LWN: Comments on "qmail released into the public domain"

This is a mistake, but he also provided a license, so it doesn't really matter

kreutzm — Tue, 18 Dec 2007 19:39:35 +0000

At least for Germany there is no such concept as "public domain".

qmail released into the public domain

OLPC — Sun, 02 Dec 2007 15:40:03 +0000

daemon-utils and ucspi-tcp are not strictly necessary to run qmail.  I find their equivalents
xinetd and tcp_wrappers superior and I've been using qmail with them for several years.

Doesn't protect DJB

BrucePerens — Sun, 02 Dec 2007 02:24:57 +0000

I've had a few short email conversations with the JRMI author, but mainly I'm going by press coverage of the case. I'm sure Eben Moglen read this finding. If it gets to an appeals court and passes there, it will have some precedential value. Right now it doesn't have that much.

Thanks

Bruce

Doesn't protect DJB

linuxrocks123 — Sat, 01 Dec 2007 19:12:05 +0000

I'm interested in this case, so I'd like to know where you're getting your information.  Is it
all off of JMRI's website?  The judge's ruling did state that JMRI might have a case for
breach of contract, but the finding that a "nonexclusive license implicitly includes a promise
not to sue for copyright infringement" is certainly unsettling.  I see that JMRI has
relicensed their code under GPLv2.  Perhaps a GPLv3.1 explicitly stating that no promise to
not sue for copyright infringement is implied by the license would be a good idea?

qmail released into the public domain

arekm — Sat, 01 Dec 2007 17:31:35 +0000

Did he really released "and a number of other packages" ? Or just said that he will do that in
future?

Looks like other tools needed for qmail (daemontools, ucspi-tcp) are not yet rereleased as
public domain.

qmail released into the public domain

djpig — Sat, 01 Dec 2007 10:14:46 +0000

At least now Linux distros will be able to ship the damn thing.
You're assuming they would want to.

I'm fairly certain there are enough qmail users still out there so that someone will care enough to put it into Debian (and Ubuntu universe etc.). It will probably not end up in the default install of any distro or something like that.

Why "oublic domain release" is a bad idea (Re: Huh?)

PO8 — Sat, 01 Dec 2007 03:52:58 +0000

First off, I Am Not A Lawyer, and nothing I say constitutes any sort of legal advice.  That
said, here's my lay opinion.

1) Lawyers I trust tell me that the whole concept of "releasing into the public domain" is of
doubtful validity in US law.  There are various ways spelled out in US law that a copyrighted
work can enter the public domain.  One of these is "abandonment" of the copyrighted material.
But it is not clear to these attorneys that you can *explicitly* abandon a copyrighted work.

2) One of the rights protected by copyright is the right of attribution.  If I were to take
public domain code and claim to have written it, this would, I believe, be perfectly legal (as
long as it wasn't part of some other actionable offense).  If you are proud of anything in
your code, you probably want to avoid that.

3) One of the things a (good) license tries to do is to control legal risks of release of
material.  Disclaimers of warranty and liability (for what they're worth) and legal terms of
use can help to protect the author from expensive and potentially damaging litigation.

4) License notices and copyright notices are a way of protecting the user of a copyrighted
work from a malicious author.  Under current US and International law, published work defaults
to "copyrighted" even in the absence of any copyright notice.  The placement of copyright and
license notices specify exactly which rights are granted by the author, and under what terms.
Anyone who uses code from a large "public domain" release may run a substantial risk that the
author may say "oh, but I didn't mean to release *that*" to a court.  I wouldn't want to be
defending myself from a copyright infringement charge over this.

Conversely, if an author publishes copyrighted work related to her previous public domain work
and inadvertently omits the copyright notice, it seems easy for someone unscrupulous to claim
that it was part of the public domain release.


Like I say, I'm not a lawyer.  But for all of these reasons, I don't plan to release any more
code for public use without licensing it somehow.

Why a license is better than Public Domain

ajross — Fri, 30 Nov 2007 23:39:57 +0000

It is better for those who use the code, because a license identifies someone who claims to have authority to grant the license. One of the problems with Public Domain is that there is no identified authority

In what was does Bernstein's announcement and the decade-long history of the product not constitute an identification of authority? This is very much a mountain out of a molehill. Sure, it's conceivable that you could get in trouble just grabbing source off the internet if you have to prove later on that it's legally useable. But it should be abundantly clear that that is not what is happening here. Really, are you honestly any more confused about where qmail comes from now than you were yesterday?

