See also: last week's Letters page.

Letters to the editor

From:	 Keith Owens <kaos@ocs.com.au>
To:	 letters@lwn.net
Subject: Exit sections and monolithic kernels
Date:	 Fri, 22 Mar 2002 14:02:41 +1100

You say
> useless "undefined reference to `local symbols in discarded section
> .text.exit'" message that accompanies a failed link

These messages are not useless, on the contrary they are detecting
coding errors where people call functions that have not been included
in the kernel.  These are kernel bugs just waiting to happen.  When
binutils started checking for dangling references, it flushed out
several coding errors.  The down side is that we have to tell the
kernel which dangling references to ignore, using __devexit_p.

From:	 Stephen.Schaefer@emis-intl.com
To:	 letters@lwn.net
Subject: Re: [m]ore GPL confusion
Date:	 Mon, 25 Mar 2002 23:55:32 -0500 (EST)

In my understanding of the GPL, there is an important freedom one
retains when using GPL software: you are under no obligation until you
transfer rights in the software (``give it'') to someone else.  If you
obtain a copy and modify it for your own use, noone's going to track you
down and require you to divulge the changes *unless and until* you
provide the software to someone else.  IANAL, but it seems you could go
so far as to provide services using the modified GPL software, again, so
long as you did not distribute the software itself to any other entity.
Otherwise, there would be no need for the Affero GPL, in which such
occlusion of the service is specifically prohibited.

Example: suppose you take GPL scheduling software and modify it to
include your cement company's trade secret algorithm for scheduling
cement trucks.  Are you in any way required to divulge your
modifications?  No.  Suppose you start offering cement truck
scheduling as a service to other cement companies.  Unless the
scheduling software was published under the Affero GPL, you can still
keep your trade secret.  Now suppose you wanted to sell cement truck
scheduling software.  It is only in this case that you would be
required to choose between replacing all the GPL licensed code with
differently licensed code (possibly with the same code, but licensed
differently from the copyright holder), or publishing the full source
and developing a service-oriented business model supporting and
refining the published software.

For reasons I don't understand, people cannot seem to comprehend that
although Free Software advocates want universal participation,
we do not, through the GPL, compel it.  The GPL is a highly pragmatic
compromise.  The world would be unimaginably richer if information *were*
free, but GPL software is no more than an invitation to a beguiling
half shadow of that world.

	- Stephen P. Schaefer

From:	 =?ISO-8859-15?Q?Leandro_Guimar=E3es_Faria_Corsetti_Dutra?=
	 <leandrod@mac.com>
To:	 letters@lwn.net
Subject: Hurd and proprietary software
Date:	 Thu, 21 Mar 2002 09:56:07 +0100

 > We are working at clarifying things further, in an attempt to
 > discover (and fairly represent) what the Free Software
 > Foundation's objections are with regard to the existing, fully
 > free distributions. Stay tuned...

	No work needed here.  The meaning of RMS' declaration is clear, and 
clearly documented in the FSF site, besides being clear in the 
history of the relation between Debian and GNU.  While Debian and 
GNU work closely together, and at one time Debian was considered to 
be part of GNU, their differences is that GNU wants to be completely 
free, while Debian has this hair-splitting about providing 
proprietary software (along with free software that depends on 
proprietary software) as add-ons to the official, completely free 
standard Debian software distribution.


-- 
  _
/ \ Leandro Guimarães Faria Corsetti Dutra        +41 (21) 216 15 93
\ / http://homepage.mac.com./leandrod/        fax +41 (21) 216 19 04
  X  http://tutoriald.sf.net./               Orange Communications CH
/ \ Campanha fita ASCII, contra correio HTML      +41 (21) 644 23 01

From:	 Leon Brooks <leon@cyberknights.com.au>
To:	 Jeroen Dekkers <jeroen@dekkers.cx>
Subject: A flying Hurd
Date:	 Sat, 23 Mar 2002 21:20:34 +0800
Cc:	 letters@lwn.net

> The following sentence doesn't make much sense to me either: "Thus,
> it seems unlikely that the HURD will mount a substantial challenge
> to the established free kernels anytime soon."

Pretty easy to understand, really, since you already have all of the pieces 
of the problem:

> Although the current implementation doesn't show it, the design of
> the Hurd and the ideas behind it really rock.

...words 2 through 7 are your answer.

When the current implementation does show it, the Hurd will have acheived 
airspeed. And impressive it will be, as well, there's lots of drool-over and 
kick-ass stuff in there.

