns and operates Tycho Softworks.
should look in the mirror.