On Tue, Mar 25, 2008 at 04:53:48PM +0100, Nils Philippsen wrote:
On Sun, 2008-03-23 at 11:43 -0400, Tom "spot" Callaway wrote:
> Free, and GPL compatible as long as the copyright holder is Eric Allman,
> Sendmail Inc, or the University of California.
That's an interesting restriction. Is there a story hidden behind it?
Not really, but here's the explanation. The sendmail license was
drafted by Eric Allman with (among other things) the intent of making
it GPL-compatible. However, the clause in question does not clearly
achieve that effect. It says you can redistribute, even for profit,
Redistributions are accompanied by a copy of the Source Code or by
an irrevocable offer to provide a copy of the Source Code for up to
three years at the cost of materials and delivery. Such
redistributions must allow further use, modification, and
redistribution of the Source Code under substantially the same terms
as this license.
It is not clear from a literal analysis how to interpret that second
sentence, but one could argue that if you incorporate sendmail code
into a larger GPL-licensed work, there are licensing terms governing
subsequent distribution of the whole which are no longer
"substantially the same terms" as those of the sendmail license. (For
example, sendmail permits binary-only distribution, whereas the GPL
requires that the recipient of a binary be provided with complete
corresponding source code.) There's enough evidence to satisfy me that
Eric Allman didn't intend that interpretation.
I think it is reasonable to consider Eric Allman and Sendmail,
Inc. licensing alter egos for purposes of this analysis.
As for the University of California, there's a 3-clause BSD-like
license embedded in the Sendmail license that applies to portions
copyrighted by UC.
If some third party licensor started using this license for some other
code, I don't think you could assume they'd share the view of Eric
Allman that the clause in question was designed to facilitate GPL
compatibility. Indeed, the best literal reading of the clause leads
to the opposite conclusion. Licenses are generally interpreted
according to the intent of the licensor.
I think this might take care of all code in sendmail that is actually
covered by the Sendmail license, but I haven't checked closely.
Richard E. Fontana
Red Hat, Inc.