Re: What is "free software"?

> "Eric S. Raymond" <esr_at_snark.thyrsus.com> writes:
> > I think I can safely assert that neither Zeyd nor myself wants to
> > prevent OpenBSD from making necessary local adaptations.
>
> That is good. Unfortunately...
>
> > That we assert a right to control the redistribution of modified
> > versions does not mean we want to forbid all such modifications.
> > That reservation is not intended to hoard the code. Rather, it is
> > intended to prevent abuses like Thomas Dickey's.
>
> The existence of the provisions that allow you to forbid modifications
> is---to be dramatic---a loaded but uncocked gun held at the head of
> projects like Debian or OpenBSD.

I understand why you might feel that way. But I also see an inconsistency.
If this is such a deal-breaker for Debian, then why do its guidelines say:

> The Debian Free Software Guidelines
>
> 1. The software may be redistributed by anyone. The license may restrict
> a source file from being distributed in modified form, as long as it
> allows modified binary files, and files that are distributed along
> with the source for the express purpose of modifying the source.

It appears to me that this paragraph specifically *permits* licenses
like that of ncurses. In fact I don't see how it could have been
written with any other intention in mind.
 
> You have just assured us that you would never fire it, because you
> like us, but until you throw away the gun, we have no guarantee of
> safety, and the fact that you have just recently fired this gun, I
> hope you understand, might give people reason to contemplate the
> origin of the term "gun-shy".

I can certainly sympathize with this position. To alleviate your unease
I can only point to my long record of service and cooperation in the free
software community. That ought to count for something. If you can't
trust me, we'd better *all* give up and go home.

> As the former Debian maintainer for ncurses, I can state that I am
> unwilling to take the risk associated with releasing modifications
> without an unequivocal guarantee that I would not be legally liable.
> And the current license does not give me that guarantee, as your
> actions against Tom Dickey have displayed.
>
> So, whatever my opinions about the current dustup, the simple fact is
> that your license *does* make it unattractive for organizations that
> do not want to incur the possibility of legal action to use ncurses.
>
> And the only thing you could reasonably do to remove this issue would
> then make it perfectly legal for Tom Dickey to fork an ncurses
> distribution.

I must respectfully differ with this assessment. The Debian
guidelines imply the possibility of an ncurses-like license which
allows maintainers a veto on modified distributions, while explicitly
granting system integrators a non-revocable right to apply local
patches as long as a pristine source tree is present.

> You also have to factor in that special licensing for Debian would not
> be adequate, because the Debian wants to be able to be used as a basis
> for other, more specialised projects, so we must be able to pass that
> modification right on to others.

I accept this argument.
 
> Moreover, what would happen if Tom Dickey sent patches to the Debian
> maintainer, and then suggested to people that they pick up the Debian
> package, instead of your distribution...

I am willing to trust the community's judgement, and compete with him
on that basis if need be.

The key thing that the Debian guidelines have shown me is a way to
craft a license which both preserves the rights of authors to veto
modified distributions, *and* which provides legal security to system
integrators making local modifications. This is a win-win situation --
a licensing style that I would not merely tolerate for the sake of
ending the crisis, but actually feel happy about afterwards.

I suggest we explore the possibility of such a formula.

-- 
		Eric S. Raymond