[License-discuss] CDDL 1.1 and GPL 2 with CPE

[Karl Fogel]

Richard Fontana <rfontana at redhat.com> writes:
>I think the typical OSI modern tradition has been to wait for the
>license steward to request OSI approval, a general issue which someone
>raised on one of these OSI lists some time ago. (I believe exceptions
>around late 2007 included GPLv3 and its siblings for the reason that
>the FSF would never submit those licenses itself, along with the
>presumed inherent importance of those licenses.)

That's my understanding too.

>So, I would assume it is up to Oracle to decide whether to submit CDDL
>1.1 for OSI approval. (Is anyone from Oracle on this list?)

...or does anyone here want to contact them about it?

>The Classpath Exception may be another matter. I don't see the value
>of OSI approval of GPLv2+Classpath Exception (ignoring the question
>whether there's really a canonical version of it) since OSI has in
>modern times generally not bothered to approve GPL+permissive
>exception permutations, to my recollection.
>
>As for approving CDDL1.1+GPLv2 with Classpath Exception as though it
>were a single license, I think that would be unprecedented. Sun never
>asked for approval of CDDL 1.0 and GPLv2 + Classpath Exception as a
>single license package, SFAIK. 

Yes; to do this would to head down a combinatoric path of no return.

An hypothesis:

If a license is already approved as open source, and the copyright
holder adds an exception that merely indicates that under certain
circumstances they will not enforce certain terms of the license, then
the distribution terms are still "open source".  

The logic is that anyone who receives a copy of the software clearly has
all the rights guaranteed them by the base license, and in the general
case no one can compel a copyright holder to enforce things they choose
not to enforce anyway.  In other words, things like the classpath
exception are not really changes to the license at all.  They are rather
promises -- a form of estoppel, in which recipients can depend on the
license holder to not exercise certain powers they might otherwise have
exercised.

Since a copyright holder can only *disclaim* powers, rather than claim
new ones not granted by copyright law or by the license they distribute
under, these exceptions therefore do not affect the OSI-approved nature
of the license to which they refer.

Does this sound sane?

-Karl