See also: last week's Back page page.

Linux links of the week

Section Editor: Jon Corbet

Letters to the editor

Date: Mon, 17 May 1999 11:44:14 -0500 (CDT)
From: Roland Dreier <droland@mail.math.okstate.edu>
To: kbahey@ab2.com, editor@lwn.net, letters@lwn.net
Subject: Voice over IP for Linux

Hi,

I saw the letter to the editor in this week's LWN asking about "internet
telephone" programs for Linux.  I have actually been working on a program
I call gphone (short for gnome-o-phone) that does exactly that.  I've been
meaning to get gphone into shape to release, but my life has been very
hectic and I haven't gotten around to it.  The program definitely works
right now: it supports full duplex and uses GSM voice compression so
talking over a modem connection should be possible.  It has a rudimentary
GTK interface and I'm planning on adding GNOME support.  The license is
GPL. 

However, the code is best described as pre-alpha.  There's a lot of stuff
that I know needs to be done.  But I'd be delighted to share what I have
with anyone who's willing to work with code in development.  And I'll let
you know when I have a release that's ready for users (ie that only
requires a ./configure && make install).

Best,
Roland Dreier
droland@math.okstate.edu

To: letters@lwn.net, Piotr Mitros <pmitros@MIT.EDU>
Subject: RedHat & xv
Date: Sun, 16 May 1999 09:20:07 -0700
From: "Zow" Terry Brugger <zow@torii.bruggerink.com>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


	Quick response to M. Mitros's note on the exclusion of xv from
RH6. While his point is well taken, I think it has more to do with the
inclusion of ee (Electric Eyes), which has all the same functionality as xv
(although it does have a few flaws that I haven't had time to
investigate). ee is written by Mr.  Enlightenment and RH Labs developer,
RasterMan. It appears that the non-free status was what prompted the
original development of ee, hence leading to its eventual extinction from
RH6.  Disclaimer: I'm only an end-user and this info's just gleaned from
what I've seen on the web.

- -- 
"Zow" Terry Brugger
zow@acm.org                 http://www.cs.purdue.edu/homes/bruggest
	"Information is easy; Tapping at my PC, that is the 
		frame of the game." - PetShopBoys



-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 5.0i for non-commercial use
Charset: noconv

iQA/AwUBNz7iJ6fuGVwXgOQkEQLpwQCgtSPezOHuPbczk+P2W9B+Y1XKvjEAoNdE
yCFXPf3ALOcEMbOUaijFI8f8
=x/pP
-----END PGP SIGNATURE-----

Date: Thu, 13 May 1999 21:26:05 -0500 (CDT)
From: John Morris <jmorris@odin.beau.lib.la.us>
To: letters@lwn.net
Subject: The Bake-Off idea

I liked the basic idea, but the proposal lacked one thing; leveraging
Linux's most important advantage.  We need to put price into the equation.
How about this:

Have an independent group list a set of jobs to be done and budgets for
each.  Get a major Linux friendly vendor (Dell would be a good candidate
right now) to donate use of their equipment.  Each team picks a server or
servers from that vendor's line, loads and equips it and then ensures it
meets the performance requirements.

Suggested jobs would be things on the order of, Build a fileserver capable
of servicing 500 users.  This will be tested by running the X benchmark on
N clients to simulate the load of 500 typical users.  Or build a webserver
capable of soaking a T-3 using static pages.  Now soak a T-3 with a
Slashdot type dynamic site. Make the tests pass/fail, either it delivers
the specified load handling or it doesn't.  Of course the real benchmark
becomes how far under budget each side brings in their project.  This sort
of benchmark would have real interest among the bean counters.

On the other hand the idea of handing each team an identical box with a
virgin hard drive and seeing which team can cook up a ready to deploy
solution in a fixed time is also appealing.  However even there the cost
factor should be hammered home by making both sides total up the cost of
the software used to build the solution and guesstimate the labor costs by
counting the number of manhours both sides use.

John M.      http://www.dtx.net/~jmorris         This post is 100% M$ Free!
Geek code 3.0:GCS C+++ UL++++$ P++ L+++ W+ N++ w--- Y+ 5+++ R tv- b++ e* r%
=========================================================+=================
#!/bin/perl -sp0777i<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<j]dsj|RSA in Perl
$/=unpack('H*',$_);$_=`echo 16dio\U$k"SK$/SM$n\EsN0p[lN*1|Using this sig is
lK[d2%Sa2/d0$^Ixp"|dc`;s/\W//g;$_=pack('H*',/((..)*)$/)  |a federal crime!

