[tlug] License regressions (GPL branch)

["Stephen J. Turnbull" <stephen@example.com>]

burlingk@example.com writes:

 > of the other GNU liscenses.  It lets you give away important
 > chunks of code, without showing off the stuff you want to keep
 > to yourself.  That is a programmer/company's right as well. Some
 > people are very private, and some are very paranoid.  For those
 > people, the LGPL is a much better solution than the GPL.

You're mixing up the "rights" of the licensor and the licensee here.
The licensee has essentially no rights *under current US law* (there's
stuff like fair use, but that is explicitly not intended to cover
commercial activity, and there's no fair use in patent).  They may be
granted a license by the licensor, but that doesn't create any
"rights", and a unilateral license like the GPL may be unenforcable by
the licensee (even if it contains language explicitly giving up the
right to revoke, as the GPLv3 does).

OTOH, if one has an *ethical* right to source code, as maintained by
the free software advocates (whether fanatical or not), a so-called
right to privacy for the developer makes never-no-mind, especially
considering that for the GPL (or similar licenses) to bind, you have
to actually distribute.

 > Anyway, onto my point.  The spirit and word of the BSD 1.0

I believe you meant OSL v1.0?  I'm going to assume that.

 > liscense covers exactly what the GNU foundation claims to want to
 > cover with the GPLv3.  All the little details are there, without
 > all the legalistic and politically correct crap.

Actually, it does not.  Larry Rosen (author of the OSL, AFL, and what
I consider the best book on the subject: _Open Source Licensing:
Software Freedom and Intellectual Property Law_) doesn't believe that
viral licenses actually "work", even if they did he thinks that would
be wrong, and the OSL is carefully worded to avoid implying a claim on
others' code unless it is textually derived from a copy of code they
received under the OSL.  In this it is closer in spirit to the LGPL.

The GPL, on the other hand, is (and always was) intended to be viral
(Bill Chen once said "if you don't like the word 'viral' because
viruses attack their hosts, how about 'AIDS-like'? just like AIDS, you
have to exchange bodily fluids to catch the GPL").  The GPL does the
best it can to make any claim on others' code that it can.  The
"magnanimous" "exceptions" like "mere aggregation" are no such thing;
there simply is no basis in (US) copyright law for making claims on
the other components of an aggregate -- if there was, they'd make it.
(See RMS's essay advocating the GPLv3 where he says "Ideally, we would
make everyone who redistributes GPL-covered code surrender all
software patents.")

The GPLv3 itself makes this goal clear, as it attempts to break down
the exec(3) boundary and inject the virus into separate processes (see
footnote 12 of the rationale).

 > In GPLv1 and 2 they intentionally avoided the issue of patents.
 > They swore that it would not be an issue they would touch.

Well, that's not what RMS has said that I've seen.  His belief was
that although the viral aspects and enforceability of the GPL are
founded in copyright (since most open source developers have no
patents in their software, while copyright is automatic), the grant of
license awarded via the GPL is a *software* license, not a "copyright
license".  AFAIK, this is true; Larry Rosen has said something
similar.  Thus it implicitly licenses any patents owned by the
licensor.  Unfortunately, it turns out that software patents are
stronger, more complex (so that the implied grants of license are
really unclear), and (now) more common than he wanted to believe in
1991.

 > [GPLv2] was not the greatest document in history, but it at least
 > was within the spirit of the GNU movement.

Well, RMS defines the "spirit of the GNU movement", along with "free
software" and "free documentation".  (The "M" apparently stands for
"Humpty Dumpty". :-)  Since he likes the GPLv3, it must be in the
spirit of the GNU movement.

 > So much of the effort waisted on the GPLv3, and the reason it
 > took them so long to hash it out, was so that they could find
 > a way to legaly take away the developers rights to make that
 > type of decision concerning their products and their code.

I, too, dislike the GPLv3 but please, no FUD.  The licensor has always
had that power.  RMS (and a fortiori, the FSF) has never tried to hide
the fact that the GPL contains important restrictions on downstream
developers' freedom.  The claim is that these restrictions are the
minimal ones that maximize the freedoms available to "downstream**n"
for n > 1.

That is still the goal of the FSF; I see no reason to believe anything
else.  The reason it took so long is that private copyright is not
really strong enough to fight public policy, as embodied in the DMCA
and software patents.  Attempts to make it do so (especially given
RMS's explicit goal of expropriating all software patents) scared big
corporate backers like IBM (inter alia), and negotiating those license
grants was hard.  Similarly with the anti-tivoization clause (with its
wonderful "if you put it in ROM you can blame the FSF license for
preventing you from providing upgrades by flash ROM" exception).
Remember, the GPL is not really a license; it's a constitution for the
sovereign state of Freesofantasia.  Writing constitutions is hard
work, just ask Mr. Abe. ;-)

However, I think that the GPLv3 is a poor instrument for achieving
such a goal.  About a year ago, I had an exchange (private) with RMS
(who initiated it) over some of the new terms.  He acknowledged that
their legal interpretation is unclear until tested in court, and the
effectiveness dubious.  "But," he said, "we *must* try to fight DRM
and software patents using the license."  Viva RMS Quixote!