License compatibility of MS-PL and MS-CL (Was: (RE: Groklaw's OSI item (was: When will CPAL actually be _used_?))

Michael Tiemann tiemann at
Thu Aug 23 12:07:37 UTC 2007

On 8/23/07, Michael R. Bernstein <michael at> wrote:
> Speaking for myself, I am not so much concerned with any particular
> incompatibility arising from different sets of permissions or
> requirements (although those can be unfortunate). It is generally fine
> with me that the submitted licenses are incompatible with the GPL or any
> other individual license.
> What bugs me is that these licenses are incompatible with *every* other
> approved license, even the BSD license. That took special effort.
> No, let me amend that: It bothers me that these licenses are
> incompatible with the entire universe of all other theoretically
> possible licenses, including future ones, regardless of their terms.
> IANAL, but it seems to me they are even incompatible with licenses that
> are word-for-word identical except for a name change.
> Is this true of any other OSI-approved license?
> It also bothers me that this incompatibility extends not just to code
> mixing and sublicensing, but also to multiple licensing. Multiple
> licensing, like forking, is an unfortunate and infrequent necessity that
> I don't think should be given up lightly.

Very well said!  I was having difficulty understanding the extraordinary
distaste that some were expressing about the compatibility problem, and if
your characterization is correct, it certainly makes the criticism very easy
to understand.

I think we need to add OSD #11 to disallow this kind of gratuitous and
> discriminatory license incompatibility. Call it the principle of
> licensing neutrality: no provision of a license should explicitly
> disallow combination with code under other licenses if the license terms
> are otherwise compatible.

While I'm not yet ready to seriously consider OSD#11 as you propose, you
also provide an excellent way to frame an aspect of the license
proliferation problem.  Today in San Francisco, several OSI folks are
meeting to talk about a License Wizard--a program designed to identify
licenses compatible with specific criteria, although the scope may also
expand to identifying compatible/incompatible licensing combinations.  (I'm
not there right now and I'm not running the agenda.)  But certainly you
raise an interesting point: if the principle objective feature of the
license (other than OSD compatibilty) is license *incompatibility*, perhaps
there should be a special category and perhaps we should recognize it as

A second comment reinforcing your argument, and your earlier appeal to the
BSD license is this: any license that cannot be mixed with BSD presents a
special case.  Granted, the OSD sets objective criteria and "BSD
compatibility" is not one of the criteria.  But the many arguments I've
heard from the copy-left and the copy-center folks come down to a basic
disagreement on whether freedom is something that should be transitively
preserved (copy-left) or something that should be exerciseable at the point
of distribution (even the freedom to take code proprietary, which is what
the copy-center people endorse).  Though I personally prefer and often
advocate for the copyleft position, I hear the copy-center position, and in
particular I respect the desire of copy-centrists to occupy the space of
maximal compatibility, even if it means being compatible at the point of
exit from open source.  That said, a license submitted to the OSI that is
not compatible with the one license that attempts to be maximally
compatible, and which many in the community believe to be The One Compatible
License (well, MIT could claim that, too), such approval would be disruptive
to that copy-center position.

Or perhaps this: A license may not discriminate against other licenses
> or groups of licenses.
> Just to re-iterate: In my opinion incompatibility with other individual
> licenses is NOT objectionable in and of itself. I do not think that
> incompatibility per-se is non-neutral or discriminatory.

I agree with you on that, too.  In many past discussions the arguments of
license-discuss and the personal appeals of OSI board members have prevailed
upon many license submitters to change their licenses so as to minimize the
harm those licenses do to the overall open source ecosystem (as best we
understand it).  I think that this is a case where the proper next step is
to take Microsoft up on their offer to discuss these points and see if we
cannot address this particular case of license incompatibility.  Again, I
agree that license incompatibility per se is not evil, but total
incompatibility with any other possible OSI-approved license is not a
feature that fosters the benefits of open source as I understand them.

- Michael R. Bernstein
Thanks for your contribution!

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the License-discuss mailing list