[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug #189164: libdbd-mysql-perl uses GPL lib, may be used by GPL-incompatible apps



On Tue, May 27, 2003 at 12:22:35PM -0400, Nathanael Nerode wrote:
> Anthony DeRobertis said:
> > I'm not sure if you're thinking of this when mentioning "public 
> >domain", but many header files (for example, ones giving simple structs 
> >and numeric defines) probably have no copyrightable work in them, and 
> >thus would be essentially in the public domain. So, using those is 
> >fine, no matter what the copyright notice says.
> I was, indeed, thinking of this. :-)

> > I point out Lotus v. Borland[0] and note that the commands used by the 
> >shell script are the same as used by a human, and thus are a method of 
> >operation, not protected by copyright.
> This is an interesting case, the details of which I was not aware.

> This case leads to the following test:  if you are 
> accessing a program through an interface which is a "method of 
> operation", referring to the "means by which a person operates 
> something", you incur no obligations, because that is not copyrightable.

> There are various possible extrapolations from this.

> First, any interface which could be used by humans is a "method of 
> operation".  This is essentially all interfaces.  Since dynamic linking 
> involves the copying of small, (usually) uncopyrightable, code segments, 
> together with the use of an interface, dynamic linking incurs no 
> obligations, and the FSF's interpretation is quite wrong.

This assumes that the FSF's interpretation depends on the claim that
dynamic linking creates a derived work.  While varies parties have
claimed this at one point or another, I have argued that the dynamically
linked work is under the purview of the GPL by virtue of the license
terms alone, *if* you are distributing the GPL library in question,
which is always the case for Debian.

There have been many well-constructed arguments here that writing to an
interface does not create a derived work, and is not subject to the
copyright of the library because the only parts being copied are not
copyrightable.  However, the advertising materials for a product are not
derived works of a software library either, yet there are software
licenses that place restrictions on the contents of advertising
materials for products containing the software -- *and* we accept such
licenses as free.  "You have no say in it, it's not a derived work" is a
null argument, because as distributors we must accept the terms of a
license that *gives* them a say in it.

The greater question for Debian, I think, is whether such an
interpretation can be regarded as DFSG-compliant.  I believe the answer
is yes: first, because it is not *merely* the act of co-distribution
that triggers this "license contamination", so DFSG 9 does not seem to
apply; second, because this interpretation of the GPL's requirements has
existed for a long time, and I'm not aware of any formal objections to
its DFSG-freeness even though it depended on a definition of "derived
work" that varies from country to country, so I don't think it suddenly
becomes non-free just because we've shown it isn't copyright law
directly that's used to enforce the requirement.  There is a certain set
of freedoms that we demand from Free Software, some of which are granted
to the user by default in copyright law and some of which are not; so
long as these core freedoms are preserved, I don't think changes in
copyright law or its interpretation should render a license non-free.

Granted, this rule of thumb has its limitations -- a change in copyright
law that eliminates all use rights by default would certainly affect the
landscape -- but I think it applies here.

-- 
Steve Langasek
postmodern programmer

Attachment: pgpyL5MtMVjtF.pgp
Description: PGP signature


Reply to: