Contractual requirements [was: request-tracker3: license shadiness]
Brian Thomas Sniffen wrote:
> Josh Triplett <firstname.lastname@example.org> writes:
> > Is it still non-free even though you are not required to submit patches
> > to them for inclusion? If you opted to never send patches upstream, the
> > condition would not affect you at all. Note that simply distributing
> > the patches could not be considered as "submitting for inclusion in the
> > work"; you would have to explicitly submit a patch to upstream for
> > inclusion. Many upstream authors refuse to include submitted patches at
> > all, or require the copyright be explicitly assigned to them, or require
> > some other arbitrary condition; this one just states that they will take
> > submitted patches and assume they hold the copyright on them.
> It's a fine policy -- I don't use it myself, but it's sure helped the
> FSF out, which is a good thing. But that policy should not be written
> into the license. This clause, even if changed to "submitted for
> inclusion by the copyright owner" is still the sort of thing which
> requires agreement from that person -- it's a contract, not just a
> license grant. That makes me reluctant to consider it free.
This comment has just clarified something that's been rattling around
half-formed in my head for a little while now, regarding Free licences. I
don't know if it's been raised before, but I think it bears discussion:
"A licence cannot be Free if it disallows actions which, in the absence of
acceptance of the licence, would be allowed by Copyright law, or imposes
restrictions not present by Copyright law".
To put it another way (and closer to Brian's wording), if the licence isn't
simply a grant of permission but requires things of me which I would
otherwise be allowed to do, it can't be free.
This would, for instance, put the kybosh on the MPL's "you must tell us if
you know of a problem" clause, because it's something that straight
copyright law doesn't require. RT3's clause under discussion would fail,
because copyright law recognises separate authorship and copyright claims in
a single work, if published as such. (It also recognises that copyright
assignment may occur, as in the FSF case, but the GPL doesn't make it a
requirement of the licence, the FSF makes it a requirement of getting your
code into it's software).
It wouldn't be a problem for the GPL's changelog clause, because ordinarily
distribution of modified versions isn't allowed by copyright -- the
changelog requirement is just setting a boundary for what extra rights you
are being given above and beyond copyright law.
Note that licence clauses which say "you can modify the whole work except
for these bits" don't fall foul of this test -- again, it's a fence-building
exercise. Incomplete modifiability is a different kettle of Free fish.
I can already recognise, just ticking through licences in my head, that this
test has a fairly fuzzy dividing line in it -- caused by differing ideas
about what copyright law is, exactly, and how it is interpreted. That's a
pity, as a bright line is always useful, but I (naively) think this test
could be of benefit in providing another benchmark for judging licence
freeness -- if it's not deluged by a long list of properly free licence
clauses which make it look stupid.
I'm interested in what more knowledgable minds think about this idea --
whether there are multitudes of otherwise Free licences which would fail
this test, and so forth. I think the wording of the test needs to be
polished a little, too.