Re: [Fsfe-uk] GPLv3 draft available

[Chris Croughton]

On Tue, Jan 17, 2006 at 11:37:14PM +0000, J. Grant wrote:

> Plugging the dynamic linking gap at last. Although it is pretty easy to 
> define an alternative interface to a software dynamic link. Anyone who 
> really wants to use a GPL'd program could just write a GPL'd interface 
> program binary, which accepts commands from a pipe, and then their 
> proprietary software can utilise the program via the pipe. They could 
> even just output there data to a file, and have a program binary crunch 
> through it, then read back in the results.

As I read it, it looks as though if your program is designed to use
other shared libraries (X, glibc, etc.) then you have to make the source
code for them available as well (unless they are both an "essential
component (kernel, window system, and so on)" AND they are "only to
enable use of the work with that system component or compiler or
interpreter, or to implement a widely used or standard interface").
While that might apply to glibc and X, it seems to preclude code which
might be designed to interface to other systems' libraries.

As far as Section 7 (other licences) is concerned, they have the option
to distributes parts I write under (aditionally) any other licence as
long as it is not more restrictive, is that correct?  And if so, does
that include the permissive licence:

  You can do anything you want as long as you don't pretend that you
  wrote the bits I wrote or that I'm responsible for modifications you
  made.

(That's a paraphrase of the Zlib licence, basically banning "passing
off" authorship.)

> I have a query Re the patent clause, does this mean that as soon as Sun 
> run some GPLv3'd software and make it available that everyone has a 
> licence to the patents it builds upon?

When running that program or a derivative version, yes, in my opinion
(IANAL).  However, it wouldn't (I suspect) cover you writing your own
code using the same patents.

For instance, say XYZ Company has a patent on an encoding method, and
they licence it to Sun.  Sun then wants to put their code under the GPL,
and asks XYZCo for permission.  Since the code is the implementation of
the public patent, there's no problem in that, nothing (except Sun's
implementation work) is being given away.  And that gives you the right
to modify Sun's code, distribute it, etc.  However, if you wrote your
own code based on the patent which is /not/ derivative of Sun's
implementation that would not give you any right to use it, because the
patent itself isn't covered by the GPL only that implementation.

(It's an interesting situation where you would want to prove that your
code was derivative, usually people want to prove the opposite!)

GPLv3 covers that, by saying:

  If you distribute a covered work knowingly relying on a patent
  license, you must act to shield downstream users against the possible
  patent infringement claims from which your license protects you.

In other words, if the author (Sun) has a licence for the patent, they
must act to protect the users (who don't have the licence) from claims.
But theu don't have an obligation to protect anyone else from those
claims, they effectively only extend the patent license to those who use
their software.

I suspect that if it were otherwise the patent holders would just refuse
to allow the infringing code to be issued under the GPL, which (if it is
an essential part of the code) would mean that it wouldn't be issued at
all under the GPL.  The wording in GPLv3 seems to allow a loophole which
is acceptable to patent holders as well as users, in that the patent can
be licenced only for specific programs.

(That's how I read it, anyway, but IANAL...)

Chris C