Re: Overview of copyright issues + Debian

[Joseph Wakeling]

Don Armstrong wrote:
> This is now my problem,[1] so I'll attempt to get it addressed at some
> point in the future. [I'd certainly like to see Lilypond at least
> clear up some of the issues so that the above can become correct.]

Hmm, I noted you were listed as the Debian maintainer on Launchpad's
Lilypond page, and brain went off: "The same Don Armstrong as ... ?"
Nice to have you involved in this discussion (and congratulations on
getting your PhD).

Disclaimer: I'm a relatively new contributor to (but long-time user of)
Lilypond, so what I say are my own thoughts and I don't speak for the
Lilypond developer community.

But, since I'm putting myself forward to try and deal with some of this,
any advice you can give would be very welcome.  From my point of view
the DFSG are a very nice benchmark against which to test code and doc
licensing and compatibility is an important aim.

> (There are a significant number of files distributed in lilypond which
> are under v2 or later, or v3 or later, as well as things like
> input/mutopia/claop.py, which isn't even Free Software, as it cannot
> be modified.[2])

If I'm not mistaken, isn't that file only used for a regression test?
How does that affect the situation?

> Furthermore, the licensing statement in COPYING is immensely
> ambiguous, as it only states "GNU PUBLIC LICENSE" without specifying a
> version, or anything.
> 
>>  (1) All new files added to the code or docs must contain an
>>      unambiguous copyright AND licensing notice: I suggest in this
>>      case GPLv2 or later for code, and the GFDL 1.1 or later for
>>      docs.
> 
> I'd personally prefer it if documentation was at least licensed under
> the same license as the code to allow for easily inclusion of code
> examples (and to obviate the problems I [and Debian] have with
> specific aspects of the GFDL.) It certainly can be dual licensed under
> GFDL >= v1.1 + GPL >= v2, though.

AFAIK the docs have always been GFDLv1.1 -- I don't think we can
unilaterally relicense them.  Can you clarify the particular issue with
GFDL?  I thought the Debian consensus was that GFDL is OK as long as
there are no invariant sections.

What does GPL imply for docs?  Would it mean e.g. that someone who
distributes a PDF of the Learning Manual would have to also be prepared
to provide Texinfo sources?

>>  (1) How well have the copyright notices for individual files been
>>      maintained?
> 
> As near as I can tell, they haven't been maintained at all. Very few
> of them actually have the boilerplate that they should have, which
> makes this kind of thing very difficult.
> 
> But by all means, please help work on this. It'll certainly make my
> life easier when I have to go through and audit the code for inclusion
> in Debian (which I naïvely assumed had already been done before I took
> over maintenance.)

What I have noted is that copyright dates have been updated but I'm not
clear whether author names have.

What I propose is that I maintain a separate branch of the code (but
keep pulling/rebasing against the Lilypond master) to insert appropriate
copyright and licensing notices.  git blame should help to give a better
idea of who has contributed to what, so I can fire of queries to authors
where necessary.

What would be good is if as many contributors as possible can reply to
this email just to OK (i) my putting copyright/licensing notices in the
files they have contributed to and (ii) their licensing preferences for
their contributions:

    -- for code, GPLv2 only or GPLv2 or later;

    -- for docs, GFDLv1.1 or later and/or GPLv2 or later;

    -- freer licenses (BSD/MIT-style, or donated to the public domain).

I think that snippets are already public domain since it's a condition
of submission to LSR.

I have Han-Wen's OK already for his contributions, but would like
others. :-)

Now, future policies -- I would suggest new contributions be requested
to follow these rules:

    -- for code, GPLv2 or later or a more liberal compatible license;

    -- for docs, GFDLv1.1 or later/GPLv2 or later dual license (or
       more liberal compatible license);

    -- when altering a file that already exists, use the same license
       as for the rest of that file (so if someone contributes a code
       file under a BSD/MIT-esque license, anyone who adds to that file
       uses the same);

    -- patches that make a significant alteration to a file should add
       the author's name to the copyright notice

    -- new files should be required to carry a copyright and licensing
       notice when added to the source code.

Note that if all this goes OK, we should then be OK to unilaterally
upgrade to GPLv3 (if that's desired).

Does this sound good to everyone?

Best wishes,

    -- Joe