Re: Copyright and License Guidelines
On Mar 22, 2011, at 13:27, Jonathan Yu wrote:
> Replying to both Jeremiah and Damyan here:
>
> On Tue, Mar 22, 2011 at 5:39 AM, Jeremiah Foster
> <jeremiah@jeremiahfoster.com> wrote:
>> Stating this policy clearly in pod, even if it is redundant, is good as long as it doesn't
>> stray from the DFSG or try to reword already established Debian policies on copyright
>> and licensing.
>
> I don't think a friendly reminder about Debian's social contract and
> free software guidelines is a bad thing, particularly since I would
> submit the most active members of our group are not (yet) Debian
> Developers.
>
>> Are there legal implications if you do not have the year of the copyright attribution?
>> Is the copyright assignment weaker if it lacks a date?
>
> Technically, a copyright statement is not actually a *must*, only
> licensing is.
According to whom? Copyright is the backbone of the GPL. The license cannot be assigned if you do not have copyright. So actually this is backwards - the copyright is a must. The license makes a great deal of difference too, at least in terms of distribution.
> However, I worded that harshly because I want to be
> strict -- moving forward, we need to have somewhere we can point
> authors to tell them that we *certainly* must have copyright
> information. Here's why:
>
> Without an explicit copyright statement, the burden is on the packager
> to determine the appropriate copyright. Who holds the copyright, you
> ask? Whoever has worked on the package, submitted code, etc. It may be
> difficult or impossible to determine who has contributed, particularly
> for team-maintained packages or where authors accept patches (without
> a CLA as mentioned).
>
> If you start a project, and I submit a patch to that project, in the
> absence of a copyright statement and a CLA, it can be assumed that I
> retain copyright on my work (that is, the patch I submitted to you).
> Now I hold partial copyright to the project (the parts that I
> contributed). You are no longer free to relicense things (as you would
> have been able to do before, since you had the agreement of all
> copyright owners -- that is, yourself :-)
>
> There are probably other implications I am not aware of. I disagree
> with Damyan on a philosophical note and do not believe the Contributor
> License Agreements are a bad thing.
The Contributor License Agreement is the wrong solution. The right solution is getting the right copyright throughout the package and honoring the copyright holder's wishes - not forcing them to abandon their copyright. Damyan's position is the right one, and the one consistent with Debian's history.
>>> Updating copyright information is easy. Examine the diff when
>>> upgrading the upstream sources and be patient. Relying on external
>>> sources seems like a easy workaround that may lead to wrong results.
>
> This seems to neglect the fact that copyright information should
> already be in the upstream package -- we are merely putting it in a
> machine-readable format. The copyright/license details will always
> accompany the source -- we are not stripping out upstream LICENSE
> files or clauses in order to *replace* them with the DEP5 copyright
> file. We are merely supplementing them.
>
>> I think there is no viable alternative aside from correctness. I don't see how we can distribute
>> software without confidence in copyright.
>
> And we cannot have confidence in copyright without upstream giving us
> the details. At the very least, if copyright information is incorrect,
> then upstream will be liable -- not us.
This is why the solution needs to come from upstream and not through Debian requiring copyright assignment through other means.
Jeremiah
Reply to: