See also: last week's Back page page.

Linux Links of the Week

Those who are into high-end sound applications on Linux may want to have a look at LinuxDJ.com. This rather utilitarian site is the home for a number of audio development projects and documents.

Section Editor: Jon Corbet

This week in history

Eric Raymond celebrated the first six months of the "open source" term.

Because if we truly desire world domination, we've got to get our LSD into the corporate elite's conceptual water supply and alter the beast's consciousness. That means we need to co-opt the media that shape decision-making at the highest corporate levels of the Fortune 500.

The development kernel was 2.1.114; work continued on the 2.0.36 stable release. Much energy went into a vast flamewar over whether the devfs patch should go into the 2.2 kernel; in the end it didn't happen, but it will be there in 2.4. The beer-drinking penguin logo was removed from the development series.

One year ago (August 5, 1999 LWN): SGI jumped into Linux with both feet, announcing a new Linux-based server system. The company also let it slip that Irix would not be ported to the Intel architecture.

Eric Raymond addresses the question of whether free software can be original:

But there is a more fundamental error in the implicit assumption that the cathedral model (or the bazaar model, or any other kind of management structure) can somehow make innovation happen reliably. This is nonsense. Gangs don't have breakthrough insights -- even volunteer groups of bazaar anarchists are usually incapable of genuine originality, let alone corporate committees of people with a survival stake in some status quo ante. Insight comes from individuals.

The development kernel release was 2.3.12. Linus Torvalds announced that the 2.3 kernel would go into feature freeze "in about two weeks." Here we are, a year later... The stable kernel release remained 2.2.10.

A Linux Lament in Salon complained about problems with the Red Hat community stock offering:

We coders had been abruptly disenfranchised, after having had silver carrots waved in front of our noses. I'd opened my first money-market account just now, in order to take part in the commercial future of something I believed in -- and the door had been slammed in my face.

Letters to the editor

Date: Thu, 27 Jul 2000 23:35:42 -0400
From: Patrick Callahan <pac1@tiac.net>
To: lwn@lwn.net, "basiclinux@topica.com" <basiclinux@topica.com>
Subject: LWN - So what happens when Linux really explodes


>So what happens when Linux really explodes, as seems (to some)
>inevitable? Just how weird is it going to get? Will we look back with
>nostalgia to 1994, when nobody knew what we were talking about? Will we
>want our old Linux back? For now this is still our revolution,  and we
>can maybe shape its future. Before long, that may no longer be true. 

I've jumped on the Linux bandwagon recently.  August 1999.  I think
things have changed radically since then.  When I first started, it was
not unusual to get a well reasoned response to a request for specific
information in a linux help chat room on irc.  Maybe lately I've been
hanging out too much on the wrong sort of channels but it seems that the
overall tone of the irc experince for geeks like me is changing
somewhat.  Has anyone else noticed this?

There seem to be many more seekers of information than givers.  There
seem to be more questions from people who haven't read the fine manual,
don't know where the manual is, and don't care... yet....  As always,
there's interesting and interested people to chat with, but the noise is
getting louder...

I wonder if the Linux Cognecenti are overwhelmed by the increasing
numbers of people arriving at linux.  Have they stopped responding to
newcomers, just because there's so many of them.  Or maybe they're just
responding to questions that interest them.  How do people who have been
here since the early days feel about people like me who are late to the
party?

Some responses in some forums lately seem quite harsh or irritated,
almost as if the information givers are getting fed up repeating
themselves to each newcomer who arrives on the scene.  

Other places are a delight to be in.  The Basic Linux Training Mailing
list is terrific.  I think its because most of its members are
newcomers, committed before they join the list, to actually working at
learning linux .  This may not be the case in other forums.

-Pat Callahan

Date: Fri, 28 Jul 2000 11:36:01 -0400 (EDT)
From: Elliot Lee <sopwith@redhat.com>
To: "Aaron J. Seigo" <aseigo@mountlinux.com>
Subject: Your LWN post

Just wanted to correct a small technical point:

> I point to Icaza's own project Gnome as an example that he is (to
> quoth him) "smoking crack" when spouting these arguments. Gnome sets
> policy, and in the right place, too: on the application level.