Date: Fri, 14 May 1999 15:46:03 +0100
From: Aaron.Trevena@msasglobal.com
Subject: Byte to host L/Apache NT/IIS benchmark
To: pschind@cmp.com 

     Paul,
     
     I have been reading Byte since I was in College, and enjoyed it while 
     it focused on design and solutions rather than towards the end of its 
     hard copy life where it became just another glossy marketing tool.
     
     The core section helped with my coursework, the programming provided 
     ideas, I was chuffed when I began to understand how technology was 
     used in business, I even had Byte articles as references on CORBA when 
     study distributed processing.
     
     Unfortunately most of the IT press now with the exception of Dr Dobbs, 
     and hopefully Byte in its new format (Haven't seen enough to judge) 
     will continue to provide a reliable independent source of information 
     for academics and professionals. With the dumbing down brought by the 
     skills crises, and lowering of bar to be able to produce 
     (unfortunately usually poor) products Independent and Professional 
     publications become even more valuable.
     
     With this in mind I ask that Byte be the venue for the benchmark 
     Microsoft/Mindcraft (same office, same apparent marketing budget, same 
     labs) are pushing.
     
     ZiffDavis although recently becoming independent of MS have suddenly 
     fallen back into line. Although I trust they wouldn't rig or foul up 
     benchmarking in the way mindcraft did, they would be happy to show NT 
     in good light and brush problems under the carpet in small side notes.
     
     I trust Byte to independent and objective enough, I know it is 
     respected by most of the Open source community. It would also be good 
     for the industry if the boundaries of the benchmark such as the 
     hardware, and the competitors were set by Byte rather than MS as they 
     currently are so as to provide a fair and more importantly USEFUL 
     result. FreeBSD, BSDI, Solaris and Zeus, thttp should be included.
     
     If a fair and useful test were done then I personally would benefit by 
     being able to know when to deploy which platform and server. Open 
     Source developers would benefit by knowing where realworld (or as 
     close as benchmarks get) problems occur rather than trying to fix 
     problems that only occur in benchmarks or in contrived settings.
     
     I have cc:ed this to Linux Weekly News. Who I am sure would agree, 
     along with most of the Linux press that such a test would be very 
     useful to all concerned.
     
     Obviously it would be less exciting to do it well and properly than to 
     have a hyped shoot out but obviously there would be many people very 
     interested to see it done properly, not least Sun, HP, Oracle, and IBM 
     who have put their money where there mouths are.
     
     Regards,
     Aaron Trevena. Inter/Intranet Developer & Administrator.
     MSAS Global Logistics, Group IT.

Date: Sat, 15 May 1999 17:27:26 +0000
From: Yoni Elhanani <biggo@netvision.net.il>
To: letters@lwn.net
Subject: Mindcraft is only one kind of test....

Dear editor,

We all know that the benchmarks that test very specific conditions are
useless.
We know that many linux servers are not run on quad xeons, nor serve so
many clients.

I'd like to suggest my own kind of test to battle against Mindcraft.
Ofcourse everyone knows there are several kinds of races,
The one i where time is constant and performance is tested (eg pie
eating contest), such as these benchmarks,
the other is where a task is the constant and time is tested (eg horse
racing).
I'd like to propose the other test.
I think an uptime test would show the power of linux.

Another kind of test is a peak test (ie spitting contest), which in our
case would be scalability test.
I want to see NT running on a 486.

And there are feature tests, such as syncronized swimming.
Now we'll see all the features apache and php has to offer!

Let's see NT trying to outperform Linux in these fields... :-)

Cheers,
Yoni.

-- 

The day Microsoft makes something that doesn't suck,
is probably the day they start making vacuum cleaners...

From: schwarzma@healthpartners.com (Michael Schwarz)
Subject: RSA keys and PGP
To: Jens.Ritter@weh.rwth-aachen.de
Date: Thu, 13 May 1999 16:52:53 -0500 (CDT)

-----BEGIN PGP SIGNED MESSAGE-----

This is in response to Jens Ritter's letter to LWN
about PGP and RSA:

It is a minor point, but...

What you say about RSA is true, however you must
remember that in PGP a 1024-bit RSA key is used to
encrypt a 128-bit IDEA session symmetric key.  Anyone
attempting to crack a PGP message will not bother
trying to recover the 1024-bit RSA key used to encrypt
the IDEA session key, they will instead concentrate
on cracking the 128-bit key.  Fortunately, cracking
128-bit keys will be take decades, even with Dr. Shamir's
TWINKLE machine...

