headers
Eugen Leitl | 9 Mar 2012 12:02

questions about GPL again


A senior guy from a rather well-known chemical abstracs service
just made the claim that GPL is contagious even for dynamically linked
code. Which I recall as not being correct.

Can somebody point me towards a definitive document or
piece of writing by a lawyer which definitevely shows
what the actual situation is?
_______________________________________________
Do not Cc: anyone else on mail sent to this list.  The list server is set for maximum one recipient.
linux-elitists mailing list
linux-elitists <at> zgp.org
http://zgp.org/cgi-bin/mailman/listinfo/linux-elitists
Permalink | Reply |
headers
Greg KH | 9 Mar 2012 15:59
Gravatar

Re: questions about GPL again

On Fri, Mar 09, 2012 at 12:02:03PM +0100, Eugen Leitl wrote:
> 
> A senior guy from a rather well-known chemical abstracs service
> just made the claim that GPL is contagious even for dynamically linked
> code. Which I recall as not being correct.

What do you mean by "contagious"?  And which version of the GPL are you
referring to?

> Can somebody point me towards a definitive document or
> piece of writing by a lawyer which definitevely shows
> what the actual situation is?

The GPL itself is very easy to read, have you tried that?

greg k-h
_______________________________________________
Do not Cc: anyone else on mail sent to this list.  The list server is set for maximum one recipient.
linux-elitists mailing list
linux-elitists <at> zgp.org
http://zgp.org/cgi-bin/mailman/listinfo/linux-elitists
Permalink | Reply |
headers
Eugen Leitl | 9 Mar 2012 16:19

Re: questions about GPL again

On Fri, Mar 09, 2012 at 06:59:07AM -0800, Greg KH wrote:
> On Fri, Mar 09, 2012 at 12:02:03PM +0100, Eugen Leitl wrote:
> > 
> > A senior guy from a rather well-known chemical abstracs service
> > just made the claim that GPL is contagious even for dynamically linked
> > code. Which I recall as not being correct.
> 
> What do you mean by "contagious"?  And which version of the GPL are you

In the corporate wonk oh-noes-we-have-to-opensource-everything-touched-by-GPL
sense.

> referring to?

They cluster v1/v2/v3 as one for all practical purposes. Is the
dynamic-linking of GPL libraries by proprietary code treated
differently across v1 to v3? 

> > Can somebody point me towards a definitive document or
> > piece of writing by a lawyer which definitevely shows
> > what the actual situation is?
> 
> The GPL itself is very easy to read, have you tried that?

Apparently (according to many online writing) many people 
find it confusing enough that they think ability to dynamically 
link (=just call a GPL library) requires a lawyer to interpret
how things are in your jurisdiction. 

I'd rather like to be able to cut through the FUD authoritatively. Is there
(Continue reading)

Permalink | Reply |
headers
Greg KH | 9 Mar 2012 16:34
Gravatar

Re: questions about GPL again

On Fri, Mar 09, 2012 at 04:19:07PM +0100, Eugen Leitl wrote:
> On Fri, Mar 09, 2012 at 06:59:07AM -0800, Greg KH wrote:
> > On Fri, Mar 09, 2012 at 12:02:03PM +0100, Eugen Leitl wrote:
> > > 
> > > A senior guy from a rather well-known chemical abstracs service
> > > just made the claim that GPL is contagious even for dynamically linked
> > > code. Which I recall as not being correct.
> > 
> > What do you mean by "contagious"?  And which version of the GPL are you
> 
> In the corporate wonk oh-noes-we-have-to-opensource-everything-touched-by-GPL
> sense.

That baseless rumor is still spreading around?  Please just ask anyone
who is making such a statement to show the relevant part of the GPL
itself that says that.

> > referring to?
> 
> They cluster v1/v2/v3 as one for all practical purposes. Is the
> dynamic-linking of GPL libraries by proprietary code treated
> differently across v1 to v3? 

There really isn't any v1 code anywhere in the world, so don't ever
worry about that.

And if you are referring to libraries, don't you mean LGPL?  That's
usually the license for them (with some well known exceptions, i.e.
readline).
(Continue reading)

Permalink | Reply |
headers
Ben Finney | 9 Mar 2012 23:54
Picon

Re: questions about GPL again

Greg KH <greg <at> kroah.com> writes:

> And if you are referring to libraries, don't you mean LGPL? That's
> usually the license for them (with some well known exceptions, i.e.
> readline).

There's no necessary connection, though. Many libraries are released
under GPL.

The FSF is not the only copyright holder releasing works under the GPL,
as you know, so the FSF's policy on which licenses to use is not really
relevant here.

For that reason, I'm glad the FSF re-named the LGPL to reduce the
implication that discourages the GPL for libraries.

> The simple answer is, if you dynamically link in GPL code (note, not
> LGPL) with your propritary application, you just violated the license
> of the library. It's really that simple.

Yes. And that's a good reason to, in some cases, use the GPL for a
library. So perhaps we should be less surprised at the suggestion that
people do use the GPL for libraries.

> If someone is worried about what happens "when we get caught" that
> implies that they know they are doing something wrong, and shouldn't
> be doing it in the first place.

To sympathise with Eugen's querent for a moment: it seems more likely to
me that they are concerned because they *don't know* whether they are
(Continue reading)

Permalink | Reply |
headers
Rick Moen | 9 Mar 2012 18:33
Favicon

Re: questions about GPL again

Quoting Eugen Leitl (eugen <at> leitl.org):

> In the corporate wonk oh-noes-we-have-to-opensource-everything-touched-by-GPL
> sense.

Which is crap.

