[LWN Logo]
[LWN.net]

Sections:
 Main page
 Security
 Kernel
 Distributions
 Development
 Commerce
 Linux in the news
 Announcements
 Back page
All in one big page

See also: last week's Back page page.

Linux Links of the Week


The NymIP Project has set itself the task of designing a set of IP-based protocols which will facilitate the use of controlled anonymity and pseudonymity on the net. The standards and their implementation are both intended to be open and freely available. If they get there, this project will have helped to protect privacy in electronic communications.

Not satisfied with using emacs to edit files, read mail, chat on IRC, partition disks, be psychoanalyzed, and edit images? Well, the emacs webring is back in action and ready to connect you with more cool stuff to satisfy your elisp needs.

Section Editor: Jon Corbet


December 14, 2000

   

 

This week in history


Two years ago (December 17, 1998 LWN): IDC reported that Linux's market share rose 212% in 1998, giving it 17% of the server operating system market.

Work continued toward the 2.2.0 stable kernel release. Linus, meanwhile, addressed the topic of raw I/O in Linux:

Quite frankly, nobody has EVER given me a reason that makes any kind of sense at all for supporting raw devices in any other way than we already do. Nobody sane uses a disk without a filesystem, and the insane people that do I feel we can and should ignore. Insanity has a way of dying off over time, when Darvin [sic] starts to look into it.

(2.4.0, of course, will include a Linus-approved raw I/O implementation).

The Debian Project adopted its constitution, which describes how the project operates. The project was smaller then; all of 86 votes were counted (once was sufficient) in the decision on the constitution. The first project leader election began, with Joseph Carter, Ben Collins, and Wichert Akkerman running as candidates.

Red Hat, meanwhile, launched its training and certification programs.

The GNOME project aims to emulate what is best about existing interfaces. "Microsoft did some things very well, and we're trying to learn from them," [Miguel] de Icaza says. At the same time, the project seeks to avoid some of Windows' annoying design peculiarities. GNOME users, de Icaza promises flatly, will not turn off their computers by clicking a button labeled "Start."
-- Technology Review on GNOME, two years ago.

The Linux Mall announced the availability of the first stuffed Tuxes. "A huggable pal to have around, or a great bed partner."

IBM released the first version of Wietse Venema's "Secure Mailer," otherwise known as Postfix.

One year ago (December 16, 1999 LWN) saw, of course, the initial public offering of VA Linux Systems. The company's stock shot up to close at almost eight times its (already increased) initial value, setting a record which remains unchallenged a year later. It was the high point of the Linux stock mania. One year later, VA's stock stands at about 5% of its first-day peak, and the IPO is being investigated by the U.S. government (which is more concerned with the underwriter's actions than VA's).

LWN predicted a flood of Linux-related IPOs to follow. Needless to say, things have not worked out that way.

VA had indeed gone out on NASDAQ -- and I had become worth approximately forty-one million dollars while I wasn't looking. Well, that didn't last long. In the next two hours, VA dropped from $274 a share to close at $239, leaving me with a stake of only thirty-six million dollars. Which is still a preposterously large amount of money.
-- Eric S. Raymond. That didn't last long either.

The Bazaar, a free software conference, was held in New York. Attendance was light, and the event has not been repeated. At the conference, Miguel de Icaza was awarded the Free Software Foundation Award for his work with GNOME.

Bastille Linux 1.0.0 was released. Debian 2.1r4 came out. MandrakeSoft proclaimed that Linux-Mandrake 6.1 was Y2K compliant. Stormix released Storm Linux 2000.

Linus released development kernel 2.3.33 with the comment: "We're obviously not going to have a 2.4 this millenium [sic], but let's get the pre-2.4 series going this year, with the real release Q1 of 2000." He was flooded by those who claim the millennium wouldn't end for another year, and responded:

The fact that our forefathers were Pascal-programmers, and started counting from one does not mean that we have to continue that mistake forever. We've since moved on to C, and the change from 1999->2000 is a lot more interesting in a base-10 system than the change from 2000->2001.