Given the time it's taken so far, the Hurd's logo really does need to be a 
squadron of winged pigs (it will win hearts, trust me), or perhaps the Spruce 
Goose, but I think most of the Hurd's team members take it too seriously to 
let that happen. (-:

Cheers; Leon

From:	 wa6cvl@sbcglobal.net
To:	 letters@lwn.net
Subject: abe lincoln and the digital pirates
Date:	 Wed, 27 Mar 2002 11:39:00 -0800
Cc:	 wa6cvl@qsl.net

       Mr Eisner:
        I just read your essay that was posted to the Financial Times. I
feel sorry for the problems that you're having. And I respect the
existance ofthe Capitalist system that we enjoy. But contrary to popular
thought, Capitalists (such as yourself) should not be protected and enhanced by
government intervention. Your tale of woe ignores the fact that copyrights
were once of shorted duration, that they were non-renewable, that "Fair Use"
was a bonafide consideration. These laws were originally written to protect
individuals, not corporations. In the spirit of freedom.  I believe that the
government should become less involved.. And mr eisner, intellectual
property rights are not synonamous with monopoly rights..
        Instead of talking about the CBDTPA as a solution, let's look at a
CBDTPA assault of the automobile and the freeway system.
        In  the following analogy:
        1. Any Vehicle = digital device such as a computer.
        2. Roadways   = networks

Be it proposed that it be prohibited to build or modify any vehicle that can
be used in in criminal activity. Such vehicle should not be function in the
commission of any crime, or even the appearance of a crime. If necessary,
the USA will fund and innovate measures to ensure that such standard can be
innovated. Fair Use will not be allowed to any vehicle that is even capable
of being  used in a criminal manner.
        Gosh, Mr Eisner, with your guidance, we could interpret a crowbar in
the trunk as a burglary tool, and insist that car automatically disable..
        If this proves fully successful, we will extend to Political
Correctness cause also..
        Perhaps the government can fund a plan to identify wrong doing in
advance. If your car stops suddenly, you must've been about to do
something bad.....

        AND SO ON.......

de jerry

From:	 "Howland, Curtis" <howlandc@kvh.co.jp>
To:	 <letters@lwn.net>, <dlc@radix.net>
Subject: Seen in the March 21st Letters...
Date:	 Wed, 27 Mar 2002 11:25:40 +0900

Dear LWN,

In the March 21st letters section, David Craig considers the statement that
"Yes, it's true: the U.S. government really wants to outlaw free software."
to be inflammatory and beneath the journalistic standards I have come to
expect from your Web site."

I consider the statement, if actually written by LWN, to be rather a mild
understatement. A cursory glance at the 5 Million laws on the books in
America demonstrate that control over ones own life, be it in terms of
labor or software, is the last thing that the various governments in the
U.S. want you to have.

Government is about control. Free software, as in beer or speech, endangers
that control.

Curt-


---
Written from, not for, work.

From:	 "jacob navia" <jacob.navia@wanadoo.fr>
To:	 <letters@lwn.net>
Subject: zlib
Date:	 Sat, 23 Mar 2002 22:22:24 +0100

zlib corrupts malloc data structures via double free. This vulnerability
impacts all major Linux vendors. It may impact every Linux installation on
Earth. Updates are required to zlib and any packages that were statically
built with the zlib code

Wouldn't be a good idea to modify free() so that it never frees twice?

When a pointer is passed to it, free should look up if the address is a valid
address, and set an error flag and do nothing if the address is incorrect.

This does not mean that a huge list of addresses must be maintained, but just
some range checking could greatly speed up the process.

I do not think that writing a better free() is completely beyond the reach of
the clib people isn't it?

This would fix all of those bugs in Linux forever, without any need for
patching all buggy applications, or waiting till all those bugs surface!

But who develops now for security?

From:	 "Robert A. Knop Jr." <rknop@pobox.com>
To:	 letters@lwn.net
Subject: A note of praise for Seagate
Date:	 Tue, 26 Mar 2002 18:36:36 -0600

I was having some trouble reading a DDS3 tape written with an HP drive
on a Seagate Archive Python 04016 DDS3 drive.  It turned out that the
solution was that I needed to issue a "mt setblk 10240" command on the
Seagate drive to get the blocksize it was using set for how the tape was
written.  However, along the way I cruised by Seagate's support site for
the drive, and saw that there was a firmware upgrade available.  I was
prepared to be annoyed, as the firmware upgarde was a DOS archive
including a program to run to patch the upgrade.  (I don't run Windows
or DOS at all on the machine with the tape drive.)

However, poking around, I discovered that Seagate's got a diagnostic and
firmware patching program available for Linux (in addition to Netware,
Windows, and Solaris).  Hence, I was able to upgrade my firmware (even
though that wasn't the solution to the problem at hand).  This utility
may be found on:

  http://www.seagate.com/support/tape/utils/stdiag.html

I don't know if there is a central information clearinghouse for how
"linux frendly" various vendors are.  You can find that information for
scanners on the SANE page, for printers on the gimp-print page, and for
other devices *sometimes* on their pages.  This is the sort of
information that would be nice to have, however, before making purchase
decisions.

-Rob Knop
rknop@pobox.com