Still no new license -- but draft text available

[Tim Peters]

[Peter_Halvorson at nfuel.com]
> Show stoppers, hmmm ....  Are Python 2.0 and/or 3000 free and clear?

I'm not sure what "free and clear" means to you.  I'll explain what I think
I know (which isn't much!), and if that doesn't cover it, ask again and I'll
make stuff up <wink>:

+ In the language of the CNRI Open Source License for Python 1.6,
  Python 2.0 will be a "derivative work".  As such, it will also be
  under the terms of the CNRI license.  Ditto 2.0's successors, so
  long as we keep building on the 1.6 codebase.  So everyone can
  expect to need to live with this license a long time (at least
  a few years).

+ The status of P3K can't be guessed at this time.  If it builds on
  the 2.0 codebase, then it will again be a derivative work.  But if
  it's a rewrite from scratch, I think the zoo of copyrights and
  licenses on the old codebase can be tossed.  Indeed, this can be
  seen as an argument *to* write P3K from scratch!  I don't see
  much in that argument, though, unless the rewrite is either public
  domain from the start, or copyright is assigned to an independent
  legal entity created for the purpose.  That is, just replacing one
  employer's "intellectual property" shenanigans with another's
  wouldn't strike me as being real progress.

> Are they subject to the CNRI license as derivative works?

As above.

> Can CNRI have a change of heart at any time and come up with
> a new license which applies to 2.0 as well?

They could *try* to, sure.  But everyone has been negotiating in good faith,
and the *agreement* is that if the proposed 1.6 license is "good enough",
BeOpen PythonLabs will release 2.0 with the exact same (modulo "1.6b1" ->
"2.0", and-- I guess --a different "handle") CNRI Open Source license.  I
wouldn't worry about this.  If CNRI tried to make substantive changes in the
license between 1.6 and 2.0, the agreement is broken so we wouldn't feel
bound by it either.  Fundamentally, trust Guido here!  If he feels the
community is about to be screwed, he'll walk.  He may have sold out, but he
doesn't come cheap <wink>.

Of course there is nothing to stop CNRI from doing its *own* 2.0 release
under a totally different license, but that would have no bearing on the
BeOpen PythonLabs release one way or t'other, and they're not going to do
anything like that anyway.

> Is this a risk to using the CPython code as a base for future
> Python development?

Study the proposed 1.6 license -- that's why Guido posted it.  Once a
release is made under that license, it's a base from which further
development can proceed.  CNRI can't *retroactively* change it, and, e.g.,
if they decide they want a new license on 2.1, and we don't like it, the 2.0
license certainly appears to me to allow us to do 2.1 as a derivative work
of 2.0, thus retaining 2.0's license in all respects.  That is, this, *now*,
is CNRI's first & last chance to force a new license on GCPython (== the
version of CPython associated with Guido).

> If 2.0 can have whatever license GvR likes, then who cares about
> the 1.6 license!

Alas, my understanding is that the licensing issues end up in court if we
can't agree on using this license for 1.6 and 2.0.  If CNRI would like to
say "but do whatever you please with 2.0", of course that would be great
(Guido has said he would move to a slight rewording of the BSD license then
(rewording just to get references to "the regents" out of it)).  Heck, if
CNRI would like to say "Ho! It just occurred to me that there was no actual
need to do any of this! Must have got up on the wrong side of the bed that
day. Oh, well -- nevermind.", that would pretty cool too <wink>.

> I trust a benevolent dictator further than I trust a benevolent
> corporation or government.

Indeed, I would love to release Python under this license:

    1. This LICENSE AGREEMENT is between the Individual or
    Organization ("Licensee") accessing and otherwise using
    Python 1.6, beta 1 software in source or binary form and
    its associated documentation, as released at the
    http://www.python.org Internet site on August 5, 2000 ("Python
    1.6b1"), and Guido van Rossum, whom Licensee, by virtue of
    accessing or otherwise using said software or documentation,
    acknowledges is their Benevolent Dictator for Life ("BDFL"),
    and whom Licensee undertakes never to displease by thought,
    word or deed.

    2. Licensee agrees to do unto others as they would have done
    unto them, unless Licensee is a masochist, in which latter
    case Licensee agrees to answer 50 questions per day on
    comp.lang.python, or to perform such other odious tasks as the
    BDFL may assign to them from time to time, at His sole
    discretion and pleasure.

    3. Licensee agrees that the BDFL is not responsible for
    anything bad that happens.  Licensee agrees to publicly praise
    the BDFL for everything good that happens.

    4. Licensee agrees that this clause is not self-referential.

    5. Licensee agrees that defining the difference between
    clicking and not clicking is an epistimelogical impossibility.

    6. Licensee agrees that there is only one way to do it.

some-dictators-for-life-are-better-than-others-ly y'rs  - tim