Re: Licensing - Mailing list pgsql-advocacy

Christopher Browne wrote:

>That's an excellent commentary on the issues.
>
>I'll poke at the "business friendly" bit a little bit because it seems
>to me that things are a _little_ more complex than that.
>
>The approach MySQL AB has taken with its "dual licensing" (I love to
>call it "dueling licensing" ;-)) is, in fact, quite "business
>friendly."  It's just that the only business that it happens to be
>particularly friendly to is MySQL AB.  Ditto for TrollTech and Qt, and
>Sun and OpenOffice.org.
>
>The "GPL + Traditional License" approach that MySQL AB is encouraging
>is compatible with the notion that the "market" will consist of a
>single software producer with exclusive ownership of the code base who
>then sell it into a traditional style "proprietary" community of
>customers/consumers.
>
>Unfortunately, in order to be able to operate under the dual licenses,
>this presents the necessity that one party has exclusive ownership of
>the application code.  That requirement of ownership prevents the kind
>of "community participation" we see with PostgreSQL, where there are
>numerous contributors working for numerous organizations.
>
>
Ownership is not strictly speaking required.  Digium, for example, does
not require that contributions to Asterisk are assigned to them, just
that the author gives them a license under BSD-like terms (which they
then relicense under the GPL and their own proprietary license).

The difference is that if I hand over my copyright to you, I am no
longer allowed to distribute my work as part of another product, but if
I license it to the company under a permissive license, I still retain
all rights to it.

>People are often willing to sign over copyright to an organization
>that operates in some form of "public interest," wherein you can see
>MANY contributions that have gone to GPL-licensed software where
>copyright is held by the non-profit "Free Software Foundation."
>
>
Completely agree.  It is a worse form of subsidizing the competition is
than normally found in the BSD community (think FreeBSD v. SunOS) ;-).
At least there the idea was that academic projects should support a wide
range of economic approaches.  With copyright assignment, one is
essentially *giving away their rights to rerelease the code under other
terms* while giving those rights to a competitor.

Note, I pick on FreeBSD for subsidizing their competition, and I realize
that FreeBSD was originally comfortable with this idea.  So the
observation is more an economic one than a practical one.  OTOH,
projects like Apache or PostgreSQL might subsidize their competition,
but the competition will rarely have any chance of superceding the main
project in terms of user base.

>There has been, in contrast, a distinct paucity of willingness to
>donate code to "dueling licenses" organizations.  Unlike the FSF
>projects, you _don't_ see a lot of code coming in from outside.
>
>
Personally, I see dual-license as a worst-of-both-worlds approach.  You
usually turn away good people who might otherwise make a contribution,
and you end up competing with your own free products.  Not a good
solution.  It can be done well (I think that Digium does a good job with
Asterisk), but that is a special circumstance partly mandated by the
fact that some standard telecom codects (G.729 iirc) are pretty heavily
encumbered under patents, so there is a real market which is
incompatible with the GPL.

Best Wishes,
Chris Travers