[H-GEN] GPL question

[David Starkoff]

Paul Gearon wrote:

> [ Humbug *General* list - semi-serious discussions about Humbug and     ]
> [ Unix-related topics. Posts from non-subscribed addresses will vanish. ]

(I realise this migrated to H-CHAT, but I thought this was well on-topic 
for H-GEN.)

> People have often told me that it is not possible to link non-GPL code 
> to a library which has been licensed with the GPL.  Hence, many 
> libraries are licensed with the LGPL instead.

In fact, it's not just "people" who say this, the Free Software 
Foundation says so in its FAQ about the GPL: 
<http://www.gnu.org/licenses/gpl-faq.html>.

> Do others agree that this is the case?  For the dissenters, what is 
> your rationale?

The FSF's FAQ says "because the program as it is actually run includes 
the library".

When a program on disk is run in a computer, a copy of the computer 
program (or at least a partial copy) is copied into RAM.  Absent a 
licence, this is an infringement of copyright.  (There's case law on 
this in Australia: Microsoft Corporation v Business Boost Pty Ltd [2000] 
FCA 1651, 
<http://www.austlii.edu.au/au/cases/cth/federal_ct/2000/1651.html>.)

When a program that is linked--whether statically or dynamically--with a 
library, the combined program in RAM will, in all likelihood, be a 
combination of both the program and the library.  If one or the other of 
those components is licensed with the GNU GPL, the entire combination 
must be licensed under the GPL.

As others have noted, clause 0 of the GPL says "[a]ctivities other than 
copying, distribution and modification are not covered by this License 
... [t]he act of running the program is not restricted ...".  However, 
clause 2--the clause that relates to derivative works (such as combining 
a GPL-licensed program, including a library, with something else) says 
"the intent is to exercise the right to control the distribution of 
derivative or collective works based on the Program."

If distribution of a statically linked version of the combined program 
is impermissible due to the GPL, then proposing licit distribution of a 
dynamically linked version smacks of sophistry.  It can be construed as 
an attempt to frustrate or circumvent (the "c" word!) clause 2, which is 
at the heart of the GPL.  There's a tension between clause 0 and clause 
2, sure, but it's not abundantly clear that a strict reading of clause 0 
(you are, after all, just running the program) prevails.  Is it, in 
fact, a de facto distribution of an infringing work?

As far as I know, there's no case law about this.  Which is 
unsurprising, as the FSF's approach, through its counsel, Eben Moglen, 
is to avoid litigation where possible.  That the FSF has been able to 
convince independently advised people that its view is, if not 
incontrovertible, plausible, is to me somewhat persuasive.

I should note that I could be completely wrong about this.  I've read 
long-running and inconclusive mailing list "discussions" about this 
issue before, but I've forgotten most of them.  I'm sure some Googling 
would find them.

David.
(IAAL, but I'm not your lawyer and this isn't legal advice.)