Of course, it looks like no 2.4.0 by the end of the millennium even by the reckoning of Pascal programmers...

Applix acquired CoSource.com, just a few days after the latter's official launch.

Linuxcare closed a large investment round.

But Linuxcare wants to get its business in better shape before it goes public. The company isn't profitable and won't be for the next year as Linuxcare pays for aggressive hiring and expansion, [CEO Fernand Sarrat] said in an interview. Shunning the method pioneered by Internet companies, Sarrat is focusing on building up the business before Linuxcare goes public, instead of using the proceeds of an IPO to fund that expansion.
-- News.com, December 14, 1999

Of course, Linuxcare filed for its IPO just one month later...

 
   

 

Letters to the editor


Letters to the editor should be sent to letters@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.
 
   
From: "Post, Erik" <epost@exch.hpl.hp.com>
To: "'letters@lwn.net'" <letters@lwn.net>
Subject: Red Hat != Linux?
Date: Thu, 7 Dec 2000 11:42:39 -0800 

Dear LWN editors,

one small piece of the 'fun with the press' topic on the December 7th issue
caused me to write this letter. It isn't a really big issue, but still. My
comment pertains to the part in which you talk about the CNet article about
RedHat not supporting Sparc anymore, and the CNet author's comparison with
Windows NT's decline in the number of supported CPUs. You conclude that CNet
made the mistake of assuming Red Hat = Linux.

However, I fail to see the point where the author of the CNet article makes
the fault of assuming Linux = RedHat. The author talks about a specific
company (Red Hat), with a OS product (Red Hat Linux), and compares it with
another OS product (Windows NT) from a specific company (Microsoft).

Besides, (and I quote) "finding a distribution that supports the Sparc is
not hard", especially if you bothered to read beyond the first five
paragraphs of the article.

I'm afraid that LWN fell for the very same thing it is warning people about:
assuming Red Hat = Linux.

Best regards,
Erik Post

   
Date: Wed, 13 Dec 2000 00:10:32 -0900
From: Fielder George Dowding <fgd@mailhost.alaska.net>
To: lwn@lwn.net
Subject: Those who live in glass houses ...

To whom it may concern:

I read with some interest the December 7, 2000, "Fun with the press"
under the headline, "Leading items and editorials." The piece labeled
"This Upside article about Plan 9" sufficently interested me to pursue
the source.

My conclusion is you should not cast stones at a fellow writer for lack
of research.

I will grant that the opening paragraphs were excessive in the use of
literary techniques for holding the attention of the reader. The real
culprit, assuming the author reported accurately, is the CEO of Vita
Nuova, Mr. Michael Jeffrey.

Mr. Jeffrey's cpu requires a bug fix in its logic unit. All of what he
is quoted as having said is almost pure marketing flack. It is clear to
me he does not understand business or the Internet development model. He
and his company will be trampled by the Hurd.

Thank you for leading me on to this situation, but please, a little more
research.

Cheerio!
-- 
Fielder George Dowding, dba Iceworm Enterprises
fgd@alaska.net            iceworm@customcpu.com
909 Chugach Way Lot 35, Anchorage, Alaska 99503-5667 US
   
Date: Tue, 12 Dec 2000 17:02:26 +0000 (GMT)
From: <S.Xenitellis@rhbnc.ac.uk>
To: <letters@lwn.net>
Subject: About FrameMaker and Adobe


I was partly surprised when I read two weeks ago that Adobe is dropping
the Linux version of Framemaker+SGML.

A lot of people do not consider typesetting to be such an exciting area of
work and show too little interest. (I personally find typesetting quite
interesting).

The fact that it is not generally an interesting area is evident from the
lack of opensource applications that can be used to create SGML documents
and generate output formats. Currently, if someone wants to use
DocBook (an SGML "application" suitable for technical documentation) to
generate output PDF/PS output, s/he has only two solutions, each of them
giving not so good results. There are to use "JadeTeX" or "PassiveTeX",
both from Sebastian Rantz. "JadeTeX" was a hack and is not developed
anymore (table support in JadeTeX is not good) while "PassiveTeX" appears
to be the right way to do things and is  being developed. However, at the
moment, PassiveTeX does not appear to be able to generate all sorts of
documents. (this is not an accusation on Sebastian)

From what it appears here, there is currently no globally acceptable
editor to write SGML (DocBook) documents and the document generation does
not offer too much sophistication.