The remedies available to plaintiff for copyright violation (_if_
proved to the court's satisfaction) are cessation of violation and
payment of damages, the latter of which can be a lot higher if the 
author paid to register the copyright, as is not often done.  There is
no precedent whatsoever for violator being ordered to give plaintiff
access to violator's other copyrighted creations.  None.

The example usually alluded to (usually inaccurately) is a case in
point, that of NeXT, Inc.'s binary-only ObjC frontend code for gcc.
Contrary to the narrative usually claimed, NeXT wasn't ordered to hand
over source code and wasn't even hauled into court.  FSF merely politely
reminded NeXT that their method of distribution was in violation of FSF
copyright.  NeXT quickly figured out that their reasonable options were
to either cease shipping the offending code (and maybe get sued for
damages, but probably not) or to voluntarily provide the source code and
thus retain the ability to ship it rather than having to find or code or
buy a substitute for gcc.  They elected to release code -- entirely for
reasons of corporate interests.

> I'd rather like to be able to cut through the FUD authoritatively. 

Read caselaw about what 'derivative work' does and does not mean in the
broad context of copyright law.  In US jurisdictions, you would start
(Continue reading)

Permalink | Reply |
headers
Rick Moen | 10 Mar 2012 13:21
Favicon

Re: questions about GPL again

Attempting to elaborate, and get more directly to Eugen's concern:

Quoting Eugen Leitl (eugen <at> leitl.org):

> I'd rather like to be able to cut through the FUD authoritatively.

Meaning no criticism whatsoever, of Eugen, he didn't exactly say what
specifically the recalcitrant faction at his firm think the problem
scenario might be (except that it involves 'we have to opensource 
everything touched by GPL').   That makes it difficult to address, beyond
pointing out that forced relicensing of another copyrighted property
cannot ever be cajoled out of any court, as that is simply _not_ a
remedy for (hypothetical) copyright violation, period.

So, as I detailed, the _particular_ fear of a non-existent remedy being
imposed for copyright violation is rubbish.  Doesn't happen, and the
people who assert that it do have failed to understand civil litigation.
However, the remaining question is:  What's the risk of copyright
violation, given that the _real_ possible remedies of court-ordered
cessation of infringement and paying some monetary damages aren't much
fun, either?

The intended scenario might be:  GPL-covered binary permitted merely to
run on same machine as precious company proprietary jewel code does.
E.g., porting over said precious company proprietary jewel codebase to
run natively on the GPLv2-covered Linux kernel and LGPLed glibc.  If
that's the scenario, the quick answer is:  How 'bout that Larry Ellison
guy?  Must be a really reckless guy, and going to be in horrific trouble
any day, now.  {crickets}
(Continue reading)

Permalink | Reply |
headers
Jeremy Hankins | 9 Mar 2012 16:46

Re: questions about GPL again

Greg KH <greg <at> kroah.com> writes:
> On Fri, Mar 09, 2012 at 12:02:03PM +0100, Eugen Leitl wrote:

>> Can somebody point me towards a definitive document or
>> piece of writing by a lawyer which definitevely shows
>> what the actual situation is?
>
> The GPL itself is very easy to read, have you tried that?

As I understand it (ianal, of course) Eugen's question isn't really
about the GPL at all, but about copyright law.  The question is whether
dynamically linking constitutes creating a derivative work.  If so, then
the constraints of the GPL would apply to that derivative work.

I have nothing to say on that question other than that, so far as I
know, this isn't a question that's been settled in court.  So it's
possible to argue over it, though most people behave as if dynamic
linking doesn't create a derivative work.

I have absolutely no idea whether it's worth it for companies to worry
about GPL "contagion" on the off chance that the courts surprise
everyone and decide that dynamic linking creates a derivative work under
copyright law.  But if the courts did make such a decision a lot of
people would be very unhappy, which I personally think is good reason to
think it's not likely.

I seem to recall that the FSF has at some point come out with a position
on this, but I'm not sure and can't be bothered to look for it atm.  But
if they have, that might be the sort of evidence you're looking for,
Eugen.
(Continue reading)

Permalink | Reply |
headers
Jeremy Hankins | 9 Mar 2012 16:49

Re: questions about GPL again

Er, ignore my last message.  I was totally not thinking clearly.

--

-- 
Jeremy Hankins <nowan <at> nowan.org>
_______________________________________________
Do not Cc: anyone else on mail sent to this list.  The list server is set for maximum one recipient.
linux-elitists mailing list
linux-elitists <at> zgp.org
http://zgp.org/cgi-bin/mailman/listinfo/linux-elitists
Permalink | Reply |
headers
Shlomi Fish | 9 Mar 2012 16:22
Favicon
Gravatar

Re: questions about GPL again

Hi,

On Fri, 9 Mar 2012 12:02:03 +0100
Eugen Leitl <eugen <at> leitl.org> wrote:

> 
> A senior guy from a rather well-known chemical abstracs service
> just made the claim that GPL is contagious even for dynamically linked
> code. Which I recall as not being correct.
> 
> Can somebody point me towards a definitive document or
> piece of writing by a lawyer which definitevely shows
> what the actual situation is?

there’s this document:

http://www.gnu.org/licenses/why-not-lgpl.html

“Why you shouldn't use the Lesser GPL for your next library” by Richard M.
Stallman.

If you could use dynamic linking (instead of static linking) to avoid having to
release your code under a FOSS, GPL-compatible licence, then I don’t see the
point of this recommendation, because most libraries nowadays are also usable
when compiled as shared libraries and dynamically linked.

The NMAP developers have gone past that and claim that running nmap, capturing
its output and processing it also falls under the GPL:

http://nmap.org/svn/COPYING
(Continue reading)

Permalink | Reply |

Gmane