Mike Schwarz
IAM mschwarz AT sherbtel DOT net
(anti spammed e-mail address)


-----BEGIN PGP SIGNATURE-----
Version: 2.6.3a
Charset: noconv

iQCVAwUBNztJpElby0f+F8BNAQG+lwP/fpNWF2qt7JwmaoHk1pCg4cQ4wX5RQZO+
VXGjFpIDfX8HXAMaf3c/t9TfvcYu5vgCzyu4V1pJaqDaANpyovtL8mFbx/DH1vWc
UfMizcVySfTyQtVJxGaSBfKswm7Eg0/U0CwJ3pihUvHYpAyl30EtVbtpgMLdihyx
gUQOBbXJJ3I=
=PQjr
-----END PGP SIGNATURE-----

From: Leandro Dutra <Leandro.dutra@globaltelecom.com.br>
To: "'drwho@xnet.com'" <drwho@xnet.com>
Subject: Restrictively Unrestrictive
Date: Thu, 13 May 1999 11:59:26 -0300

	Re: your article at http://www.daemonnews.org/199905/gpl.html
	I won't try to refute any little point in your article first, but
your general, US "culture" conditioned, strictly "capitalist" viewpoint.

	It is easy, but not very honest, to call RMS a communist.  It shows
you no nothing about Communism, and little about RMS intentions.

	First of all, let me define a word you from US misuse: Liberalism.
Liberalism is an European concept about freedom to live under the laws,
freedom to enterprise, and keeping the state interference in private and
economical issues at a minimum.  What you from US call "Liberalism" is
called Leftism all over the world.

	Communism has two faces: one is the ideal, unrealistic goal of
communality, the idea that everything should belong to everyone.  You can
subscribe to this desire while still believing  Liberalism better for
practical reasons.  It's only if you think Leftism it the true way that you
are being a Communist proper, believing everyone should refrain from gain
and share all their goods and money.

	There is no evidence that RMS is a Communist, nor does GNU GPL favor
Communism in any way.  It is just a defense against software hoarders, which
almost ruined BSD if you remember History.  GNU GPL is to free software what
an Army is to a pacific country: only a way to make sure the country stays
free and peaceful and independent, not an end in itself.

	When you say things such as "the General Public License is not so
much about ``keeping free software free'' as it is about forcing us to
accept the extreme Communistic political philosophy of Richard Stallman and
others at the Free Software Foundation. The very spirit of the GPL is to
attack the very concept of Capitalism and individualism. There is no concept
of intellectual property under the terms of the GPL. Your hard work is no
more your property than everyone else's.", you get to misunderstood
everything.  This phrase I want to refute point by point.

	First, it is easy to call someone Communistic.  I've shown you they
aren't.  Political?  Yes, you are also.  It is just that your politics is
mainstream, our is marginal.  This shouldn't be a term of abuse.  Extreme?
This carries no meaning, unless you really meant "radical".  Radical yes,
because GNU goes to the root of the problem ensuring the continuation of
freedom.  In this sense, BSD is lukewarm, because it counts on the good will
of everyone to keep freedom against hoarding.

	Second, no one is forcing no one.  You can do it all by yourself
with BSD software, with GNU software, or with proprietary.  Just do not
hoard GNU software.  If you want to keep others away from the freedom of
having source, you won't be able to use GNU software, why is that such a
problem?  It's like some kinds of virus, if you do not want to be sick do
not go near the source of the virus!

	Third, what do you understand by Capitalism and individualism?  Do
you know that Capitalism is a term of abuse invented by Karl Marx?  The
bigger values in Western society shouldn't be capital, richness, but freedom
and justice.  It is freedom and justice that ensured the conditions to
capital accumulation and richness.  But to put richness before freedom and
justice can kill the chicken of the golden eggs.

	Also, individualism is not a value, it was created as just the idea
that the individual should have defense against the state or any other
collective power.  In this original sense the GNU GPL could be considered
very individualist, as it protects the individual free software developer
and user against corporate or government software hoarding.

	About intellectual property, it is just a concession in the former
of copyright or patent so as to encourage invention, it isn't a fundamental
concept of law or ethics, and it can certainly be abused.  I find the
concept to be widely misunderstood.  It is astonishing to see that just now
that richness in the Western world is so great, that the periods for
copyright and patent proctection are being enlarged.  That is hoarding for
sure, as the original idea of intellectual property should call for a
shortening of the proctection periods as richness grows, for public benefit.

	Finally, your work belongs to you and there is nothing in GNU GPL
against your owning it.  It just ensures you that if you create a program
for free use, it will remain free.  It is still your code, so that if
someone wants to use it in a proprietary way he will have to license it
directly from you, and then you can set your price for that alternate,
proprietary licensing.

