Re: GPL flaw?

[bts@alum.mit.edu (Brian T. Sniffen)]

Mark Schreiber <mark7@andrew.cmu.edu> writes:

> I have a small quibble with the GPL on a point that seems that it
> could be improved, unless, of course, my interpreation is incorrect.
> The existing license seems to produce undesireable behavior in a
> particular (admittedly, unusual) case.

You are horribly confused.  I think you meant to send this to
licensing@fsf.org.  However...

> Let us suppose that a company (Small Company) produces a software
> package (Smart Writer).  Small Company sells their software for years
> under a closed-source license.  Small Company hears about the value
> that GPL-licensing their software provides, and decides to GPL
> (v2)-license their software.  Small Company produces several releases
> of their Smart Writer.  They do not use any code contributions from
> the community.

Keep an eye on that last sentence: if they used no code contributions
from outside, the problem you mention below seems unlikely.
However...

> Smart Writer is a very good package, and large chunks of code from it
> are used in a stunning array of GPLed packages.  RMS himself includes
> ten thousand lines of code from Smart Writer in core code in his new
> package (Fast Lisp Interpreter), interweaving it throughout Fast List
> Interpreter's codebase.
>
> Small Company is then purchased by a much, much larger company
> (Macrosoft, Inc).  This includes "all current intellectual property
> assets".  Macrosoft wishes to prevent others from using its new IP
> freely in GPLed software.  It goes back, and discovers that one of
> Small Company's developers included (in a version of Smart Writer
> predating GPL release) twenty lines of code from a software package
> that Small Company did not have a license to, and is not available
> under a GPL-compatible license.  Nobody in the GPL community has used
> these twenty lines of code in their software.

It's questionable whether anybody can copyright twenty lines of
typical code.  Sure, there are 20-line snippets that have creative
work, but that does seem a bit unlikely in a
tens-of-thousands-of-lines work.  However...

> Normally, this is not a problem -- a company may remove offending code
> and do a re-release.  However, Small Company no longer controls Smart
> Writer.  Macrosoft is in charge.

You are confused about several points.  Here's the first one: Small
Company gave a license to use its copyrighted works to those who
received copies of Smart Writer (GPL Edition).  If it couldn't have
granted a license to use part of that package, that's fine: the
license to the rest persists.  So RMS and the users and distributors
of FLI are not affected by this.

> Macrosoft asserts that because Small Company did not have ownership of
> the full Smart Writer codebase, Clause 7 of the GPL applies
> (prohibiting distribution under the GPL unless such distribution may
> be done legally) to all releases of Smart Writer that Small Company
> attempted to perform.  Macrosoft thus claims that the Smart Writer
> releases were never legally released under the GPL.

Macrosoft needs better lawyers: clause 7 does not apply to Small
Company, because they are the original authors.  They don't *need* a
license to distribute or modify Smart Writer, because they hold the
copyright.

> RMS asks Macrosoft to re-release a Smart Writer release with the
> offending twenty lines of code removed.  Macrosoft refuses, and says
> that the GPL-using community must instead remove all instances of
> Smart Writer code from all GPL software, or it will sue for copyright
> infringement.  In the meantime, the GPL-using community must cease
> distribution of all software using Smart Writer code.

This sounds a lot like SCO's recent nonsense; perhaps they confused
you.  Either way, this scenario is not plausible.

> Even more disturbing, let us assume that RMS passes away from computer
> monitor radition before he can finish excising the ten thousand lines
> of Smart Writer code from his Fast Lisp Interpreter package.  Since he
> owns the Fast List Interpreter copyright, and has never done a legal
> release of Fast List Interpreter, this package has never been GPLed.
> His heirs refuse to GPL-license the Fast List Interpreter package.
> Thus, Fast Lisp Interpreter is not GPLed.

Not quite: RMS had a license from Small Company to modify and
distribute all the code he used for FLI, so it's available to anyone
who has a copy under the GPL.

> In the GPL v2, clause 6 states:
>
> "Each time you redistribute the Program (or any work based on the
> Program), the recipient automatically receives a license from the
> original licensor to copy, distribute or modify the Program subject to
> these terms and conditions."
>
> Perhaps clause 6 should specifically state that license(s) are granted
> to all portions of a Program released under the GPL.  This would mean
> that even if the case of the GPL license applying to the Program being
> shown to be invalid, any GPLable portion of the Program is still
> available to anyone using source from that Program.  This way, if
> someone in the GPL community had used the twenty offending lines, they
> would have to remove...those twenty lines.  The remainder of the
> codebase would still be GPLed.

That's already the case, because of how combined and joint works are
treated under copyright law.

-Brian

> Thoughts?  Perhaps I've misinterpreted the GPL, or missed some portion
> of a clause that applies.  It would be nice to know that this isn't an
> issue.  :-)

-- 
Brian T. Sniffen                                        bts@alum.mit.edu
                       http://www.evenmere.org/~bts/