See also: last week's Back page page.

Linux links of the week

The Free Software Jobs Page is maintained by the Free Software Foundation, so only jobs which involve working on free software will be listed.

LinuxToday also has a jobs page with slightly looser criteria (and correspondingly more listings).

Section Editor: Jon Corbet

Letters to the editor

Choosing letters to publish was unusually hard this week, due to the large number of them that we received. A number of quality letters were left out. We can only apologize to the senders, and encourage you to write to us again when the inspiration strikes you.

Date: 6 May 1999 21:25:46 -0000
From: bruce@perens.com
To: lwn@lwn.net, malda@slashdot.org, nicholas@petreley.com,
Subject: Bake-off proposal

Instead of griping over this mindcraft benchmark, let's hold a bake-off at
LinuxWorld in August.

We will challenge NT and Linux developers to each field teams to compete
at LinuxWorld. The two teams get identical hardware provided by a
non-competing third-party: Dell, IBM, whoever. The NT team and the Linux
team have a day or two to bootstrap their systems and tune them. Then they
compete on a number of issues. Besides how fast they can serve, etc., let's
make rapid application development one of the issues too. Each team could
be handed a list of tasks to develop using their respective environments,
and could be judged on time to completion, bugginess, features and elegance,
speed, etc. Cover it with live webcasts, etc.

It's sort of like Deep Blue vs. Kasparov. You win on the publicity front
even if you lose.

Pass this on to anyone you like if you think it's interesting.

	Thanks

	Bruce Perens

Date: Fri, 7 May 1999 15:02:32 -0400 (EDT)
From: Ray Jones <rjones@pobox.com>
To: bruce@perens.com, lwn@lwn.net, malda@slashdot.org, nicholas@petreley.com
Subject: Re: Bake-off proposal

> [Bake-off suggestion by Bruce elided...]

There are two things I like about Linux compared to other OSes (and
the accompanying software):

1- high quality
2- low hype

A bake-off doesn't do much for either of these.  Yeah, Mindcraft's
benchmark was pretty useless and made us all angry, but it doesn't
really matter to Linux's continued success.  The results have been
shot down sufficiently well (by people outside the Linux community!)
that we should just be ignoring them and working on making Linux
better, so the next time someone like Mindcraft wants to put out a
benchmark that makes NT or whatever else look better/faster than
Linux, they won't be able to.[1]

IMO, a bake-off as described by Bruce would produce less good than an
actual bake-off with the food being sold and the money being donated
to the FSF.

Ray Jones


[1] I realize that's impossible.  But it can be made more difficult,
and tuning information can be collected and made easily available so
that the "no documentation" argument falls apart.

Date: Fri, 7 May 1999 01:09:44 +0200 (METDST)
From: Morten Welinder <terra@diku.dk>
To: letters@lwn.net
Subject: Attribution


In the May 6 issue, I think you are misrepresenting the quote

   "While the NT Testanlage an intensive Tuning experienced by
    Microsoft specialists, at the Linux version one did not screw."

I am quite sure that this quote should be attributed to Yoda the Jedi,
not Babelfish.

Morten

Date: Thu, 6 May 1999 08:20:03 +0200 (CEST)
From: Jens Ritter <jens@hilbert.weh.rwth-aachen.de>
To: letters@lwn.net
Subject: RSA not obsoleted


Dear Editors,

I guess you confused DES (Data Encryption Standard) and RSA in the current
security section (6th may issue). It is true that DES features only 56 bit
key length and that it is obsoleted by the the eff's descracker machine.
See:  http://www.eff.org/pub/Privacy/Crypto_misc/DESCracker/ 
So the current announcement of Shamir only puts another nail in the
already sealed and buried coffing of DES. 

RSA in contrast is the public key encryption scheme used in pgp as default
method up to version 2.6.3. It uses the product of two large prime numbers
and the security is based upon the fact that factorising such a product is
a difficult (or hard) problem. While it is clear that to factor the
product of two prime numbers with say 3 digits is a matter of seconds on
current hardware, the problem becomes difficult if one uses products in
the range of 1024 and more binary digits.

The result of Shamir only pushes the number of digits up, if one wants to
get a key which can be considered secure for some years.  As Bruce
Schneier in his book "Applied cryptography" notes, if you want to have a
long lasting RSA key, you have to use as many bits as possible. 

Yours sincerely,

Jens Ritter  

P.S.: Please vote against Spam! At
             http://www.politik-digital.de/spam/
(Sorry Europeans only)
---
Jens.Ritter@weh.rwth-aachen.de   grimaldi@debian.org
Key ID: 2048/E451C639 Jens Ritter
Key fingerprint: 5F 3D 43 1E 24 1E CC 48  1E 05 93 3A A7 10 73 37 

Date: Thu, 06 May 1999 03:47:52 -0400
From: Piotr Mitros <pmitros@MIT.EDU>
To: letters@lwn.net
Subject: xv

Quick comment on Red Hat 6.0 and xv. I would guess that the reason that
Red Hat left it out is because xv is not free software. It comes with
source, but it is distinctly shareware. Red Hat is phasing itself over
to be a pure free software company. I think the good of having a
completely free OS outweigh the disadvantages of loosing a little
utility like xv. Xv is convenient, but it's not that big a deal.

