[LWN Logo]

 Main page
 Linux in the news
 Back page
All in one big page

See also: last week's Back page page.

Linux links of the week

SSC has put up a new GIMP site at thegimp.com. It contains parts of Michael Hammel's book, tutorials, etc.

And once you get good at the GIMP, you can try to put together something for themes.org...

Section Editor: Jon Corbet

March 18, 1999



Letters to the editor

Letters to the editor should be sent to editor@lwn.net. Preference will be given to letters which are short, to the point, and well written. If you want your email address "anti-spammed" in some way please be sure to let us know. We do not have a policy against anonymous letters, but we will be reluctant to include them.
Date: Wed, 17 Mar 1999 11:56:40 -0500 (EST)
From: Conrad Sanderson <conrad@hive.me.gu.edu.au>
To: lwn@lwn.net, editor@lwn.net
Subject: clearer definitions of "free software"

It's about time we should, as a community, have clearer definitions
of what free software is.  We know what it is, but there is a lot of
confusion to the outsiders and newcomers, as well as the millions 
of clue-less article writers, who for some strange reason use the term
"shareware", or only get the "free beer" aspect.

I propose we start using the clearer French (?) versions of free 
(no wonder it's the preferred diplomatic language) :

Libre = "liberated" -> free to modify source code
Gratis = "free beer"

Conrad Sanderson - Microelectronic Signal Processing Laboratory
Griffith University, Queensland, Australia

Date: Mon, 15 Mar 1999 10:14:55 +1000
From: geishan <geishan@ozemail.com.au>
To: editor@lwn.net
Subject: Free software.

I'm an small independent software developer. The company consists of
me.  That's all just me.  My work provides for my family. I create
proprietary software.  I am therefore the enemy in some peoples eyes.

I like Linux. I like the open source. Hell I'll proberly even
contribute to Linux.  But to have all software free?? That sucks.

The whole problem is one of levels. and degree.  At the bottom layer
is the OS, the kernel, then libraries and shells that make use of it,
then tools that in turn use the libraries. Now at this level there is
a really thick black line.  Because above this are applications.  I'll
contribute below the line, after all I, and my applications, can
benefit from this by improving stability and so on.  Above the line is
my eating and beer money. My livelihood.  Opening this up will stop
the beer, and food.

In short. The free and proprietary models are NOT exclusive, they
complement each other.  The free/open source just occupies a level
below the not free.

human nature: nothing for nothing - something for something.  I can
gain via contributions to open source at the low level and lose at the
high level.

Food for thought

Sean Hennessy

Date: Sat, 13 Mar 1999 22:58:36 -0500
From: "Jay R. Ashworth" <jra@baylink.com>
To: editor@lwn.net
Subject: Who's hand is on the tiller?

In this weeks' LWN, you point to two Larry McVoy pieces from the kernel
mailing list on the topic of what should be done about architectural
debates concerning the Linux kernel, and whether or not such things
should merely be left to Linus, or also debated in public.

The consensus appears to be both, and this ought not to be much of a
surprise to anyone who follows that list.

However, there's a subtle point that I think is being missed here.

Probably _the_ fundamental reason that Linux has lasted as long as it
has, works as well as it does, is as popular as it is... and is in only
release 2.2 after _ten years_ is because there is _one_ hand on the
tiller: Linus'.

Yeah, he gets lots of help from hundreds of people, and I'm not remotely
trying to downplay their contributions -- especially since I'm not one
of them.  But in the end, the person who decides what does _not_ go in
the kernel (as Larry does note) is Linus. 

No one else.

Fred Brooks, author of the seminal work in programming project
management, _The_Mythical_Man_Month_ (which you really should read if
you haven't already) makes this point very clearly in his book: any
given system, if it is expected to work well and survive for a long
time, can tolerate one, or _maybe_ two, hands actively exercising
control over it's architecture.  No more.

One is best, for a job one can do -- and apparently this is such a job.

Of course, things will get interesting when Linus decides to retire.  I
can't see any way to avoid a fork at that point, but then, I don't
follow the kernel list.  Perhaps the architect apparent is already
apparent and I simply don't realize it.

In any event, while it's not necessarily a pleasant thing to think
about, The Linux Community<tm> would do well to give some consideration
to the thing that kills most family business by it's lack: succession
planning.  (Read: "Hey, Linus! You figure out who gets the reins, yet?"

Architecture isn't easy.  If it was, architects wouldn't get paid so

-- jr 'will stir up firestorms for food' a
Jay R. Ashworth
Ashworth & Associates
Date: Fri, 12 Mar 1999 17:50:12 -0500 (EST)
From: "Ben 'The Con Man' Kahn" <xkahn@cybersites.com>
To: Craig Goodrich <craig@airnet.net>, John Kodis <kodis@jagunet.com>,
Subject: Who can you sue?

	The latest attack on Linux is the "Who do you sue?" question.
Taking the question at face value misses the point.  I don't believe that
the people asking this question actually imagine that they can sue other
software companies.  They are asking a different question.

	They want to know who is responsible -- who they can lean on when
things go wrong.  You see, large corporations like to have a process they
can reliably follow when things go wrong.  

