Subject: Re: Tarball licenses (was: Free documentation licenses)
From: Rick Moen <rick@linuxmafia.com>
Date: Fri, 1 Dec 2000 13:51:09 -0800

begin John Cowan quotation:

> And of course the reason homeopathic pills don't work is that the efficacy of
> pills depends on the current state, not the historical origin.

Don't tempt me towards an off-topic disquisition, but you'll want to
look up the "Law of Similars" and the "Law of Dilution".  It was a
decent attempt at a scientific theory when Samuel Hahnemann came up with
it a couple of hundred years ago, but the ad-hockery still being used to
justify it has become amusing but somewhat sad.

>> Are you saying that interpretation is wrong, and that, but for the
>> lawsuit settlement, BSD would have remained subject to AT&T's licence
>> terms even after all AT&T code was gone?
> 
> No.  AT&T's claim was that some of their code was *not* gone,
> particularly in certain drivers.

That's _not_ the question I was trying to ask.  I meant, "_Had_ the
lawsuit not interrupted the process, and all AT&T code been
replaced...."  I was saying that it would _appear_, from your posted
criteria, that 4.4BSD was a derived work from AT&T code just as much as
any A1 module would have been of Alice's module A, regardless of any
subsequent changes.

> The University claimed independent invention, and pointed out that
> AT&T's hands were not clean either: they had appropriated some old-BSD
> licensed code while ignoring the advertising requirement.

I am quite aware of the latter, having heard (among other things) the
long version of McKusick's talk on the subject.  But the question of
whether CSRG's derived work -- regardless of _which_ AT&T codebase it
branched from -- retains licence obligations to the original still
strikes me as significant. 

Weren't you saying that a derived work remains a derived work, even
after all the original content has been replaced?  If so, then why
should it matter if the original modules were eventually replaced by 
wholly new modules that interact in wholly new ways?

> It was you who said "it's pretty clear that Beirne doesn't object".
> How is it clear?

Well, maybe it's not clear.  Maybe that was a vague handwave of my own, 
having been infected by the sloppiness of sundry programmers.  ;->

So:  Beirne issued makedoc7.cpp without an explicit licence.  I
_suspect_ that he did likewise for his MakeDocJ (but don't have a copy).
Krzysztow says somewhere (his Changelog?) something that made me think
Beirne was involved in some way, but I can't remember details.  But, the
more I've been trying to properly and scrupulously classify my PalmOS
software collection, the more skeptical I've become of programmers'
vague licence handwaves -- and of my own earlier ones.

I _am_ attempting to track down Beirne, out of a stubborn desire to 
fix the licence status of what is, objectively, some fairly
insignificant code.  But the need (however reduced by the antiquity and
insignificance of the project) to _attempt_ this, with no guarantee of
success, underlines why licence information really needs to accompany 
the affected code in the first place.

-- 
Cheers,                                      "Reality is not optional."
Rick Moen                                             -- Thomas Sowell
rick@linuxmafia.com