Re: Linux and GPLv2
On Sun, Mar 13, 2005 at 03:24:24PM +0100, Måns Rullgård wrote:
> We have to consider the possibility that GPLv3 will say something we
> don not want. Then we do not want people distributing it under those
> terms. Never give permission to do something you do not know what it
There's a more significant problem: if you say "or later", and you don't
like GPLv3--either because it allows things you don't like, or (as
recent events suggest may be more likely) includes restrictions you
don't like, people can take your work, modify it, and place their
modifications under GPLv3-only. If you want to keep your code available
under GPLv2, you can't incorporate those changes, since they're not
available under v2.
Considering that a primary motivator of the GPL is to prevent the case
where you can't incorporate other people's enhancements to your work
due to licensing, this seems like a potentially major failure.
> Only if all the copyright holders agree. Suppose A has accepted
> contributions from B, with the "or later" option, and it turns out
> that A does not approve of v3. Now B refuses to drop this, so A is
> effectively forced to distribute his code under a license he does not
> approve of.
You can always drop the "or later" from *your* contributions (not
retroactively to previous licensees, of course, but to your future
modifications). The only way the "or later" clause is applicable
to an entire work is if every piece of its code has it.
> The "or later" gives the FSF more flexibility to change the license
> terms for a vast amount of software they really have no connection at
> all with, with or without the approval of the copyright holders of
> said software.
The "or later" is intended, as I understand--from rational logic, as
I don't believe I've seen any statement from the FSF--to allow the
FSF to fix problems in the GPL. Without "or later", it's impossible:
the only way a "bugfixed" GPL could be used is after getting explicit
permission from every copyright holder of a GPL work. Further, and
just as importantly, the "bugfixed" GPLv3 would be incompatible with
the original GPLv2! That would fragment the GPL at a fundamental level.
That would be fine, if the FSF had maintained its traditional consistency.
Frankly, they broke that trust with the GFDL, and so I'd sympathise with
people no longer willing to use the "or later" language. The above
problem doesn't go away, though.