	When they buy software from a large company, they know who they
can lean on -- that large software company.  They don't really believe
that they can effectively sue this software company, but they like to know
they can put pressure on the company to fix the problem.  (By refusing to
buy their products in the future, causing bad publicity, etc.)

	Because of this, businesses would rather deal with other
businesses.  The system, for the most part, works.  And looked at from
this perspective, free software (err...  Open Source(tm), sorry!) is
scary. It isn't always clear who can be blamed.

	Of course, Open Source(tm) software works in a different way.
Because the code is available, blame isn't an issue.  Instead of focusing
on who to blame, you get to concentrate on who can fix it.  And the answer
to that question is: anyone can fix it!

	Once companies grasp this concept, Open Source becomes far more
accepted.  I've seen it happen.


------------------------------------ |\      _,,,--,,_  ,) ----------
Benjamin Kahn                        /,`.-'`'   -,  ;-;;'
(212) 924 - 2220                    |,4-  ) )-,_ ) /\
ben@cybersites.com --------------- '---''(_/--' (_/-' ---------------
	Drawing on my fine command of language, I said nothing.

From: "F.Baube(tm)" <fred@rodan.moremagic.com>
Subject: RMS 2020
To: editor@lwn.net
Date: Fri, 12 Mar 1999 14:42:52 +0200 (EET)

At this point in time, when licensing models are in a state
of flux, the role of RM Stallman is absolutely essential.

Right now, the Internet is transitioning from being merely 
huge to being stupefyingly all-pervasive, and this creates 
ever more pressure for the commercialisation of software.
But contemporaneous with this, licensing models are being 
devised and settled upon that will live on into the future.  

An absolutist is required to press home the point that no 
quarter can be yielded on the fundamental issue of citizen 
access to essential tools.  The programmers of ten and 
twenty years in the future must continue to have basic GPL 
tools available, lest the craft of programming be relegated 
to those who can pony up the price, whatever that future 
price might be.  Society's cashless and cash-shy must not 
be denied tools.  Let RMS remind the community of this.

I speak from experience.  I upgraded my skills during 
a bout of unemployment using GNU/Linux.  (And also the 
Java SDK, but that's a whole 'nother dispute ...  :-)

Best regards,

Fred Baube


F.Baube(tm)        * "The record labels are middlemen.
G'town U. MSFS '88 *  In the age of the Net, middlemen 
fred@moremagic.com *  are roadkill.  Let's kick out the jams ..."
+358 (40) 737 6934 *    -- music attorney Ashwood Kavanna
#include <std_disclaimer.h> 

Date: Thu, 11 Mar 1999 12:16:06 +0100 (MET)
From: David Kastrup <dak@neuroinformatik.ruhr-uni-bochum.de>
To: editor@lwn.net
Subject: Commercial support of Linux

In the last few years, corporateinterest in Linux is all the rage.  I
want to point out a few things in connection with that.

Linux is constituted of free software.  Only what has been released as
free software (redistributable with source) can find its way into
every distribution and become a part of Linux.  The rest can stay an
optional add-on at most.  If a corporation loses interest, this add-on
will no longer be available for newer kernel/library/processor.  It is
not a value added to Linux, as it can be taken away again.  It is
lustre, not substance.  Of course, lustre can be important too, and
sometimes needed, but it's the bones that count.

Some people think that proprietary software is good enough.  If this
stance would have been prevalent always, there would be no Linux.
There would be no GNU software, either.  This software is available
precisely because some people cared about free software, and it will
stay available beceuase of that.  Since no good free software gets
lost or buried, the free software pool can only improve.  This is what
has made Linux great mainly from the work of private contributors.

Now companies start getting involved with Linux because it is a good
platform, and on the rise.  They don't care about freedom of software
per se, they are trying to ride the waves.  What is in it for them?
Money.  What is in it for us?  Depends on what the company does.  We
have several forms of involvements.

One is the offer of proprietary software under Linux, like the Oracle
database.  This buys Linux nothing except more employment and a bit of
advertising.  It adds nothing to the value of Linux, it makes use of
the value of Linux.  It can be used for pointing out that Linux is a
viable platform for proprietary software too, but that's about it.

We have companies like Creative that start developing drivers for
their hardware without wanting to release them in the source.  They
are strictly adding value to their products, not to Linux.  You can
use such stuff only at their whim on platforms they choose to support,
on products they choose to support.  Once they think their interest
insupporting a certain board is over, you might no longer be able to
get it working with newer systems.  Limited lifetime.

Then we have contributors like Compaq/Digital/Sun, IBM, SGI and Intel
that support or do porting and development work on basic free Linux
code for hardware they are producing.  The results of that efforts are
available freely, thus they add to Linux (and potentially other free
systems) permanently.  If they chose to discontinue their support of
certain hardware, whoever wants to can take this up.  The fate of this
software is not at the absolute whim of the contributing companies
anymore.  This is a true and permanent contribution and thus is of an
entirely different quality, even though it is primarily intended for
pushing certain hardware.  This still makes excellent business sense.