There are reasons, this isn't particularly compelling

kmself — Fri, 30 Nov 2007 20:37:40 +0000

Proof and/or demonstration of authorship is one of the key parts of virtually any copyright dispute. In most cases it's sufficiently trivially accomplished that it's not given much thought. There are significant legal sanctions against false claims of authorship, and for a work as well documented and publicly distributed as qmail, chances of success would be low (though cooperation of djb in any defense would help).

And having a license over a codebase doesn't necessarially free you from conflicting claims of ownership. Hell, it's even possible to imagine a world in which some two-bit thuggish company tried to claim a copyright interest in the Linux kernel.

The better reasons for license, of some sort, are:

A framework for distributing the work, including:
Clear statement of copyright interest.
Clear statement of rights of recipients to modify, use, and redistribute the work.
A clear statement of any additional rights of authorship which are or are not retained (rights not explicitly granted are implicitly retained under US statute).
Clear statement of obligations (if any) when receiving, modifying, using, and/or distributing a work.
A disclaimer of warranty and liability

The cover of an approved license also greatly relieves the burden on those who would like to engage in further distribution, modification, and use of the work. For many FSF Free Software projects, having a clear check that works are granted under an approved and accepted license provides a very real and useful protection against real and imagined legal threats, from both outside and inside the organization. Mirror sites and CD/DVD services for CentOS, Debian, Fedora, and Gentoo come to mind. Likewise, odds of an effective attack against adopting a tool for use within an organization merely because it is licenced under GPL, BSD, or MIT license (and is hence "legally risky") wouldn't get far in any sane environment -- in almost any case such code is already in extant use. "Public domain" is a slightly harier threat at present.

qmail released into the public domain

lovelace — Fri, 30 Nov 2007 20:08:58 +0000

At least now Linux distros will be able to ship the damn thing.

You're assuming they would want to.

Doesn't protect DJB

BrucePerens — Fri, 30 Nov 2007 19:47:31 +0000

DJB and Phil Zimmerman's crypto cases were early ones, and they had lots of room to make case law invalidating legislation. In contrast, DJB would hardly be the first party to be sued for infringing a software patent, the precedents he could use to invalidate the law are already tried and gone.

Bruce

Doesn't protect DJB

rsidd — Fri, 30 Nov 2007 19:15:32 +0000

From DJB's history, I suspect he would enjoy the fight.

not necessarily

rsidd — Fri, 30 Nov 2007 19:09:24 +0000

DJB knows what public domain is. I really don't see what people gain from insulting his intelligence.

Why a license is better than Public Domain

jreiser — Fri, 30 Nov 2007 18:29:30 +0000

How is it "always best to provide a license"?
It is better for those who use the code, because a license identifies someone who claims to have authority to grant the license. One of the problems with Public Domain is that there is no identified authority. Anyone can claim that something called "public domain" instead is really theirs (was stolen from them.) A license provides users with a defense, and somebody to sue for not having legitimate authority to grant the license. Public domain opens users to attack from all quarters, and requires users to mount their own defense. Suppose that [a court decides that] the code really does belong to somebody else. Then a user who claims "public domain" is liable for intent to steal the code, while a bona fide licensee is not [excludes collusion between licensee and licensor to steal the code by creating a bogus license, etc.] In either case the user might be forced to discontinue using the code, but having a [supposedly valid] license tends to limit other losses.

Doesn't protect DJB

BrucePerens — Fri, 30 Nov 2007 18:20:30 +0000

Public domain won't give DJB any protection from malicious patent holders. The author of JRMI is feeling the pain from using the Artistic license. He had a manufacturer use his software in a product and sue him for patent infringement. Classic example of what Free Software developers don't want to happen. He's not doing well in court because his license was read by the judge as a broad permission rather than a license at all. He would have been in better shape with a license that protected him from patent aggression by people who redistribute his code.

Bruce

This is a mistake, but he also provided a license, so it doesn't really matter

erwbgy — Fri, 30 Nov 2007 18:12:54 +0000

In Placing documents into the public domain DJB says:

"I've seen a few people claiming, without justification, that a clear written dedication of the work to the public domain doesn't actually abandon copyright. Nobody, to my knowledge, has ever wasted a judge's time trying to make this silly argument in court."