Date: Fri, 7 May 1999 15:40:09 -0400
From: ajc@uncnsrd.mt-kisco.ny.us (Art Cancro)
To: letters@lwn.net
Subject: wxWindows, Qt, and other cross-platform toolkits

Dear Editor,
 
   I came home from a vacation this week to discover that the latest "war"
in the open source universe appears to be wxWindows vs. Qt.  It's ironic
that nearly every time another debate surfaces in our community, Troll Tech
AS seems to be involved, although that's not what I intend to discuss today.
 
   I recently began a non-trivial development project that requires a
cross-platform toolkit, and I went through the same search that Matt Heck
and many others did.  I have to say that Mr. Heck's conclusion is a good
one: wxWindows and Qt are both very good toolkits, and you can't make a bad
choice.
 
   Yes, I do have a preference.  I chose wxWindows.  Right out of the box,
it can target Linux (using GTK), Microsoft Windows (using Win32), and
commercial Unix (using Motif).  It's a wonderfully complete, portable
toolkit that is very sensibly classed.  It's free of commercial
licensing restrictions, and a Macintosh version is also in development.
Given a selection of several high-quality toolkits, why not go with the
one that will please the most people?
 
   I'm rather excited about the potential for products like wxWindows to
blur the lines which can divide developers.  I've heard that there is a
version of wxWindows in development which can link to Qt.  This would be a
perfect bridge between the GNOME and KDE worlds, as well as giving the
hordes of Linux programmers an opportunity to get backports to Windows and
Macintosh with very little additional effort.
 
   When two or more great products go head to head without one having an
unfair advantage, there can be only one winner: the consumer.  (Yes, I'm
talking to you, Bill.)  The truly important message which I gleaned from
this week's debates is this: cross-platform toolkits are becoming an
increasingly important part of the development world.  Developers are
realizing that simply targeting one platform is no longer a safe option. 
Like many Linux users, I had high hopes for Java, but in its current
incarnation it seems to have failed in the "standalone app" universe
(although it has done quite well in the "small front-end embedded in a web
page" universe).  I'm confident that Sun will eventually figure out how to
make Java's success more widespread -- GPL'ing the entire JDK and JVM comes
to mind -- but in the meantime, we must look elsewhere.
 
   That "elsewhere" is currently the several cross-platform toolkits that
are available.  I would encourage all developers to give wxWindows, Qt, and
the others a good close look, and keep portability in mind when starting a
new project that doesn't necessarily have to be Linux-only.  Unlike some
people, we don't need to create platform lock-in.
 
  Art Cancro                              UNCENSORED! BBS
  ajc@uncnsrd.mt-kisco.ny.us              http://uncnsrd.mt-kisco.ny.us

Date: Fri, 07 May 1999 19:26:20 -0700
From: David Smead <dsmead@cyberhighway.net>
To: letters@lwn.net
Subject: OS Battle is won, war lost

Greetings,

Sure I'm a Linux user and hope to stay one, but while us peons are
scratching in the OS barnyard, M$ is developing a new twist on an old
strategy - delivery of information using proprietary formats/protocols.

For the few of us who use computers to enhance engineering productivity,
Linux will eventually be the OS of choice.  In the meantime, look for
the masses to remain addicted to some variant of M$ CE, with content
delivered not only encrypted, but also in formats that are not public.
Their investment in AT&T doesn't look so innocent to me.  Just send
payment of your information utility bill the Bill.

So we win the OS war - so what? Perhaps the Justice Department should
start developing foresight and make sure that delivery of information
isn't monopolized by anyone in the future.  That means if I want to use
an information and entertainment appliance that happens to run Linux,
then I should be able to connect to any delivery service in like manner
to the Internet today.

-- 
Sincerely,

David Smead
Please visit our web site  --  amplepower.com.

Date: Sat, 08 May 1999 21:27:51 +0300
From: "Khalid M. Baheyeldin" <kbahey@ab2.com>
To: editor@lwn.net, letters@lwn.net
Subject: Voice over IP for Linux

This is a call for help and/or pointers.

If you can publish this letter, and maybe some reader will
be able to point me in the right direction.

The issue is that I am looking for an application that works
under Linux (preferrably KDE, but will consider all options)
that would allow me to talk (in voice) to someone else with
Linux and KDE.

The world of Windows95/98 abound with such application from
all spectrums (open source, free, shareware, commercial),
such as Speakfreely, Netmeeting, CU-SeeMe, MediaRing, ..etc.

However, when it comes to Linux, there is a dearth in such
applications.

I have tried most of those I know of, but some do not work
at all, and others provide low quality voice.

Those I tried include:

- SpeakFreely (http://www.speakfreely.org) doesn't work under 
  Linux since the sfmike program bombs out. half Duplex only.

- fphone (http://rodeo.inria.fr). Has a nice interface, but
  the voice is very jittery. Half Duplex only.

- KVirc (http://www.kvirc.org). A super program, but there
  seems to be severe lag of voice (in minutes!!). Half Duplex
  only.

