LWN: Comments on "GPLv3 and the kernel"

Problems with Linus are to be expected...

sammythesnake — 2006-02-10T17:33:34+00:00

It's true that "laziness" is a key feature of Linus' character, but I submit that it's a good thing, given that it's generally quite well marshalled. Laziness leads (via wisdom) to efficiency, and the evidence so far seems to be that Linus' laziness has actually had exactly that effect! :)

I remember reading in some o'reiley book that the three most important characteristics to have as a programmer included Hubris and Laziness (couldn't remember the third, but http://www.io.com/~shiva/ninevirtues.html suggests impatience)

Cheers & God bless
Sam "SammyTheSnake" Penny

Improvement

forthy — 2006-02-09T09:56:03+00:00

GPLv3 is still a copying license, not a usage license. It allows usage restrictions to be compatible with licenses like ApacheL or CDDL from Sun. This is an optional "module" of the license.

The DRM part is non-optional, because I'm quite convinced that Linus simply is wrong, and the majority is correct, that the anti-DRM statement really is already in GPLv2, though not explicitely. Linus is just telling blue-eyed rubbish. That includes himself putting the "GPLv2 only" notice into COPYING. Didn't he read the "Thou shalt not alter the COPYING file" part of the GPL? It's right below the copyright statement!

"or any later version"

pjgrok — 2006-02-07T22:36:41+00:00

If you'd like to write about this subject for Groklaw, would you please contact me? I would love
to have an explantion developers can understand why this language isn't to be feared. It was, I
understand, such fear that caused Linus to avoid the wording in the first place, and look at all the
issues that resulted from that.

You're a contract lawyer, so you understand what the wording means, and how little it means,
but to developers it can look spooky, and it is my belief that it matters that developers and
lawyers figure out how to communicate better to avoid unncecessary issues.

Hope you don't mind, Jon, and if so, publish it here. I don't care, as long as it is published. I'll
gladly link to it instead. Either way is fine with me.

"or any later version"

giraffedata — 2006-02-04T01:54:10+00:00

Yeah, for a contract that's true. I'm a contract lawyer. When you agree to abide by a document you haven't seen, you aren't normally held to that (but if you see it soon after and don't object, and it's a customary sequence of events, you become bound -- that's why shrink wrap agreements work).

But there's no contract here. This is a unidirectional license binding only on the licensor, and the rules are probably different.

And it's not like licensees can lose anything. If I get Linux under GPL2 today with a "or any later version" clause, and FSF produces GPL3 next year, it doesn't take anything away from me -- it just means I have one more option if I want to redistribute my copy. The person it hurts is the copyright owner who now has a little less control over his code because I have more options.

A contract can refer to future events that are uncertain, even if they are under the control of some uninterested third party. Like when the interest rate on a loan is based on some number published in the Wall Street Journal each month. So one could probably say "whatever the FSF thinks is a fair license" and still be enforceable.

"or any later version"

shane — 2006-02-03T11:00:53+00:00

I occasionally wonder about the "or any later version" clause.

IANAL, but I thought that you could not agree to something that you had
not yet seen. For instance, when you start a job and you are asked to sign
a paper saying that you agree to abide by a human resources policy that
you have not been given.

This clause strikes me as similar. GPLv4 could read, "the code may be
hidden and sold for profit by companies building land mines who have given
more than $10000 to Nigerian banks". (Okay, not very likely.) But the idea
that you are agreeing to some future, unknown license seems like it would
add confusion if at some point the copyright holder did NOT agree to the
future license.

Problems with Linus are to be expected...

hingo — 2006-02-03T09:01:41+00:00

I attended an Open Source seminar last Wednesday, and heard the following rather to the point quote: "The comments by Linus on GPL3 should be understood in the context that the new license is quite long and wordy and Linus is very lazy." ;-)

Problems with Linus are to be expected...

zblaxell — 2006-02-03T03:29:34+00:00

Linus follows an even/odd release cycle. Even-numbered months are stable, but during odd-numbered months Linus undergoes a lot of active development and never enough testing. Sometimes serious bugs are introduced during odd-numbered months, causing Linus to produce incorrect or unexpected results. During even-numbered months, tireless Linus maintainers work to fix the bugs and produce a stable kernel developer, then an odd-numbered month starts and the whole process starts over again.

Note that in the current case and the 4GB highmem case, Linus's categorical statements were issued in odd-numbered months. The highmem stuff IIRC was merged in an October, an even-numbered month, when Linus had been better debugged and was a lot more stable.

;-)

Improvement

liamh — 2006-02-03T00:44:17+00:00

Maybe. But after reading his comments I find myself mostly agreeing with his reasoning. And even before this, it did bother me that the GPL appeared to have made a shift from being only a copying license to also being a usage license. This is a worrysome jump. However, as almost no one has mentioned, other than (of course) LWN, this is a draft --- the v3 license doesn't exist yet. The draft was made public for precisely these kinds of discussions.

Improvement

cyd — 2006-02-02T22:19:51+00:00

Or maybe Torvalds is simply spouting rubbish, which is not unknown for him.

GPLv3 and the kernel

Baylink — 2006-02-01T15:29:53+00:00

"Linus groupie RSS feed".

/me chuckles

GPLv3 and the kernel

nurhussein — 2006-02-01T06:52:50+00:00

LKML.org has a Linus groupie RSS feed which grabs the latest Linus Torvalds emails from lkml:

http://lkml.org/groupie.php

Improvement

man_ls — 2006-01-31T22:17:33+00:00

And, don't forget, it can lead to a better GPLv3. Torvalds' rejection surely has some legitimate grounds: maybe there is something wrong with the ideas behind the changes, maybe just the wording is not clear enough.

Marketing

darrint — 2006-01-31T19:19:52+00:00

True enough. Linus "rejecting" GPLv3 is only going to increase the attention focused on it, on the Linux kernel, and ultimately, on some of the finer points of software licensing.

GPLv3 and the kernel

smitty_one_each — 2006-01-31T17:51:47+00:00

The free advertising alone is well worth saying a firey thing or five.
Who says the techies can't market?

GPLv3 and the kernel

dankohn — 2006-01-31T17:25:54+00:00

Here's a procmail formula for those Linus-hungry journalists (and others curious about what he's saying). It forwards his posts to a different address (I use a Bloglines email subscriptions folder), and kills everything else. It's a nice complement to, but of course no substitute for, LWN.

:0
* ^X-Mailing-List:.*linux-kernel@vger\.kernel\.org|\
  ^X-Mailing-List:.*git@vger\.kernel\.org

  {
   :0
   * ^From:.*torvalds@osdl\.org
   ! forwarding-address@example.com

   :0
   /dev/null
  }