[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: GPL and LGPL not acceptable for Fedora!

Hash: SHA1

Tom "spot" Callaway wrote:
> On Thu, 2007-08-16 at 14:56 +0200, Adel Gadllah wrote:
>> Hans de Goede wrote:
>>> GPL+, the version in the COPYING file is irrelevant.
>> this sounds wrong ... if the source says nothing and the copying says 
>> gplv2 it _is_ gplv2.
> COPYING does not signal licensing intent (it might not seem intuitive,
> but this is what Red Hat legal told us, and we're going by that).
> The order of operations goes like this:
> 1. What does the code say? If it specifies a version, that's what it is.
> 2. Does the code conflict with itself? (file1.c and file2.c are compiled
> together but have different licensing)
> 2A. Are the conflicting licenses compatible?
> 2AA. Does one license overpower the other one? (GPL/LGPL does this) If
> so, the strictest license wins.
> 3. What does the documentation say? This signals the author(s)
> intentions from a legal perspective, although, not as binding as in the
> source. If the documentation specifies a version when the source does
> not, then we can use the documentation as our source. NOTE: COPYING does
> not count as documentation, since the author(s) didn't write it.
> 4. If neither the source, nor the upstream composed documentation says
> anything about the license version, then it could be under _ANY_ version
> of the GPL. The version listed in COPYING is irrelevant from this
> perspective.
> Now, keep in mind that most upstreams are probably leaving the
> versioning out by accident. If you get to case 4, you definitely want to
> let upstream know that you are unable to determine the applicable
> version(s) of the license from the source and documentation. They'll
> almost certainly let you know what their intended license version is,
> and (hopefully) correct it in the upstream source.
Actually spot, I have to disagree with you slightly here.  If COPYING
doesn't signal licensing intent, then those programs which have no
license information in the source (like abe, apparently) *are not
licensed*.  Which means that we can't distribute them.  This could even
be carried out to mean that a program which lacks licensing information
in one necessary source file would make the whole work not licensed for
us to do anything with.

Either COPYING has a meaning or it doesn't.

OTOH, it is probably more common for there to be a COPYING(v2 or v3)
file in the source tarball and a source header that says "GPL".  In that
case, I would say we *should not put GPL+ in the License field* if at
all possible.

The reason is that whatever the legal technicalities of the situation
are, the author of the software included a GPLv3 license file and wrote
something that was a common shorthand for that license file at the time
the software was written.  Whether or not we're legally correct in
writing GPL+, we don't want to piss off (and possibly get
[unsuccessfully] sued by) an author that says, "I meant GPLvX who the
fuck are you to steal my work?"

In an even more practical vein, if we put GPL+ in the license field and
someone takes our word for it, goes off and starts using it as a GPLv1
instead of GPLv3, then we're right in the middle of causing everybody a
lot of pain when the licensing is clarified for all involved.  If, OTOH,
we write GPLv3+, (which is legal for us to do even if the original is
technically GPL+) and someone licenses their code as GPLv3 because of
it, the most harm we've done is caused someone to use the GPLv3 when
they could have used GPL+.

Please, go back to common sense with this one and do what the vast
majority of authors intended by putting a GPLv2 COPYING file in their
source tarball, not what we know is the most liberal licensing that we
can legally distribute their code under.

- -Toshi
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]