GPL compliance

[Mario Vukelic]

[Please follow up on sounder, it's cc'ed]

On Sun, 2006-07-02 at 13:30 -0700, Gary W. Swearingen wrote:
>    (at your option) any later version.
> 
> NOTE^^^^^^^^^^^^^^  and no usage of "both" or "and" in the key phrase.
> 
> THAT (and/or clause 9) permits a GPL licensee to distribute his
> derivative under ONLY ONE version.

I don't think that anything prevents him from distributing one copy
under v2 and one copy under v3 (or at the very least person A to
distribute v2, person B v3)

>  It's my rather fuzzy
> impression that most GPL licensors pick only one, v2, glad to say, but
> many don't.) 

But the point is that if a program is already licensed under v2, the
copyright holder holders can relicense it as v3, but this does not
remove the earlier rights you already had to the latest v2 version. You
can fork off there. It is my understanding that it is not possible (alos
not for the copyright holder) to use patches that are v3-only in a v2
codebase - but I might be wrong here.

> And that could be incompatible with licenses that
> were compatible with v2 and still in use on non-GPL parts of the
> software.  That's what I meant by becoming illegal or whatever I said.
> Of course, it would be the publishing of that derivative that would be
> illegal. 

I'm sure if such cases exist they will be figured out long before the
real v3 (not the strawman "v3b" you hinge your argument on) is released,
and the FSF will publish instructions that will be peer-reviewed.

> It's derivatives that can be closed, because
> derivers are given more freedom to control _their_ work.)

And it's exactly the same concept with v2 and v3. If you don't accept v3
(say, because it's too closed for you), can, "at your option" stick to
the latest v2 version and fork off.