Il mer, 2004-04-07 alle 16:14, W. Borgert ha scritto: > On Wed, Apr 07, 2004 at 11:47:20AM +0200, Federico Di Gregorio wrote: > > i've read the FSF writings about GPL and interpreters multiple times and > > still i don't understand the exact meaning (or even what they would like > > to enforce). _my_ interpretation is that as long as you don't directly > > link with psycopg and only use the python api you're safe. you can write > > and distribute a proprietary python program as long as python and any C > > module calling psycopg are GPL compatible. you can even have proprietary > > C module in python as long as they don't call psycopg code (by "linking" > > i mean linking, not just coexisting in the same address space :) > > OK, if this interpretation holds, I'll go on in using > psycopg. Note: RMS explicitly does not allow any GNU Emacs > modes that are not GPLed. GNU Emacs modes are written in > elisp, so the situation might be comparable. Confusion? yYou have just one layer with emacs, two with python. In emacs the interpreter itself is GPL'ed. Python is just GPL-compatible but not GPL'ed, so you have: emacs (GPL) <-> elisp code (must be GPL compatible) but: psycopg (GPL) <-> python (GPL compatible) <-> your script (??) Confusion? Yes. Python is used as a bridge. Is this allowed under the GPL? I don't know. The problem is complicated by the fact that you're not calling psycopg directly but a generic API (the DBAPI-2.0) that will be the same even without psycopg. Maybe I'll just license psycopg 2 under LGPL (but what I wanted to avoid using the GPL is people using psycopg code to write proprietary drivers and with LGPL they can do that!) -- Federico Di Gregorio http://people.initd.org/fog Debian GNU/Linux Developer fog@debian.org INIT.D Developer fog@initd.org Viviamo in un mondo reale, Ciccio. -- Lucy
Attachment:
signature.asc
Description: Questa parte del messaggio =?ISO-8859-1?Q?=E8?= firmata