www-legal-discuss mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Robert Burrell Donkin <rdon...@apache.org>
Subject RE: Comments on GPLv3
Date Tue, 19 Jun 2007 21:52:02 GMT
On Tue, 2007-06-19 at 15:32 +0200, Santiago Gala wrote:
> El vie, 15-06-2007 a las 11:19 -0400, Jeffrey Thompson escribi??:
> > 
> (...)
> > I think that's a real concern that Apache needs to figure out how to
> > handle.  Even though the FSF has declared GPLv3 and Apache to be
> > compatible, the projects cannot cross-pollenate.  If a GPLv3 project
> > takes some Apache code and evolves it, the Apache committers need to
> > be on the lookout for people trying to contribute those changes back
> > to Apache by mistake.  I don't know how they are going to do that
> > since the code will likely still have the Apache copyright and license
> > notice in it.  Has anyone thought about how to do that? 
this is a general problem for apache licensed code. our source is
already included within more restrictively licensed works. i hope that
the mechanisms that apache already uses would be strong enough but
perhaps this is a good time for them to be reviewed.

apache code is a mosiac patchwork of contributions by individuals
covered by CLAs, CCLAs, software grants or the AL2.0 grant-back clause.
apache tries to act in good faith and if contribution mistakes are
uncovered would back out the suspect patches and create alternative
clean room implementations of the required functions.

apache tries to encourage developers to contribute only original works
created for inclusion within an apache work. external code should be
imported only through the incubator and should be checked by the
incubator PMC.

i hope that this gives apache reasonable coverage but i'd be interested
to hear opinions from those who are better informed.

> My read is that small changes (bugfixes and the like) would not be a
> problem for the people contributing back to license under the AL 2.0, as
> the issue tracker reminds them every time they use it to attach a patch,
> etc.

if a contributor submits code through JIRA ticking the inclusion box
which is derived from source covered by the apache license, 2.0 and
which has the apache license header then i think that IMHO it's
reasonable for apache to assume that this contributor is knowingly
submitting code to apache under the apache license.

if that code has been developed in a GPLv3 project after being forked
from an apache project then AIUI AL2.0 requires that the source should
include indication of the changes made and the licenses for these
changes. if the source file contains license information that indicates
that some changes are GPLv3'd then it must not be included within an
apache code base. if not, then the source is AL2.0'd, not GPLv3's and
there is no issue.

> If people is drastic in their views and they insist in the changes being
> GPLv3, that means they have effectively forked the code they used, and
> they are due to maintain their changes... and upstream forced to
> duplicate bug fixing effort, etc.


apache is generally comfortable with downstream derivative works and
IMHO this is very likely to be more common as time goes on. maintaining
a patched version is reasonably onerous.

i can see that this may get messy from a personal perspective: there are
only a number of ways that particular bugs can be fixed but providing
that developer stick to the IP clearance requirements, it shouldn't be
too messy legally.

> This is *not* happening currently with lots of basic infrastructure
> projects that are not under GPLv2, see for instance x.org, python, ...
> So I guess there is plenty of space for all the communities to live.

it is possible that a small number of people may want to try to
undermine apache by submarining submissions and then claiming that we
must GPL the whole. provided that the rules are followed and we are
covered by the license or by a CLA then apache has acted in good faith
and i hope that the FSF would be happy for us just to remove the code
contributed in error or in bad faith.

- robert

View raw message