That is incorrect - all desktop-generic policy is set in gnome-libs and
the other Gnome libraries, not in the applications themselves. gnome-libs
and related pieces would generally be accepted as part of the operating
environment, rather than part of the application.

And an opinion:

> Well, look at BeOS/Mac/Windows. They each enforce policies on
> programmers and users at the system level, but because of that they
> are each cordonned off into their own space of the computing arena.
> Unix is a substrate that strives to be flexible enough for _any_
> policy.

You may be ignoring a few things:
	. BeOS/Mac/Windows are intended primarily to meet the end-user's
	needs n the desktop, which is why they have to set policy.
	. The unavailability of a widely used group of UNIX libraries that
	set user policy is a large reason for the failure of UNIX on the
	desktop so far.
	. Gnome's goal (and I believe the goal of all the desktop
	projects) is to make UNIX viable as an end-user desktop
	platform, which *requires* setting policy across apps.
	. For what it's worth, it is possible to define an aweful lot
	more policy on BeOS/Mac/Windows than your post would imply -
	you are making a lot of assumptions based on having used them
	rather than any real facts. (Not that they aren't sometimes
	a pain to use :)

I think a main point of confusion lies in the fact that you see "the
system should set policy" and start to incorrectly think that the kernel
will suddenly start to know about my mouse settings or root window
background. This is not what is being advocated. I think you are also
confusing "setting policy" (which is a good thing) with "the system is not
transparent or flexible". The latter is the case with BeOS/Mac/Windows, is
a reason I like UNIX so much, and has nothing to do with policy being set.
It is entirely practical to both set policy _and_ be transparent/flexible.

If you wish to argue that ever setting any policy on an inter-application
basis is evil, that is most likely because your goal could to continue
using the same old (and definitely fun from a hacker's perspective! :)
UNIX that has been around for ages, rather than bring UNIX to the desktop
and other new frontiers, as Miguel's goal is. If you choose to argue this,
that is fine - you may not agree with this goal and the changes it
requires, but recognize & respect the reasons for which the opinion was
expressed.

Not a Miguel fanboy, but annoyed at random rants,
-- Elliot
The best way to accelerate a Macintosh is at 9.8 meters per second per second.

Date: Wed, 02 Aug 2000 17:02:49 -0400
From: "George B. Moody" <george@mit.edu>
To: letters@lwn.net
Subject: Fred Moody's story, "Linux Sux Redux"

The story indicates that the numbers of vulnerabilities reported on BugTraq
for "Red Hat and the other Linuxes" were 122 in 1999 and 47 so far this
year, and notes that Windows NT's counts of 99 and 37 are significantly
lower.  The error is that the numbers for Linux *include* those for Red
Hat, so that adding the Red Hat numbers to those for Linux results in
counting the Red Hat vulnerabilities twice.  In fact, the correct numbers
for all versions of Linux put together are 84 for 1999 and 30 for 2000, and
for Red Hat they are 38 for 1999 and 17 for 2000.  In round numbers, the
numbers of vulnerabilities in Windows NT are about three times as high as
those for Red Hat.

Fred says, "If you look this list over, and measure each system's number of
vulnerabilities against the number of its customers, Linux is arguably the
worst operating-system product in history, and Microsoft's the best."  A
more bizarre way to assess quality would be hard to imagine.  If I
understand him correctly, Fred is suggesting that quality is proportional
to market share, and that having more customers in some way can overcome
having more bugs.  This is no more true of software than it is of food.
The greasy spoon in the mall may attract more visitors despite high prices
and poor sanitation, but those who are lucky enough to enjoy a friend's
home cooking are not only getting a free lunch but a better one, and they
get to inspect the ingredients if they care to do so.  Those who are so
thoroughly in the grip of the belief that what costs more must be better,
and that anything free is therefore worthless, might spend their money on a
nice bunch of flowers for the cook; or they can throw a brick through their
friend's window and go eat the best mystery meat in town at the greasy
spoon with Fred.

"As Linux zealots are beginning to find out, it's a lot easier to
masquerade as a better product than it is to go out and be one."  Earth to
Fred: Get a clue!  We Linux zealots(TM) know that marketing can make people
believe that expensive and shoddy products are better than superior free
alternatives, and guess what?  Anyone who has ever paid too much for
something just because it comes in a shiny box knows it, too.

-- George Moody (no relation to Fred, as far as I know)