[Nix-dev] Re: Specifying licenses on Nix packages

[Sander van der Burg - EWI]

My goal is not to develop a system that could verify everything and see whether all the licenses are used properly or not. I know this is only possible when a license is defined in a formal language which we can use to verify a system, which is not the case in practice.

What I do want to solve is that we currently have no notion at all about what licenses are applicable to certain components and what eventual problems/issues there are.

The advantages/benefits I see of using these notations and conventions is:

- We have a notion of what licenses there are specified or what licenses there are missing on packages in the nix repository. Currently, we don't have any notion.
- We could immediately see when we install a package whether or not it matches a policy, such as installing a non-free package or a package with a "problematic/vague" license. e.g. a user could decide not installing packages that are (partially) non-free (e.g. a package available under a proprietary license)
- We have some (but no complete) insight why a composition is possibly non-free.
- We could decide to disable certain builds on the Hydra that we don't want to distribute based on some custom defined constraints

I know that there are a lot of cases that we cannot solve, but by having a convention for specifying licenses we at least have some insight/notion in these problems, which is better than not having it all IMHO.

As a start we need to specify these license more concise I think, which I proposed in my initial message. From these conventions we can derive other possible use-cases. I know this will not solve all our problems/issues, but it will give us more insight in the licensing structure of Nixpkgs and also offers us more options in dealing with them.

-----Original Message-----
From: nix-dev-bounces at cs.uu.nl on behalf of Ludovic Courtès
Sent: Mon 2/22/2010 4:53 PM
To: nix-dev at cs.uu.nl
Subject: [Nix-dev] Re: Specifying licenses on Nix packages
 
Hi again,  :-)

ludo at gnu.org (Ludovic Courtès) writes:

> What are you replying to?

For some reason, your initial message didn't make it through Gmane (!).
For those who missed it, here it is:
http://mail.cs.uu.nl/pipermail/nix-dev/2010-February/003911.html .

We quickly discussed it at FOSDEM, but I'd like to summarize my
position.

If find the "license calculus project" you have in mind interesting, but
I'm skeptical about its practicality.  The reason is that this whole
idea assumes that legal texts written in natural languages, subject to
human interpretation, can be reasoned about in a mathematical way.
I think this assumption doesn't hold, unfortunately.

Even if the interpretation of law and licenses were unambiguous, stable
in time, context-independent, etc., other practical issues would make it
very hard to automatically reason on software licensing.

A few examples:

  * What is "the license" of OpenOffice.org?

  * What is "the license" of Teeworlds?  (It's home-made.)

  * What is "the license" of GNU Guile once you've typed
    '(use-modules (ice-9 readline))'?  (libguile is LGPLv2+, but that
    module is GPLv3+ as it dlopens GNU Readline.)

  * Is the Open Font License (OFL) FSF-free? (See
    http://thread.gmane.org/gmane.linux.distributions.gnu-linux-libre/406
    for a discussion.)

  * Can you write a predicate that tells whether a given library is a
    "System Library" according to the GPL?

    ...

As a Free Software supporter, I'm taking the straightforward approach of
removing software that doesn't correspond to my policy:
http://repo.or.cz/w/nixpkgs-libre.git .  Join us now, share the
software!  :-)

Thanks,
Ludo'.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.science.uu.nl/pipermail/nix-dev/attachments/20100222/eb0b98c0/attachment.html