[gentoo-dev] Re: Watch out for license changes to GPL-3.

[Duncan <1i5t5.duncan@cox.net>]

Richard Freeman <rich@thefreemanclan.net> posted
469129CF.9040301@thefreemanclan.net, excerpted below, on  Sun, 08 Jul 2007
14:15:43 -0400:

> Seemant Kulleen wrote:
>> If you can really show some way that GPL3 provides a compelling case to
>> move to it, then we can start talking about that.
>> 
>> 
> I wasn't aware that gentoo practiced copyright assignment.  You might
> want to make the disclaimers clear - if somebody submits a patch on
> bugzilla and doesn't expressly assign copyright they would legally
> retain it unless it were a clear condition of using the site.  Also, it
> would help avoid people submitting patches that aren't
> GPL-2-only-compatible from other projects.  But then again, I'm not a
> lawyer...  :)

Choosing here to jump in, tho this could go elsewhere in the thread.  
I've done a bit of research on this for my own (scripted) code.

Trivial isn't copyrightable.  It has to express creativity and etc.  
There's a bit of a gray line as to what's "trivial" vs what's not, but 
the position the FSF takes is that if it's just a few lines, it's 
"trivial".  I've seen numbers thrown around as low as three lines or as 
high as 20, on the "arguing on the low side" end (so some saying as low 
as 20 may consider the norm higher but admit there might be a /few/ cases 
for as low as 20 lines).

More specifically, in their licensing recommendations, the FSF suggests 
that it's /not/ appropriate to use the GPL/LGPL on works short enough 
that incorporating the whole of the license would make the license the 
bulk of the work in question.  They strongly recommend that works 
incorporate the whole license in word, not just by reference as to a URL 
or the like, since those change over time.  (This is in contrast to the 
CC licenses, which encourage incorporation by URL reference, and pledge 
to keep a more or less stable URL for each version.)  The FSF says on 
such short works, it's better to release "in the public domain" or under 
some other less restrictive license.

Thus the questions of whether many/most individual ebuilds /could/ be 
copyrighted or if so whether it's worth doing so.  Certainly, it's the 
tree that contains the license, not the individual ebuilds, etc, which 
give the copyright statement but little more.  Gentoo policy would seem 
to be, then, that it's the work of the tree as a whole that's 
copyrighted.  Individual ebuilds may or may not be, and it's /implied/ 
(which isn't necessarily legally binding) that if they are, there'd be 
little attempt at enforcement unless a significant portion of the tree 
was copied/modified.

Of course, there's also the question of whether an individual ebuild is 
all that useful in practice, without the rest of the supporting tree 
structure (not necessarily the individual applications including those 
developed by Gentoo such as portage, the tree).  Certainly without the 
eclasses, many ebuilds would be in practice almost worthless.

So the copyright is on the tree.  Note that actual Gentoo apps such as 
portage, catalyst, etc, are copyrighted individually.  The Gentoo policy 
/does/ state that apps are GPL2ed AFAIK, as is the tree.  Then there's 
documentation, which is not GPLed but generally CC-AT-SA (Attribution 
Share-alike).

> I guess one reason to move would be that it is the goal of the FSF for
> this to become the "default" GPL.  So, if there was a compelling case
> for adopting the GPL at all (one presumes there was since we're GPL
> currently), then there is a case for migrating to GPL v3 by that virtue
> alone.  Does that mean that we HAVE to?  Certainly not.
> 
> I'd ask the question why we're GPL at all?  If the reason is because we
> generally agree with the principles of free software and copyleft, then
> the GPL v3 is only an improvement over the GPL.  If we don't really like
> copyleft as an organization then it would make more sense to just adopt
> BSD, rather than stick with a copyleft license that just has a few
> loopholes in it.

That's a long and predictably controversial debate.  See all the 
electrons spilled on it debating the Linux kernel, for instance.  While I 
personally support the FSF and GPL3, there's a definitely valid position 
held by some that the code return requirements of GPL2 are sufficient, 
that Tivoization should be specifically allowed, because the code is 
returned, even if it doesn't work on their specific product without the 
signing keys and etc.

Apart from the more specifically enumerated patent protections and wider 
compatibility of GPL3, which might be worthy shooting for, I don't think 
the anti-tivoization clauses are much that Gentoo needs to worry about 
for the tree (possibly for some of the apps) anyway.  Of course, there's 
also the point that what's in the tree is scripted and therefore 
inherently in source form, and that changing it sufficiently to put it in 
compiled language form would be a rewrite and of questionable "derived" 
status.  Certainly, the work to put it in compiled form would be 
significant.  It's also not likely as the scripted form is a major part 
of the point.  If it were compiled and therefore more opaque, it'd lose 
the distinctiveness that makes it Gentoo and is coming close to being any 
other (binary) distribution.

There's also the hassle of changing.  Many contributors could argue that 
they contributed under the statement that it'd be GPL2, period.  How that 
might turn out is anyone's guess, but I just don't see that there's any 
benefit in moving the tree to GPLv3, with the possible exception of 
patent protections and I don't believe they are likely to be worth the 
switch on their own.  Thus, for the tree as a copyrightable work, I just 
don't see it being worth even attempting to change.

The case could however be made for portage, catalyst, etc, all the Gentoo 
apps.  They have a narrower contributor base so the hassle of switching 
should be less.  Their usage is such that the GPL3 may arguably be of 
benefit over the GPL2.  However, I know little of the feelings of the 
major contributors.  If they don't feel it worth switching, or are 
definitely against it, it's unlikely to happen.  If they favor switching, 
with the narrower contribution base, it might indeed be possible, and the 
benefits could indeed arguably outweigh the cost.

> In terms of pros/cons with GPLv2 you'd have compatibility with GPL3 and
> GPL2+ licenses, as well as the the Affero GPL.  There is of course the
> closing of the tivoization loophole, and that can be considered a pro or
> a con depending on your personal beliefs.  However, if you really are
> pro-tivoization, then why use the GPL at all?
> 
> Oh, there exists another option - we could also relicense as GPL2 or 3 -
> that gets rid of the "what if it changes to something bad" issue while
> allowing others to adopt the code under either license.

The 2/3 option will have lower cost than 3 only, certainly, as those who 
favor GPLv2 are less likely to be strongly opposed to the dual license 
2/3.  However, again, I don't see it being even worth serious 
consideration (beyond the current thread level) for the tree.  Possibly 
for one or more of the apps, but not the tree in general.

Again, I'm generally pro-GPLv3 switch, but an optimistic realist as 
well.  If a generally pro-GPLv3 guy doesn't see it as worth switching the 
tree, I don't believe it's going to happen, period, because there are 
certainly those that are more adamantly GPLv2 only than I am GPLv3 only, 
and they have the present situation on their side.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

-- 
gentoo-dev@gentoo.org mailing list