[Python-Dev] Still no new license -- but draft text available

Tim Peters tim_one at email.msn.com
Fri Aug 4 01:08:09 EDT 2000


[From the proposed CNRI Open Source License]
> 2. Subject to the terms and conditions of this License Agreement, CNRI
> hereby grants Licensee a nonexclusive, royalty-free, world-wide
> license to reproduce, analyze, test, perform and/or display publicly,
> prepare derivative works, distribute, and otherwise use Python 1.6b1
> alone or in any derivative version, provided, however, that CNRI's
> License Agreement is retained in Python 1.6b1, alone or in any
> derivative version prepared by Licensee.

[Gary Momarison]
> I see a show stopper there.  Well, I think it SHOULD be a showstopper,
> but I doubt that it will be.

What you go on to describe is not a show-stopper for BeOpen PythonLabs,
because, as you say later, the CWI license was similar in the respect you're
worried about (FWIW-- which is nothing --I think the CNRI license is more
clearly in your favor than was the CWI license on this point).  For that
matter, so is the BSD license, which is what we would have *liked* to put on
Python 2.0.

Is there *any* OSI-certified and GPL-compatible license that does not rub
you the wrong way?  If not, and you want to call this a showstopper, then
our only alternative is to put Python to sleep -- fun while it lasted, but
that's that.  Ain't gonna happen, cuz Fred Drake has way too many kids to
feed <wink>.

> That "...provided, however, that CNRI's License Agreement is
> retained" clause looks deadly to me.  If a derivative work retains
> that License while also getting some new License which most people
> would want to apply, then conflicts seem inevitable.
>
> For example, say I want to use some Python function in the
> interpreter of my own new-and-improved language which I want to
> license as a binaary under royalty to BigCorp.  So now the
> derivative work carries two licenses, one which says royalty-free
> and one the opposite.
>
> Or, say I just want to modify the syntax a little to create my
> new Cobra language to license for royalties or release without
> source or whatever.  Same problem.

When people asked Guido about stuff like this wrt the CWI license, his
response was that the inherited license clearly <wink> applied only to the
inherited part of the combined product, in effect to keep a record of the
legal history of the software you started with and to prevent CWI from
getting sued.  But note this from the CNRI license text:

    Licensee [has a] nonexclusive [etc] license to reproduce [etc]
    and otherwise use Python 1.6b1 alone or in any derivative version
    [yadda yadda yadda]

It does *not* say Licensee has a license to do all that stuff with your
derivative work, it explicitly names Python 1.6b1 as the only thing they're
getting a license for (and they should!  you most certainly are not getting
the right to prevent anyone else from using Python 1.6b1, derivative work or
not).  The CWI license was much muddier on this point to my eyes.

> Some will say that if "A" licenses some rights over code to the
> public (say "B" and "C") like this license does, then "B" can then
> license a derivative to "C" with his new license covering only his
> work.
>
> Others will say that "B" can license the derivative as long as he
> satisfies the original license by retaining that license (for
> historical purposes only?) and as long as his new license doesn't
> try to give away more rights than the original did.
>
> Both of these rationalizations seem bogus for reasons I'm too lazy
> to go into now.  I'll just say that I too would like them to be
> reasonable, but I see no legal justification.

Are you a lawyer?  I don't see a justification for much of what goes on in
Perl either, but then neither am I Perler by training <wink>.  That is, if
you're not trained in the law, how could anyone satisfy you?  A killer
argument to a lawyer isn't a Usenet-y argument from first principles, it's
usually based on a long string of citations into the massive body of case
law, statutes and precedents that are *their* language of discourse.  I know
I couldn't tell a good legal argument from a bad one if it bit me on the
ass.

> There is a whole other line of discussion regarding derivatives
> and the separation of differently-licensed code, but, again, it's
> too much for now.
>
> But I ask this: What does "is retained" mean?  What are the
> implications?  By what legal theory do licenses "stack"?
> If the derivative retains the original license which gives me
> a right, why must I observe some other license which trys to
> take it away?  I have no way of knowing what code is original
> and which is not.
>
> P.S. The old Python and X11 licenses have the same problem. The
> BSD license might leave a loophole.  It seems that people have
> "known" what the licensors "meant" to imply by these and nobody
> has cared to get finicky lawyers to be nasty about these licenses.
> And I know enough about lawyers to know that most of them don't
> bother thinking through an issue if nobody else is worring about it.
>
> P.P.S. Please don't tell me that Clause 2 above is OK with RMS.
> I've read that and it seems irrelevant without an explanation
> that addresses my concern.

We asked RMS to negotiate with CNRI solely to resolve the issue of GPL
compatibility, and I agree that's got almost nothing to say to people
interested in proprietary uses.

> I care about what courts will think first,

Well, obviously, this isn't a court, and AFAIK the old CWI license was never
even mentioned in a court.  The proposed CNRI Open Source License may or may
not be, but has not been yet.  Can only suggest you retain a lawyer or ask a
law professor if you want a *competent* opinion about what courts will
think, and aren't persuaded to take a chance based on the courts' failure to
kick anyone in the teeth yet based on your reading of similar licenses.

> then what the Python owners think,

Well, CNRI is the Python copyright holder, and nobody here can speak for
CNRI either.  Have you tried asking CNRI directly?  I don't believe they
read this newsgroup, so posting here is not asking them (and since I've said
several times in the last week here that I don't believe they read this
newsgroup, if they *do* read this newsgroup they're clearly not interested
in letting people know that).

> and THEN what some derivative licensor (which could be me or
> RMS) might think.

BeOpen PythonLabs will be a derivative licensee, same as you.  An upbeat
statement from BeOpen's CTO was posted here a few days ago, for what that's
worth to you.  If I were to use Python in a proprietary way on my own, I
would not be worried about this, but I would retain a lawyer to study the
license first -- as would any prudent businessperson entering into a
contract or using any other licensed material in their business.

> And, of course, I care what you think.

You shouldn't!  I'm incompetent to address legal issues, and so is Guido.
These aren't matters of plain fact, and your guess is as good as ours.  I
can promise *I'll* never sue for shipping a derivative work of the kinds you
described.  I expect it would be easier to extract several billion promises
like that from everyone on earth, one at a time, than to satisfy you that
you'll never get screwed in court on a matter that's never been tested in
court <0.3 wink>.

or-if-this-pacified-you-start-wondering-about-what-the-laws-
    say-in-germany-and-swaziland-too<wink>-ly y'rs  - tim






More information about the Python-list mailing list