Re: Hypothetical situation to chew on
> I'm vaguely aware of a piece of software which contains both GFDL
> licensed material, and possibly code which was dropped in without
> actually checking the licence for compatibility with the GPL.
> A gargantuan number of people over the years have contributed code to
> it, and many have claimed copyright for their contributions. No policy
> of copyright-assignment has been used.
> So here's a hypothetical situation; say the current upstream maintainer
> was to announce in a very public place, with Cc's to all known
> contributor e-mail addresses, his intent to change the licence of the
> code to GPL-2 (including documentation) and give a full list of
> everything that would fall under it. And then was to give a period (say
> 28 days) for objections to be raised.
> If none were raised, could they then change the licence?
Yes, this is what SUCKS about current copyright law. The presumption is "All
rights reserved unless you have explicit permission".
>If not, what procedure would be needed to make the software DFSG-free?
>I'm going to guess clean-room rewrite of all of the documentation, and
>of any code that could be affected?
Not *quite*. But close.
(1) Every piece of code must be audited to determine the copyright holders.
(I *hope* they kept track of that, but many don't.) The author of any major
portion or major collection of changes -- or his/her employer if it was work
for hire and the author was in a country with that doctrine -- is a copyright
holder for that portion. (Unless that person has explicitly released it to
the public domain, or it was published without copyright notices in the US
prior to, I believe, 1986 -- in those cases it's in the public domain.)
(2) If the author of a piece cannot be identified, and the piece is not
clearly in the public domain, it must be clean-room rewritten.
(3) If the author can be identified, and agrees explicitly to relicense the
code (or the license for the code is already OK), it can be kept.
(4) If the author can be identified, but does not agree explicitly to
relicense the code (whether this is because he/she can't be reached or
because he/she refuses or because he/she is incomprehensible), it must be
So, (3) is the exception to your rule; the work of anyone who can actually be
tracked down, and who agrees to relicense, can be relicensed. Perhaps
surprisingly, many people can be tracked down, and *do* agree to relicense.
For large chunks, this may be easier than clean-room rewrites. For small
chunks, probably it's harder.
I know of several packages which fall into this rather nasty category, and I
haven't had the heart to work on most of them. I wish you luck.