Adobe could have had the chance to invest into a simple GUI product that
can aid in
a) the writing/editing of SGML documents
b) generation of all sorts of PDF/PS output

Projects like KDE and GNOME make extensive use of DocBook and most other
OS documentation is converted into DocBook as well.

My recommendation to Adobe is that when they consider entering the
Linux market, they should think primarily in capturing the market and
getting the money latter from services by companies who make commercial
use of the product. Linux is stuck with no complete (easy to use,
supporting all features of standards, supporting many local languages)
typesetting tools.

Thanks,
Simos Xenitellis

   
From: Peter Samuelson <peter@cadcamlab.org>
Date: Fri, 8 Dec 2000 00:55:02 -0600 (CST)
To: jra@baylink.com
Subject: Re: Universal RPMs.


[Jay Ashworth]
> In general, anytime that part of an installation involves "put this
> file in the right place" or "change this [parameter in] this system
> control file (inittab)", there should probably be a program that does
> the work, the call to which can be standardized across systems, and
> the underlying actions can be specific to a distribution.

Hey, you're describing Debian:

  $ ls {,/usr}/{,s}bin/update-*
  ls: /bin/update-*: No such file or directory
  /sbin/update-modules
  /usr/bin/update-menus
  /usr/bin/update-ppd
  /usr/sbin/update-alternatives
  /usr/sbin/update-dlocatedb
  /usr/sbin/update-fonts-alias
  /usr/sbin/update-fonts-scale
  /usr/sbin/update-inetd
  /usr/sbin/update-ispell-dictionary
  /usr/sbin/update-mime
  /usr/sbin/update-passwd
  /usr/sbin/update-rc.d
  /usr/sbin/update-vfontcap
  /usr/sbin/update-xaw-wrappers

Now granted, a few of these are Debian-specific ("update-alternatives"
manages the alternatives system by which you can have both nvi and vim
installed, either of which (but not both) can have a /usr/bin/vi
symlink) but many of them -- update-inetd, update-rc.d, update-passwd,
update-mime -- could easily be applied to other distributions.

There is actually a Debian package 'file-rc' which replaces the SysV
runlevel symlink tree with a single 'runlevel.conf' file -- and thanks
to the update-rc.d abstraction layer, it works transparently, without
any changes to other packages.

Peter
   
To: letters@lwn.net
Subject: Re: Universal RPMs
From: Marc Lefranc <mlefranc@libertysurf.fr>
Date: 07 Dec 2000 23:01:29 +0100


Dear Editors,

I would first like to thank you for the best Linux news source in the
world.

In the issue of LWN dated Dec. 7, 2000, Jay R. Ashworth complains that
the Red Hat distribution does not provide a simple user interface for
starting and stopping SysVinit services. He notes that he created a
simple script, called svc, which provides a wrapper around the script
located in /etc/rc.d/init.d.

I would just like to point out that:

1) It has been some time since RH has been shipping such a user
   interface. Just look at.../sbin/service which does exactly what the
   above-mentioned svc does.

2) starting with RH 7, there is a symbolic link /etc/init.d ->
   /etc/rc.d/init.d, which makes the /etc/init.d valid on RH, Debian,
   most certainly Mandrake (not checked), and probably soon most other
   distributions. This implies that /etc/init.d/script start might be
   even simpler that calling the svc|service interface.

Marc Lefranc

   
Date: Thu, 7 Dec 2000 14:58:58 -0500 (EST)
From: Joseph J Klemmer <klemmerj@webtrek.com>
To: "Jay R. Ashworth" <jra@baylink.com>
Subject: Re: Universal RPMs.