This (short) document explains that you can abandon copyright in Europe or the USA by placing something into the public domain.

not necessarily

JoeBuck — Fri, 30 Nov 2007 17:27:17 +0000

It used to be common in the early 90s, before people were as sophisticated as they are now about licenses, to see code posted on Usenet with a comment like "this code is in the public domain. You can do anything you want with it as long as you don't make money off of it". That's contradictory, of course, and there's ever a dispute, a judge might find that the second statement trumps the first.

If DJB's statement placing the code in the public domain contains any words indicating other terms, then he hasn't really put it in the public domain. I haven't parsed his message carefully, so I don't know.

The main reason for a software author to prefer the BSD license to "public domain", especially if he/she is an American, is liability. The BSD license has a loud "no warranty" clause; without terms like this, if a defect in DJB's software causes anyone any harm, it's conceivable that DJB could be held to be legally responsible.

This is a mistake, but he also provided a license, so it doesn't really matter

charlieb — Fri, 30 Nov 2007 17:01:29 +0000

Providing a license is mutually exclusive with placing into the public domain.

You are entitled to your opinion as to what DJB should have done, but the decision is his to
make. He's not "in error" - he's just done something other than what you would have done.

> ... though this requires interpreting what he means by "etc."

"etc" is not defined (no longer defined) by DJB, it's defined by the legal definition of
"public domain".

Huh?

thumperward — Fri, 30 Nov 2007 16:15:36 +0000

How is it "always best to provide a license"? I've heard the argument that a contrivance like
the three-clause BSD ensures that the code is available even in states which do not recognise
public domain, but I've never heard of anyone who experienced this difficulty.

SQLite is also PD and has considerable uptake in large free software projects, commercial
endeavours and in US government software. It doesn't appears to have been hindered by being
PD. Do you have anything to back up your claim that it would have been "simpler", and that it
would have "avoided needless confusion", by picking a contrived set of terms instead of a
perfectly valid copyright status enshrined in law?

Furthermore, BSD explicitly ties a work to an author. Perhaps this was the precise reason for
choosing not to hold copyright over the work in this case.

 - Chris

qmail released into the public domain

lkundrak — Fri, 30 Nov 2007 15:36:10 +0000

In my humble opinion he shouldn't ever release his software to public, in any form :)

This is a mistake, but he also provided a license, so it doesn't really matter

shieldsd — Fri, 30 Nov 2007 15:29:15 +0000

I suggest Mr. Bernstein is in error in releasing the qmail code as "public domain." 

"Public domain" is, roughly speaking, that which has no copyright. Many folks feel this is the
best way to make code available in the most open form as possible.

However, it is always best to provide a license. In the present case, Mr. Berstein writes, "I
hereby place the qmail package (in particular, qmail-1.03.tar.gz, with MD5 checksum
622f65f982e380dbe86e6574f3abcb7c) into the public domain. You are free to modify the package,
distribute modified versions, etc." In effect he is both releasing the code into the public
domain, and also providing the code under terms that are close to those of what is known as
"BSD-style" license, though this requires interpreting what he means by "etc."

Simply using the BSD license would have been simpler, and would have avoided needless
confusion. Hopefully Mr. Bernstein is a better programmer than he is an authority on
open-source licensing.

Then again, there is a comparable program, Postfix, that is available under an open-source
license, the IBM Public License. It was written by an IBM colleague, Wietse Venema. See
http://postfix.org.

thanks,dave
http://daveshields.wordpress.com

qmail released into the public domain

mbottrell — Fri, 30 Nov 2007 14:34:06 +0000

meh...  something Daniel should have done a long time ago.

Release into PD or even under a GNU GPL license.

At least now Linux distros will be able to ship the damn thing.

Whilst Qmail was nice (though you had to patch the hell outta it to get it to a workable
solution), it has long since been moved on by many to something that's supported outta the box
by their distro of choice.

Too late, too bad.

finally...

niner — Fri, 30 Nov 2007 14:30:41 +0000

...a clear path to a qmail that at least somewhat works out of the box and integrates at 
least a little with the rest of the system. At least in theory (yes I've given up on qmail a 
long time ago...)