The good news:
> > Can you give me any kind of feel for how this re-examination is
> > progressing? If a release under a new license is likely within a short
> > period (I'd consider a month to be a short period), then we'd be happy
> > to wait for the new license. If the release under a new license is not
> > likely, or likely to take a long period (I'd consider a year to be a
> > long period), then we'll need to ignore this potential notice.
On Tue, Sep 12, 2000 at 08:18:23AM -0700, Lori Stevens wrote:
> Hopefully, within a few months. I'm hoping for closer to 1-2 months.
[Jaldhar is pursuing what the new license is going to be.]
The bad news:
> > Do you maintain that "this software" in "Permission to use, copy,
> > modify, and distribute this software and its documentation for
> > any purpose and without fee is hereby granted" is no longer "this
> > software" if it has been modified?
> Yes, the phrase "this software" definitely refers to our distribution,
> not to any derivative works based on it.
And, after having exchanged quite a bit of email from her, I'll add:
 She'll only answer specific questions, and
 Any questions on this topic will yield either
"Anything that's not specifically permitted is forbidden",
or a variation of the above (basically claiming that "this"
only refers to an exact copy of what UW distributes).
Finally, since she's granted Debian specific permission (the permission
had limits, but they're not the sort of limits that a Debian developer
would violate without also violating debian policy), they're not going
to be suing Debian.
I'm tempted to accept their logic. But what does that mean for something
like amanda, bind, blt, bonnie, bootp, cfs, clips, cutils, cxhextris,
etc., which use a similar phrasing (or bible-kjv which uses different
phrasing but where the same underlying logic applies).
[Actually, in passing, bind has a part of it which is distributed under
a DNSSAFE license, and I'm not sure that's DFSG -- it has restrictions
on modifying it for other uses. I should note in passing that DNSSEC is
almost completely useless in terms of protection from DNS forgeries --
though that could change, if the root servers started supporting it.
[And, before you hold your breath: the DNSSEC rfcs are seven years
old already, but no rfcs have been issued to define an authentication
infrastructure which would let the root name servers support DNSSEC..]]
Also, note that ckermit has an explicit requirement that permission be
sought before changes can be made (there's a claim that permission won't
be unreasonably denied). [I've not filed a bug requirement on this one,
by the way.]
Also, note that edict relies on this same phrasing, but makes it clear
by implication that distribution of modified versions are permitted.
[And, cweb uses similar phrasing in a similar fashion.]
Finally, note that I only looked at licenses for packages begining with
[a-e]. There's over 400 more that should be looked at. [And not just
because of this issue -- I had to file half a dozen bug reports on
packages that we include in main which didn't grant any permission to
distribute modified versions.]