[copyleft-next] copyleft-next 0.2.0 released

Richard Fontana fontana at sharpeleven.org
Thu Feb 21 15:11:30 UTC 2013


On Thu, Feb 21, 2013 at 09:17:27AM -0500, Pamela Chestek wrote:
> 
> On Thu, Feb 21, 2013 at 4:59 AM, Gervase Markham <gerv at mozilla.org> wrote:
> 
>     >> Who decides? That sounds like a recipe for lawyer argument. Is it better
>     >> to refer to the OSI license list as of a certain date instead?
[...]
>     I think that having a clearly defined list is much easier. Then it's
>     pretty obvious to all. In the current set up, I can imagine someone
>     wanting to take advantage of that clause thinking "Hmm, I don't think
>     their pseudo-open license meets the OSD, but they do. So what now?"
>     Pseudo-open is currently observed (if not common), and we should plan
>     for it possibly remaining so.

True. I suppose at some point I felt that for purposes of keeping this
provision from being too complex some lack of precision or
introduction of pathological possibilities was tolerable. (I note that
the OSI has, in the distant past (long before the tenure of current
board members :), approved some rather questionable licenses, and has
shown no inclination to revisit past decisions.)

But the 'prevent complexity' argument supports referring to the OSI
list too.

> What about the Fedora list?

That is out of the question, since Fedora has an intimate relationship
with $DAYJOB and I have personally worked on the maintenance of that
list with my $DAYJOB colleague Tom Callaway. It's bad enough that this
mailing list is hosted on fedorahosted.org, which is only because I
couldn't figure out how to set up GNU Mailman with a mail server in a
working manner and this option happened to be readily available.

> Or on any one of the Debian, OSI or Fedora lists?

In an alternate universe Debian would be nice, but in ours Debian is
out of the question, because it has no true 'list' other than
"whatever covers what we distribute in main". It is quite clear that
Debian engages in little if any scrutiny of such licenses. At least
the OSI has a well-defined process for approval of licenses. 

So the best option is the OSI list, for all its faults. One that I
also considered is the FSF list (or the bkuhn-style intersection of
those two lists, but the FSF list of free and nonfree licenses has
never aimed for comprehensiveness. That's true in a different sense
for the OSI list, to be sure, but at least the OSI has a publicized
process for what gets approved. The FSF list seems to be more like a
set of interesting examples. 

> Even if you disagree with the judgment of a particular list it will be a very
> fringe case, and I agree with Gerv that certainty is better than uncertainty in
> this case.

That is true for the OSI list, and indeed the fringe case problem
would exist under the current formulation, since one can very
compellingly argue that a given fringe license that happened to be
approved by the OSI in the early 2000s must meet the OSD which is
maintained by the body that approved such license presumably in
relation to the OSD. 

I believe I may have been worried about the perception of a fairness
problem, as hundreds of clearly-FLOSS (I would like to say
'clearly-OSD-compliant') licenses have not been and undoubtedly never
will be approved by the OSI. But that's probably not a big problem,
and in any case where it would be the solution would be to seek OSI
approval for the license in question, which might be worthwhile for
everyone.

- RF


More information about the copyleft-next mailing list