- ephone. Seems to require a fast connection (Ethernet?)

What else is out there?

Thanks, and feel free to edit the message above.
--
Khalid M. Baheyeldin

Date: Thu, 06 May 1999 17:44:26 -0500
From: "Spencer T. Kittelson" <spencertk@abasys.com>
To: lwn@lwn.net
Subject: Mindcraft suggestions

Here's my email to Mindcraft:

-------------------------------------------------------------
Spencer T. Kittelson
VP Systems Engineering
Advanced Business Automation
-------------------------------------------------------------

----------------------------------------------------------------------
 
Mr. Weiner,

There two essentials that are required to perform a reasonable and
useful test of Linux vs. NT:

1) Test using NT Workstation for clients since many business users
   deploy NT on the desktop (they don't like the instability and
   lack of security in Win95/98).  If you want a real test, use
   real configurations.

2) Use absolutely current, state of the art releases, patches, etc.
   The choice of 20-Apr-1999 based upon your unreleased second test
   is an arbitrary date.  It serves no purpose to restrict the 
   software source dates in such a manner.  As a matter of opinion, 
   since the original testing was handled so badly, it looks like 
   you are trying to use a cutoff date to prevent the most recent 
   Linux based performance enhancements from being used.

It is unlikely that Mindcraft (and any of the personnel involved in
the first test) will be able to completely recover their credibility.
Trust and confidence, once shattered, is difficult to restore.  It
remains to be seen if Mindcraft et al. have integrity.  Until then,
all my clients are advised to dismiss any and all Mindcraft "tests"
as cooked, bought and paid for propaganda.

Date: Fri, 07 May 1999 17:23:34 -0700
From: Ariel Faigon <ariel@cthulhu.engr.sgi.com>
To: letters@lwn.net
Subject: An idea for Mindcraft second chance

While Mike Abbott is working on speeding up Apache (stay tuned)...

Since we are comparing Linux and NT here, why not open the game
and try a truly fast web server like thttpd or even better Zeus on Linux?
Having a lot of prior experience with Apache and Zeus, I suspect this will
put NT and IIS in the right perspective even on a very favorable (to NT)
basic hardware config.  Since according to Jef's graphs at:

  http://www.acme.com/software/thttpd/benchmarks.html

Zeus outperforms Apache by a factor of 3 to 20 depending on load
on static small files, The 3x Mindcraft claim might be solved right there.

--
Peace, Ariel

Date: Thu, 6 May 1999 01:22:15 -0500 (CDT)
From: Dave Finton <surazal@nerp.net>
To: letters@lwn.net
Subject: Vague feelings of uneasiness with Mindcraft


It didn't surprise me at all that Microsoft, with its vast arsenel of
marketting assault techniques, would do something like fund an
"independent" benchmark study with NT against Linux.  It surprised me even
less that NT "won" in these benchmarks.

At first I wasn't too bothered by all the hubbub, since anyone with half a
brain takes everything Microsoft says these days with a grain of salt (the
much-lauded security rating Microsoft got with NT disconnected from a
network comes to mind :).  What started bothering me, however was the
negative publicity.  Keep in mind I was glad someone out there would care
enough to refute the results rationally and logically.  What bothers me
now is the *amount* of publicity this is getting.

Think about it:  Microsoft's position is pretty sweet right now.  Linux
developers and enthusiasts are arguing with some backwater benchmarking
company, and all of this is very public.  Mindcraft looks bad because
their benchmarking results weren't very credible.  Linux looks bad both by
proxy and by looking like a bunch of guys screaming "No fair!"  Not only
that, but now this study is getting tremendous amounts of publicity that
it probably would not have generated on its own.  And all Microsoft has to
say is "Hey don't look at us, we just hired them."

We've got to be a little bit more careful in choosing the targets that
Linux wants to hit.  Was the Mindcraft test all that dangerous really?
It's hard to say, but now all of the sudden it's clear that the test
now has a lot more potential to do damage to the credibility of Linux and
Open Source in general that it would have had without front-page
feature-length articles on e-zines like Salon.

Linux enthusiasts have always been ready and able to respond to perceived
threats.  But it can be too easy to allow ourselves to jump at every
opportunity for a cause to defend.  The image I have in my mind right now
is a little hamster running in a wheel chasing after a carrot tied to a
string.  The hamster is very determined to continue its existence by
chasing and eating the carrot, but it's completely blind to the cat
stalking up behind it.  Populist movements can be very powerful, but they
can be easily swayed into making rash decisions.  We have to keep that in
mind in the future (and in the present for that matter).

                          - Dave Finton

---------------------------------------------------------
| If an infinite number of monkeys typed randomly at    |
|   an infinite number of typewriters for an infinite   |
|   amount of time, they would eventually type out      |
|   this sentencdfjg sd84wUUlksaWQE~kd ::.              |
| ----------------------------------------------------- |
|      Name:      Dave Finton                           |
|      E-mail:    surazal@nerp.net                      |
|      Web Page:  http://surazal.nerp.net/              |
---------------------------------------------------------