Subject: Re: compatibility and the OSD
From: Nick Moffitt <>
Date: Mon, 20 Sep 2004 08:32:19 -0700

begin  Bob Scheifler  quotation:
> >Provided both licenses pass the OSD, this is entirely acceptable.
> There may be confusion here. I'm not constraining what the other
> license is; you are free to use whatever license you want. Whether
> you, for a specific executable derivative work, choose to license it
> under an open source license or not, should not affect whether the
> license I sketched is itself an open source license?

	Consider the MIT X license (or the shorter, newer BSD), which
puts no restrictions on derived works.  That is effectively the same
thing.  Many proprietary software vendors have taken BSD or MIT
X-licensed code and included it inside derived works that use some
other license entirely.  I could this very day start a new
distribution of FreeBSD licensed entirely under the GPL, and it would
be perfectly legal.

	The FSF (not the OSI) has ruled that the phrase "DO WHAT THE
<redacted> YOU WANT!" is an acceptable Free Software license*, and
their standards are much more strict and subtle than the OSI's.  You'd
have a hard time convincing me that "Do whatever you want with this
work" is somehow in conflict with the OSD.

> >Of course, if you really are coming up with some license that does
> >not meet the OSD in the event that the user does not pass the test
> >suite, then that is no more acceptable than Sun's Java license.
> The sketch I provided *is* the hypothetical license (informally
> stated).  The question is, if and how does it conflict with the OSD?

	The big restriction seems to be that all distributions that
fail the test suite must include an unmodified copy of the suite.  It
appears that they may, also, include modified copies.  This would be

	Cumbersome requirements to include data of one sort or another
are not unheard of in Open Source licensing.  For example: most
licenses require that attribution information follow a work around.
The GPL technically requires that you include some indication of how a
work has changed, although many people omit changelogs without fear of
litigation (or perhaps I misunderstand those clauses).

	My impresion (not any sort of binding conclusion, as I have no
official relationship with the OSI) is that so long as you allow
modified copies of all code to be distributed, the requirement to
include the pristine sources is cumbersome but not necessarily in
contradiction with the OSD.  I believe the OSD permits licenses that
allow modifications to be distributed only in patch form, for example.

*: Per a conversation with Bradley Kuhn outside Linxuworld San
    Francisco in I believe 2002 or 2003.  Acceptability refers to
    legal uses, not social situations.

Support your droogs!                        Nick Moffitt