[DB-SIG] Re: [Psycopg] GPL or LGPL

Federico Di Gregorio fog@initd.org
02 Oct 2002 09:15:00 +0200

Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable

Il mer, 2002-10-02 alle 08:52, Magnus Lycka ha scritto:
> At 01:19 2002-10-02 +0200, Federico Di Gregorio wrote:
> >let's restate the problem: bash is GPL'ed. should *any* sh script be
> >released under the GPL? or only the ones using bash extensions to
> >standard bourne shell? and why is python (and its extension psycopg)
> >diferent from a shell?
> Do you suggest that a python program that is written
> for vanilla DB-API 2 and happens to work with psycopg
> could be proprietary, but a python program that uses
> unique features in psycopg can not?
> The issue is (I suppose) whether the python program
> is a derivate work of psycopg.

is the following script a derivative work of bash and should then be
placed under GPL?

pushd /tmp && popd

> I don't think there is any doubt that python code that
> uses unique features of psycopg must be considered a
> derivate work, just as a novel about the hobbit Frodo
> is a derivate work of Tolkiens books, and thus a
> copyright infringement if published without permission.

i think there are *big* doubts. python bytecode and the (possibly
dynamically linked) python executable are on different domains. nobody
really thinks a single, proprietary bash script breaks bash license.
now, tell me *where* python is different from bash. your program
bytecode *uses* psycopg, it does not link with it. not even on "import",
import just instructs the python interpreter to link itself agains some

Federico Di Gregorio
Debian GNU/Linux Developer & Italian Press Contact        fog@debian.org
INIT.D Developer                                           fog@initd.org
              All programmers are optimists. -- Frederick P. Brooks, Jr.

Content-Type: application/pgp-signature; name=signature.asc

Version: GnuPG v1.2.0 (GNU/Linux)