[forum] GPL-incompatible license

[David Dawes]

On Thu, Feb 12, 2004 at 09:57:45AM -0500, Richard Stallman wrote:
>    XFree86 contains code from many sources and contributors, and under a
>    range of licences with many copyright holders.  There has never been
>    a single _specific_ licence covering everything.
>
>Thank you, I stand corrected.
>
>    The general preference has been for licences like the BSD and MIT
>    licences.  By BSD, I mean the original BSD licence in common use when
>    XFree86 began.
>
>The original BSD license is incompatible with the GNU GPL.  I worked
>for a couple of years to convince Berkeley to switch the BSD source
>code to the revised BSD license.  I hope we can avoid bringing back
>that sort of "advertising clause".  Its problems are cumulative at the
>level of the whole system, and thus rather severe.  (Also, it is
>incompatible with the GPL.)
>
>Are there any files in Xlib that are covered by the original BSD
>license?

I do not believe that any files in the client-side libraries are covered
by the original BSD licence, although I have not made a thorough check
for this.

>    So, I am surprised to hear that you were not aware of this, especially
>    since the FSF web site's "Various Licenses and Comments about Them" page
>    <http://www.fsf.org/licenses/license-list.html#OriginalBSD> links to
>    the online copy of the XFree86 3.3.6 COPYRIGHT document
>
>Alas, I'm only human, and I can't keep up with all the work that would
>be useful for me to do.  I remember noticing that link at one point,
>and thinking "that's an unusual place to find the original BSD
>license", but never thought of reading the page it pointed to.  Being
>in a hurry, as always, I just moved on to the next task I had to do.
>Of course, I forgot all about it until just now.

That is understandable.  I am just pointing out XFree86's licensing
policy is fairly well established, and it is this policy that has not
been changed.

>It looks now I had better fetch and read that page.
>
>    I could see a potential case for arguing that our library licensing
>    policy be tightened, i.e. become more restrictive in what licences it
>    will accept, and reduce the licensing choices available to contributors
>    to XFree86 libraries.
>
>If that eliminates the GPL-incompatible licenses from Xlib, it would
>eliminate the incompatibility with applications.  There might still
>be incompatibilities with GPL-covered software that people might want
>to use with the X server in various ways.  Freetype does not cause
>a problem because its dual license provides GPL compatibility.  So it
>becomes a significant question which other GPL-incompatible licenses
>are really used, and where.

Yes, such a change in library licensing policy would ensure that the
client-side libraries, and other critical dependency libraries, whether
they be shipped by XFree86 or not, would continue to be compatible with
both XFree86's general licensing position and the GPL.  There are several
ways that this could be achieved -- either by a single licence compatible
with both requirements, or dual licensing as in the FreeType case.

I will make this recommendation to the XFree86 board. In the meantime, I
will defer applying the licence change to any client-side library code.

I would also like to suggest that consideration should be given to
allowing the notion of "attribution equivalency" to fall within the
scope of GPL compatibility.  The goal of the XFree86 licence change is
not to require special treatment, but only treatment equivalent to that
given to other third party attributions.  I don't believe that this is
a large burden when compared to the requirement that all of the various
licence notices be published.  Furthermore, if publishing license notices
in the documentation accompanying a distribution is the only form of
third party attribution, then that would in itself satisfy the requirement
of our modified licence.

I appreciate your input on these issues.

David