> The files on a Red hat distribution (among others) in the
> /etc/rc.d/init.d directory constitute a sort of "service manager
> interface", in conjunction with SysVinit, they're one of the few
> ideas stolen from NT that I like.  But, while many Linux
> distributions provide the "chkconfig" command for setting services in
> this directory to be enabled or disabled in specific runlevels, that
> command doesn't provide a user interface for turning something on or
> off, or restarting it, *right now*.  I created my on, called svc:
>
> /etc/rc.d/init.d/$1 $2
>
> Real complicated, right?  People do that all the time, right?  So why
> hasn't RH already added that to the distro?

	FWIW, there's a little known utility called "service" on RH (don't
know if it's on any of the others) that does exactly what you're referring
to.  It seems to be an undocumented little thing but, as it's a shell
script, it shouldn't be to difficult to figure out.  Just try this as
root:

[root@billy /root]# service
Usage: service < option > | --status-all | [ service_name [ command ] ]

The script is in /sbin and makes it real handy to do things like -

service httpd restart
service sendmail stop

Just thought you'd like to know.  :-)

---
There are just 25 days till the beginning of the 21st century and the next
millennium!


   
Date: Thu, 07 Dec 2000 13:42:15 -0500
From: Pierre Baillargeon <pb@artquest.net>
To: letters@lwn.net
Subject: Elevator algorithms

Hi,

Just a comment about the conclusion Thomas Sippel - Dau
(t.sippel-dau@ic.ac.uk) reached about the elevator algorithm.

When you strip down the math notation from his letter, you get the
simple claim that applying two sorting algorithms in succession is
wasteful because only the last one will prevail.

This is patently false. A sorting algorithm does not need to affect all
elements. An example of such algorithm is the stable sort: equal
elements are not moved. This algorithm is useful to sort by two criteria
in succession: for example sorting mails first by sender then by date,
so that letters received on a single day will be sorted by senders.
Using a single sorts would not produce that result.

Elevator algorithm are part of such breed. Only the elements that meet
certain criteria will be moved around. So the effort is not wasted. The
kernel can optimize with the criteria it consider important, and the
disk controller will sort them with its own optimizing criteria.

The second flaw of the argument is the assumption that both the kernel
and the disk controller see the same set of requests. This is not
necessarily so. In fact, the elevator algorithm is there to provide the
most cohesive set of requests to the controller, so that its own
internal algorithm can do an effective job.
   
Date: Thu, 7 Dec 2000 13:03:15 -0500
From: "Jay R. Ashworth" <jra@baylink.com>
To: letters@lwn.net
Subject: Disk Elevators redux

In last week's letters column, Thomas Sippel made an excellent point
about OS-based disk elevator transfer-reordering algorithms: disk
controllers with onboard cache reorder transactions themselves, anyway.

I wanted to make a different point, and it's still valid, so I will.  :-)

It wouldn't matter if the drives were uncached, anyway: the Cylinder,
Head and Sector address information with which the drive talks to the
outside world has been fictional for many years, anyway.  Many, and I'm
tempted to say all, but can't back it up, drive use translation
mappings such that the sectors on one "cylinder" do not all reside on
the same physical cylinder.

This alone would seem to make OS-based elevators not only useless, but
in some cases, the pessimal solution to the problem at hand, no?

Cheers,
-- jra
-- 
Jay R. Ashworth                                                jra@baylink.com
Member of the Technical Staff     Baylink
The Suncoast Freenet         The Things I Think
Tampa Bay, Florida     http://baylink.pitas.com                +1 727 804 5015
   
Date: Mon, 11 Dec 2000 19:07:26 -0800
From: Thornton Prime <thornton@jalan.com>
To: mfoley@zdnet.com, letters@lwn.net
Subject: Re: Open-source Backers


Dear Ms. Foley,

I know as an avid industry watcher you recognize the enthusiasim and
growth of open source projects. Open source is clearly already an
important force in the market, and will continue to be as long as
individuals find rewards in the model. At the same time, though, I
submit that the financial interests for a corporation that actively
contributes to open source projects is substantially greater than any
reward available to individuals.

