Subject: RE: BSD and MIT license "compliance" with the MS-PL
From: "Wilson, Andrew" <andrew.wilson@intel.com>
Date: Fri, 17 Apr 2009 17:24:56 -0700

 Fri, 17 Apr 2009 17:24:56 -0700


Matthew Flaschen wrote:

>>> But the real question is, "What is your definition of copyleft?"
>>> and who supports it.
>>
>> Evidently Andy Wilson and Chuck Swiger...and whomever else they
>> remember during those discussions.  How many folks are required and
>> why isn't the FSF good enough?
>
> Because in this case, I think the FSF (or rather the one unknown but
> well-intentioned person who updated that license list) is contradicting
> their own definition (http://www.gnu.org/copyleft/).
>
>> My definition of strong copyleft is the same as yours.  My definition
>> of "weak" copyleft is "something less than strong copyleft" and
>> fuzzier because weak copylefts are "weaker" than strong copylefts in
>> different ways.
>
> Weak copyleft is no excuse for a weak definition.

Matt, it does seem a little disingenuous to say one FSF citation is
wrong (the one which explicitly calls MS-PL a copyleft license)
and another is right (RMS's personal definition of copyleft, which
is quite ideological and involves many tests that full freedom
of the code is preserved).

My personal definition of copyleft, for what it might be worth, as
a term of art in open source licensing as opposed to the RMS ideological sense,
is that copyleft is one step removed from permissive.  A permissive license
allows derivatives to be sublicensed under terms and conditions of
the licensee's choice.  A copyleft license does not.  A copyleft license
says the only valid source license for derivatives is the original license.

MS-PL is sui generis because it is copyleft (under my definition
above) but not reciprocal.  MS-PL imposes no requirement for sources
of derivative works to be made available; however, it insists that if
"any portion of the software" is released in source, it must be under MS-PL.
This makes MS-PL unique since all other OSI-approved non-permissive licenses
are also reciprocal and enforce some measure of code sharing.

Andy Wilson
Intel open source technology center