Stephen Harpster wrote:
The big bonus is that GPLv3 will open us up to a whole new audience.
Linux aside, there are plenty of other big projects that will combine
with OpenSolaris more easily if we're dual licensed. To be successful,
you want to reach out to as many communities as possible. The more
friends the better. GPLv3 will give us that.
That's an interesting collection of assertions, but I'm not sure they
are entirely correct. It really depends on what you mean by "combine
with Solaris". We already have a significant amount of GPLv2 code
shipped with Solaris, so Solaris not being GPL obviously isn't an
impediment. Any existing candidate projects will be GPLv2, so as I
understand it they won't be able to "combine" with a GPLv3 Solaris
unless they switch to GPLv3 first, and we won't know the level of uptake
of GPLv3 until it's been in existence for some time.
Very unlikely that a source fork will happen. Let's face it. Most of
the people who know and understand all the intricacies of OpenSolaris
source code work at Sun. Who's going to fork? How will they maintain
that fork? Constantly chase opensolaris.org? And what happens if their
new incompatible changes don't work with the changes they pull from
opensolaris.org? It's not practical and I can't imagine it happening.
What is to stop someone producing an OpenSolaris distribution where the
only significant difference is that they've ripped out the CDDL? That
would seem to be fairly easy to do, and I don't think it would be a good
thing if it happened.
An assembly exception is sort of a way to neuter a license. Suppose I
have two files, gpl.c and harpster.c. gpl.c is dual licensed under CDDL
and GPLv3. harpster.c is licensed under the Harpster license, a
proprietary license that solves world hunger. ;-)
Now normally, linking gpl.c and harpster.c would force harpster.c to
also be licensed as GPL. That's the "viral" nature of GPL everyone
talks about. If I wrote gpl.c, I can place upon it an assembly
exception that says, "when you link gpl.c with a Harpster licensed file,
don't force the Harpster licensed file to be GPL." Because I wrote and
own the original gpl.c, I can modify the terms of the license to be more
restrictive. (Or less restrictive depending upon your point of view.)
If we "neuter" the GPLv3 license with an assembly exception we'll
immediately be accused of playing marketing games with licensing, which
will defeat the entire purpose of the exercise. In fact it will make
things *considerably* worse, not just in terms of the negative PR hit
we'll take, but also in terms of the unnecessary complexity and
confusion we will have saddled ourselves with.
There is a direct analogue we can look at - MySQL is dual-licensed under
both GPLv2 and a Commercial license. It's a cause of great confusion to
anyone trying to use MySQL, and I don't see hordes of people switching
from Postgres (BSD license) to MySQL as a result of MySQL being
dual-licensed, in fact I suspect the flow is the other way - anyone who
want's an open source database will pick Postgres, not MySQL as MySQL is
only pseudo-open at best. For GPL purists the only acceptable license
is pure GPL. A dual-license, especially one that contains a crippled
GPLv3 is more likely to drive them away than it is to attract them. By
sticking to pure CDDL we can make a reasoned defence of what I consider
to be a very good license. I strongly suspect bolting GPLv3 on the side
will only make things worse, not better.
We can't ditch CDDL for all the reasons we put it there in the first
place -- and we don't want to alienate the community we have. There are
still folks who will want to embed OpenSolaris in appliances and create
proprietary solutions. CDDL allows for that very nicely. GPL does not.
If you force the GPLv3 issue against the wishes of the community you are
going to alienate many of them anyway.
If in 12-24 months time a significant proportion of the open source
world has switched to GPLv3, then that would be a good time to consider
a switch for Solaris. The upsides of switching Solaris to GPLv3 at this
point in time are massively outweighed by the downsides. For now I
think we should leave the OpenSolaris licensing alone.
--
Alan Burlison
--
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
