Re: Decision on R datasets
"Paul R. Tagliamonte" <paultag@gmail.com> writes:
> The real question is it less *distributable*? If its GPL and contains
> things *not* in preferred form, I'd say yes, that would be less
> distributable, since its a GPL violation and results in breech of
> license - which means we can't distribute.
I think this is where one also applies the point that the law was intended
to be understood by humans, not lawyers. I don't find it particularly
likely that redistributing the exact same files that upstream releases
under the GPL is going to be found to be a copyright violation by any
actual court.
There are scenarios where this could still be a problem, of course. Most
of them involve a transfer of ownership from one copyright holder to
another, who then starts using license terms to try to pull back all the
code. That's (one of) the reasons why we're rightfully nervous about such
things in the main archive. (Another is that, regardless of what the
license says, it's bad for a free software distribution to distribute
software that cannot be easily modified, which is one of the major reasons
why we insist on having the preferred form for modifications.) However,
we've always previously treated Alioth and similar places on a looser
basis, with the understanding that we'll follow reasonable precuations to
comply with upstream's license and take things down if someone complains
about it.
I am certain there is technically undistributable stuff in Git
repositories on git.debian.org. If someone notices, we should obviously
remove it, but I'm not going to lose a lot of sleep over it, or worry
about cases like this one where the potential license conflict is
questionable and not upstream's intent.
In other words, I think people are jumping on Charles because he took a
confrontational position, but when one looks at the details of what he's
talking about doing, it's not atypical for Alioth use in general.
--
Russ Allbery (rra@debian.org) <http://www.eyrie.org/~eagle/>
Reply to: