linking BSD and GPL code via a plugin

[Chris Travers]

On Wed, Jan 19, 2011 at 9:07 AM, John Cowan <cowan at mercury.ccil.org> wrote:
> Kevin Hunter scripsit:
>
>> I'm confused (yet again!) about linking to libraries and other programs.
>> I am working with a BSD licensed project, and I've been tasked to
>> write a plugin to call functions of a GPL project.
>
> The FSF claims that such a plugin is a derivative work of the GPLed
> code, unless there is a non-GPLed version (separate implementation or
> differently licensed copy) that is plug-compatible with the GPLed code.
> Thus code written to use readline does not have to be under the GPL, given
> that it can work just as well with the equivalent BSD library editline.

Of course, such a view, if upheld by the courts, would require
permission from Microsoft for any piece of software written for
Windows unless that program ran perfectly under WINE.
>
> Whether there is legal warrant for this claim is more than doubtful, as
> ordinary copyrighted works such as books do not normally "interoperate" in
> this way.  One hypothetical would be this: given a book like a textbook
> that asks questions of the reader, is another independently written
> book that answers those questions a derivative work of the first book?
> I know of no cases on point.
>

There have been a number of law review articles which have discussed
this issue and every one that I have seen has taken a dim view of the
FSF's expansive view of copyright law at least in the US.  For one
thing, current copyright law in the US only protects expressive
elements which are not too closely tied to practical elements.
Interoperability of software in fact has had a fairly large number of
cases involving attempts to control it via copyright law.  Many of
these have been resolved by looking to fair use (Sony v. Connectix),
some via de minimis considerations (Lexmark v. Static Control), etc.
The really interesting cases which are directly on point never got
resolved as a matter of law though.

More interestingly, the cases which have gone the other way (Midway v.
Arctic for example) have involved other issues, such as the protection
of audiovisual copyrighted works separate from the software copyright.

IANAL, but I think it is very safe to say that the GPL is not as
restrictive as the FSF claims in areas of interoperability.

However, outside the legal side, there's also the social side.  Most
of what we gain from doing FOSS is directly related to having a good
name, and the goodwill of the community.  Consequently it's important
to seek to conform to community expectations regardless of the state
of the law.

However, to my knowledge, there are no community expectations against
taking your own contributions and releasing them with more permissions
than the work as a whole might be expected to have.  The GPL presents
a floor, not a ceiling, in terms of permissions.  While the GPL v3
raises some questions about this in terms of what RMS calls
"relicensing" requirements in terms of compatibility, current
community norms are generally a bit more reasonable.

In general, communities tend to be pretty reasonable about accepting
BSD-licensed contributions into GPL'd code.  Even if they didn't
choose to be reasonable, it's not clear who could enforce a copyright
claim against your contributions except you anyway.

> In any case, it's not about who distributes what, it's about whether and
> when one piece of code is a derivative of another when it contains none
> of the other.  Clearly in some cases a derivative work is formed: if you
> translate an Albanian novel into French, the French work is derivative
> of the Albanian work even though it contains no words from the original.
> In other cases, it's quite different: we can both draw the Eiffel Tower,
> or even photograph it, and neither work infringes the other no matter
> how closely similar they are, provided there was no intent to make a copy.
>
> Sorry not to be more definite, but the state of the law is not definite.
>
> --
> John Cowan            http://www.ccil.org/~cowan     cowan at ccil.org
> Uneasy lies the head that wears the Editor's hat! --Eddie Foirbeis Climo
>