Glynn Foster wrote:
Hey,
Stephen Harpster wrote:
I'm also not asking to replace CDDL. I'm asking if people think it
would be a good idea to dual-license OpenSolaris CDDL code with GPLv3.
Of course that depends on what the final outcome of GPLv3 is, but
assuming it looks close to what it is today, would you like that, not
like that, or not care?
I don't really believe I'm enough of a stakeholder in OpenSolaris (ON) to feel
like I have a say in the matter, but what I'd really like to see is a set of
scenarios of how this would work - in terms of committing code back,
distributing code, and linking to the current closed sources.
If incoming code is licensed as CDDL or GPLv3, then we would ask that
you dual-license it CDDL/GPLv3. If incoming code is another license,
then that license needs to be able to work with both CDDL and GPLv3.
You don't want to disadvantage one over the other. If someone wants to
take OpenSolaris under the CDDL license, then they need access to all
the pieces. Likewise with taking OpenSolaris under the GPLv3 license.
In terms of the current closed sources, if we, Sun, wrote it (install,
storage, clustering), then eventually it will be opened under CDDL. And
if we, the entire community, decide that dual-licensing is a good thing,
we would also dual-license that.
For closed 3rd party code (mainly device drivers), we provide an
assembly exception to our GPLv3 code that allows it to be mixed.
Essentially, we're saying that for code Sun wrote, we'll allow our GPLv3
code to combine with these proprietary drivers without a viral effect.
Because we own the code we wrote, we can grant such exceptions.
Of course, this also means that incoming contributions that are licensed
under GPLv3 will also have to grant an assembly exception or you're
stuck with contributed code that doesn't work with other pieces of
OpenSolaris. (Note that this really only applies to the kernel. You
can still have userland GPL code just like we do today.)
As a random aside, I'd be worried that dual licensing would attract more people
to the code base that we still haven't been able to get to an operational level
for non-Sun contributions - perhaps that's a good worry to have, but I'd really
like to see serious progress being made before such a move is possible.
I imagine there will still be quality gates to get code in. The
constitution outlines that --- your average hacker just can't check code
in willy nilly. Communities will dictate some criteria for checking
code back and limiting check ins to qualified contributers.
Dual-licensing would not change what the constitution says.
Glynn
--
Stephen Harpster
Director, Open Source Software
Sun Microsystems, Inc.
_______________________________________________
opensolaris-discuss mailing list
[email protected]
