Apache vs. BSD licenses

[Brian Behlendorf]

On Tue, 20 Mar 2001, Rodent of Unusual Size wrote:
> Ian Stokes-Rees wrote:
> >
> > If we were to do this, and subsequent developers were to do the
> > same, then each developer who redistributed the accumulated code
> > base with new files and entirely new code which they put under
> > _their_ version of Apache License, the credits list would grow
> > and grow.
>
> That is a consequence of the ASFisms or their replacements, not
> the licence terms themselves..

Actually, Ken, Ian is right.  If there is a bundle of code that contains
ASF code, which has an Apache License 1.1-style license on it requiring
attribution to this other party, then both the ASF and this other party
need to have their credits remain.  The license terms do require this, and
it's not likely to change in the near future, since people in the ASF
believe pretty strongly in attributing credit where earned.  If you have a
string of derivative works, you have a string of attributions.  You can
remove an attribution from that list only if you're sure none of that
contributor's work is being used anymore.

Personally, I think this is a pretty small thing to ask.  Even with 1000
contributors representing a chain of 1000 derivations (whereas these days
you rarely have two or three) such an atribution would take under 100K of
text, or 10K compressed.  Giving credit is an inexpensive way of spread
the benefits of participation in an open source project, and costs the
current developers essentially nothing.

Note that this "advertising clause" is much different in 1.1 than in 1.0.
In fact, it was carefully written to *be* GPL compatible.  The clause
"Alternately, this acknowledgment may appear in the software itself,
if and wherever such third-party acknowledgments normally appear" can
apply to sourcecode as well - that is, if you have a GPL tarball that
contains Xerces code, you are of course distributing source, and that
acknowlegement appears in the source to the Xerces code, where such a
"third-party acknowledgment normally appears".  I've discussed this with
Stallman and he agrees, grudgingly, that clause 3 is not GPL-incompatible.
Such notices will persist, as well, because the GPL does not allow you to
remove copyright notices on included code (I believe...).

He still has an issue with clauses 4 and 5, though, which are a device to
help protect us against someone creating a proprietary fork and calling it
"ApachePro", or "Apache++" or whatever.  Stallman believes such things
should be enforced through trademark law.  I think anyone who's been
through trademark law proceedings would tell you to avoid it at all costs
- trademark law is all about who has the more expensive lawyers arguing
your case.  :/  By making it a term of copyright, we protect that interest
in a much more direct way.

Stallman has indicated to me that clause 4 ("Apache" may not be used to
endorse) will be compatible with the GPL v3, but clause 5 ("Apache" may
not appear in the product name) will not.  I think this is unfortunate, as
in a digital environment, your good name is your only asset, and
protecting it shouldn't be hard.  I don't see asking someone to choose a
different name for a derivative work as not qualifying as "free" as the
FSF defines it.

	Brian