Questions for Guido van Rossum (Was: ...Tim Peters)

[Tim Peters]

Henry, as Guido says here from time to time, if you have a question you feel
*must* be answered by him, posting to c.l.py is a low-probability way of
getting his attention.  So I'm copying him on this directly; he does get
around to reading his email, and usually no more than a few years after he
gets it <wink>.

[Henry Jones <joneshenry at my-deja.com>]
> ...
> In brief, has the copyright holder CNRI of Python 1.5.2 and
> 1.6a2 actually agreed to the purported license included with
> the code, and do they agree the license is legally valid?

Guido can no longer speak for CNRI.  I was told that CNRI said they do *not*
believe the CWI license is valid, but can't point you to any public
statement on the matter from them one way or the other.

> CNRI's agreement as the copyright holder is the only thing that
> gives anyone else the right to distribute, use, or modify the code.

Which is why BeOpen is negotiating with CNRI about CNRI's new license; we
would have released Python 2.0 under the BSD license (on top of the CWI
license) if CNRI hadn't opposed that plan, umm, vigorously.

> If there is any doubt, in my opinion, potential distributors of
> Python need to be told so that they can make an informed decision
> on whether to risk infringement.

For versions of Python after 1.2 (the last released by CWI) and before 1.6b1
(which was released this weekend from CNRI under the new CNRI Open Source
License), both of these are true:

+ The CWI license is the only license there is, valid or not.

+ CNRI is the copyright holder.

IANAL, but that suggests to me that any binding statement about this must
come from CNRI.

> Debian for example has made the decision that because there might
> be a legal risk that the GPL is not compatible with the QPL, they
> cannot distribute KDE linked with Qt.  This is despite the extreme
> unlikelihood of either Trolltech or the copyright holders of the
> KDE code ever suing Debian for such distribution.  On the other
> hand, Red Hat has made a different decision.
>
> I would hope that the final agreement between CNRI and BeOpen would
> include a clear indication one way or another whether the purported
> license of Python 1.5.2 and 1.6a2 is valid in the eyes of CNRI.  After
> all, 1.6 is not released yet, so many distributors will be including
> 1.5.2 for some time.

Well, I can't argue against you on this!  If CNRI claims they released
software without a valid license, the legality of using 1.5.2 and 1.6a2 is
muddy (indeed,  CNRI may not even agree 1.6a2 was "a release" in the sense
of 1.5.2).  Since it seems very unlikely they'll agree to say that the CWI
license is valid, perhaps they could be persuaded to promise not to press
any claims based on the presumed invalidity of the CWI license excepting
claims against BeOpen PythonLabs.  I've copied my bosses on this, but,
again, it has to be taken up with CNRI directly.  Have you asked them?
Since 1.5.2 and 1.6a2 were not BeOpen releases, and there's been no hint of
trouble with them yet, I don't *think* BeOpen will be highly motivated to
press this.

> Thanks to Tim Peters for his honest, informative, and
> thoughtful answers to these troubling questions.
>
> Henry Jones

I'll cop to all that, provided you add "useless" too <wink>.

hoping-to-reincarnate-as-a-sentient-gatling-gun-ly y'rs  - tim