>From your article posted at
http://www.zdnet.com/zdnn/stories/comment/0,5859,2662295,00.html I see
that you have wondered about how companies can gain from participation
in open source projects. I am writing to offer you what I believe to be
reasons why companies in some parts of the IT industry should consider
participation in open source projects.

I see six basic financial interests in corporate sponsorship of open
source projects. Most of these depend on a particular business model,
but all apply to a company as large as IBM which is pursuing multiple
business goals simultaneously:

1. Contributing to open source projects is absolutely essential for any
hardware company hoping to make sales in the rapidly growing open source
operating system market. To guarantee the best driver support for your
hardware, you need to write your own drivers. To guarantee equal
protection and treatment of your drivers in a community like Linux,
these drivers need to be open. For companies like IBM and Compaq,
investing in open source is investing in a substantial investment they
have made in CPU development. They are guaranteeing future markets for
their PPC and Alpha processors. Smaller hardware manufacturers stand
just as much to gain.

2. Contributing to an open source project enables a company that sells
associated services or projects to guarantee computability and help
shape future directions of a software that is key to their future. IBM
again provides a great example of a company that benefits in this way.
By participating in the Apache product, they guarantee compatibility
with WebSphere. they have also substantially moved forward the Apache
Group's next generation server, moving it from a forking Unix codebase
to a model that provides multiple multi-process architectures, including
threads, pre-forking and a hybrid.

3. Contributing to open source projects helps develop internal expertise
and enhances the credentials of a company who provides consulting
services for the software or a related product. IBM, by contributing
performance patches to Apache has demonstrated themselves as an
authority on Apache and web servers in general.

4. Developers come and go. Open source developers remain loyal to their
code and support it after they have left. IBM licensed the postfix mail
server under their open source license. The developer, Wietse Venema,
continues to actively improve and enhance his code, even though the
license remains IBM's. While it started as a research project it has
turned into a secure and stable alternative to sendmail now used by
Compaq and many other companies around the world. IBM continues to enjoy
a return on their research investment even though the principal author
is no longer in their employ.

5. Good systems get better faster when they are open. If there is no
real market for a software project, open sourcing it can create a market
and create opportunities. Once code is available to the public it
matures more quickly and at less cost than a comparable commercial
project as long as there remains good management of the project. 
Developers outside of an organization offer abilities and enhancements
that often are unthought of within a company of origin. Again, I use the
example of postfix. IBM had less than a snowball's chance in hell of
penetrating a mail server market dominated by Sendmail, Lotus, Novell,
and Microsoft when Mr. Venema started his project. Still, developing the
expertise internally was critical to other business processes. While the
code had no value commercially, releasing it publicly opened competition
to Sendmail and created a new market. The code improved dramatically,
gaining database and LDAP support, while increasing in security and
performance once other developers had an opportunity to contribute and
once system administrators deployed, tested, and suggested enhancements.
Another good example of this is Mozilla and Netscape. While many
proclaim Netscape as an example of a failed open source project, I think
few would contest the fact that Microsoft's growth in the browser market
was only halted once Netscape open sourced their browser and created
Mozilla. While Mozilla may not be the best example of an open source
product that creates opportunities, it is a clear example of a move that
saved a dwindling market share from dwindling further. In the end
Mozilla may turn around the browser market when and if AOL adopts
Mozilla as the core of their product. In fact, because Mozilla is
available on so many platforms it may help AOL enter markets they
probably never considered, including Linux, Be, and the embedded browser
markets.

6. Most open source developers code for two simple reasons: ego and fun.
In an increasingly tight job market, both of these motivations can be
key to employee acquisition and retention. Companies like IBM, who only
a few years ago were considered too buttoned-up for most developers, are
now able to make attractive employment offerings to a market that seems
less concerned with money and more concerned with fringe benefits.

Thornton Prime
 

 

 
Eklektix, Inc. Linux powered! Copyright © 2000 Eklektix, Inc., all rights reserved
Linux ® is a registered trademark of Linus Torvalds