	It is this possibility of dual licensing that shows how your idea of
Capitalism and Communism is weak.  In fact the GNU GPL does more to protect
the gains of the programmer than BSD.

	Please be more careful, read more, think more.


Leandro Guimarães Faria Corcete Dutra
Amdocs (Brasil) Ltda

Date: Fri, 14 May 1999 10:51:15 +0800 (CST)
From: Hung(2) Chao(2)-Kuei(4) <ckhung@cyut.edu.tw>
To: drwho@xnet.com
Subject: BSD Licence vs GPL

Dear Mr. Maxwell,

I found your article
    http://www.daemonnews.org/199905/gpl.html
    "Restrictively Unrestrictive: The GPL License in Software Development"
misleading and likely to do harm rather than service to the open source
community.

To understand GPL, one has to understand the "political opinion" from
GPL which you omitted in your article.

Yes, Richard Stallman is against software intellectual property. I bet Mr.
Stallman would be happy if one day he wakes up to a world where software
copyright laws -- something GPL's enforcement depends on -- does not exist.
Whether you agree with his opinion (and the opinion of many people in the
open source community) or not, let us understand his point: software
should not be copyrighted.

So why did he (cooperatively?) created GPL that seems so much more
restrictive than the FreeBSD style license, and seems so much more
dependent on the copyright law?

I bet you heard "proof by contradiction" in mathematics. (OK, I know
there is a Latin phrase for it but I don't remember.) You don't agree
with a statement X. You _assume_ that X were true. You base your
arguments on X, and come to a contradictory conclusion. Your arguments
are all fine and logical. The only explanation left is that the
statement X was indeed absurd to begin with.

Now if someone jumps into the middle of the proof without understanding
the absurdity of X, he is bound to view the statements in the proof
as totally incomprehensible or at best "confusing". He is bound to see,
of course, absurd intermediate statements. It would not be wise to
claim that the flow of proof is incorrect when in fact it is X that is
absurd.

I see the open source software movement as a time-consuming proof of
the absurdity of software intellectual property, initiated by FSF's
GPL by way of contradiction. Please understand that, and you will see
the mathematical beauty in GPL. By the way, I would use the term
"recursive" instead where you used "infective". I personally see
GPL as a piece of work to be appreciated by the mathematically-minded.

Of course, you may not agree with Mr. Stallman's opinion that software
copyright should not exist. Whether Stallman is correct is not the
topic of this article. I am just trying to explain how the "proof
mechanism" works.

On the other hand, I don't see how anti-software-copyright implies
anti-competitiveness and anti-capitalism. You seemed to jump into
unrelated and far-fetched conclusions that seriously begs explanation.

Sincerely,

Chao-Kuei Hung
http://www.cyut.edu.tw/~ckhung/

Information Management Department
Chaoyang University of Technology

#include <std.disclaimer>

Date: 17 May 99 12:25:01 PDT
From: Drew Davis <drew3@netscape.net>
To: letters@lwn.net
Subject: "Red Hat goes upscale"??

Your May 6, 1999 "Commerce" page leads with an article "Red Hat goes
upmarket" claiming that RedHat has increased the price of their
distribution.  That same assertion shows up in:

        http://computers.miningco.com/library/weekly/aa042699.htm

whose author acts like the price of RedHat 6.0 is $100.

Reading www.redhat.com's info, it looks to me that the $99.95
option is for a box that has Red Hat 6.0 Linux, plus a bunch of
other CD's, including their "Linux Power Tools" product and a
CD full of a ton of Linux documentation.

I believe the software is still free if you're willing to take the
time to download it.  That is, I think you can download, already
compiled and ready to install, all the Red Hat 6.0 software.
(And source code too, if you like).  But I know I don't have the
patience for doing that much downloading.

$39.95 gets you the 2 CD's of Red Hat 6.0, but no boot-floppy and no
support.  I'm told that it is easy to make a boot-floppy if you have
an already working PC.

$79.95 gets you the 2 CD's of Red Hat 6.0, and 30 days telephone
support.  (I assume this edition includes the boot-floppy too).

Seems like a fair spectrum of choices to me, and with a choice of
buying a bundle that includes telephone support, or getting RedHat 6.0
for less money than I paid for RedHat 5.2 (which was $50)..

I don't work for Red Hat, but I think there's a different story here
than a "price increase".  "RedHat broadens product line" would be a
more accurate description of what they apparently are doing.

R. Drew Davis
drew3@netscape.net


____________________________________________________________________
Get your own FREE, personal Netscape WebMail account today at http://webm=
ail.netscape.com.