And then we have contributors like SGI with the contribution of its
GLX code, IBM with contributions in Apache development, RedHat with
the contribution of GNOME development.  This is code that can be made
to benefit *every* Linux system even from competitors.  It is code
that advances the state of art of Linux for everyone.  It helps
everyone, establishing the contributor as a technology spearhead
instead of merely a small claims technology dealer.  It is
contributions like that that *really* advance Linux significantly and
permanently.  Since a widely accepted standard operating system freely
available is very desirable for companies, this again makes business
sense, but it makes long-term business sense and benefits even those
that do not play by the rules.

Involvements like that are of the highest quality and value for the
advance of Linux.  I find it pretty sad that corporate involvement
with Linux gets all-hailed currently without much differentiation.
All people, please take care to pinpoint always which class a Linux
involvement of business is in when reporting about it.  The true
contributors deserve your respect, your acknowledgments, your public
support and your praise in proportion to their contribution.

If we indiscriminately applaud everyone regardless of their true
accomplishments, we discourage real contributions, real investments,
real dedication.  So please grade the amount of enthusiasm you show
for various achievements.  We currently have enough to start getting
a bit less indiscriminate.

David Kastrup                                     Phone: +49-234-700-5570
Email: dak@neuroinformatik.ruhr-uni-bochum.de       Fax: +49-234-709-4209
Institut für Neuroinformatik, Universitätsstr. 150, 44780 Bochum, Germany
Date: Tue, 16 Mar 1999 21:41:15 +0800
From: Tom Atkinson <tom@tyco.net.au>
To: editor@lwn.net
Subject: What Linux needs next

It seems that the world is crying out for an escape 
from Microsoft but Bill Gates knows that he will 
continue to dominate the desktop while Linux/Unix
configuration can be done only by the highly skilled.

It is generally agreed that for Linux to make inroads
"on the desktop", it needs a unified and consistent
configuration/administration system, which can be
accessed via either a graphical or text-based frontend.

If ALL the system and application configuration could 
be done through this system, then Linux would be as 
"useable" as Windows95 for 100% of the world's computer
users.  At present, only about 5% of users would 
consider Linux to be as useable as Win95.

By "useable", I am talking about the ease with which 
the user can configure, or reconfigure, the way
applications work, be they "system" apps such as CRON,
or "user" apps such as KBIFF.

With useability equal to Win95, ports (fully supported
ones at that) of commercial software such as Photoshop
become a considerably more worthwhile business

If the configuration system I am proposing is the
missing ingredient to desktop success, then why hasn't
it been done?

As you are probably aware, there have been a number of
attempts at "configuration/administration suites" for 
Unix and Linux over the years.

Most of the proprietary unixes have their own, 
graphical administration tools.  Works in progress for
Linux are COAS and Linuxconf, plus, I presume, a tool 
for each of the distributions.  (A RedHat distribution
I once used had a tool called "Glint").

None of these attempts have been terribly successful,
which is not surprising, given that they keep making
the same mistake, over and over!

In Linux Journal issue 58, Olaf Kirch, talking about
COAS, ruefully admits that "updating and maintaining
configuration software for Sendmail is almost a full 
time job for a programmer".  Clearly, the wrong
approach is being followed here.

COAS and the others are a step in the right direction
but are doomed to failure because they simply are not
"doing it right".  As each of the current crop of
suites dies through lack of adoption, sadly, others
will spring up, in their place, repeating the same
mistake.  Forgive my pessimism, but this has been the
pattern over the last several years.  Remember the 
"dotfile generator"?


You make big problems for yourself when you create an
admin system which operates on the ASCII config files.
The programming required, in order to support each 
different config file, is significant.  This is 
because the programming needs to be very "defensive", 
so as to allow sysadmins to continue to use vi on
the file.  The programming work required, considering
the number of different applications you wish to
support, is simply too large.  This is the simple 
reason why ALL attempts in this area have thus far 
been failures.

In place of the hundreds of existing config files 
needs to be one database that holds all of the 
configuration info.  (Or should the system consist of
system-wide, and per-user databases?).


This means that all the applications have to be 
modified but this is the only realistic way!  It may 
seem like hard work at first glance, but just think 
of how much work has already gone into those failed 
admin tool attempts.  Think also of how the internet 
and the open source ideal make the job much more 
possible today than 5-10 years ago.

Hopefully, it should only be necessary to get the 
major application authors onside (Sendmail, Samba, 
Apache, etc), then the ball would be rolling.  
Ironically, life would actually be a little easier for
these authors, because less code would be required in
their application to read and write config data.

I do not wish to get into the technicalities of the
database layout and interface - that is a task for 
those more expert at this subject than I.

If the major apps went to such a system, it would also
cause a force for further unification of all unixes,
because they would have to adopt the system too.

Eventually, this dream of all unixes would be realised
- the graphical, bulletproof, configuration suite that
configures ALL software, not just /etc/passwd and 

A distribution of Linux that is as useable to the "mums
and dads" as Windows 95 would suddenly be possible.

Tom Atkinson
(tom AT tyco.net.au)


Eklektix, Inc. Linux powered! Copyright © 1999 Eklektix, Inc., all rights reserved
Linux ® is a registered trademark of Linus Torvalds