Re: asterisk and mysql_cdr
On Sat, Feb 26, 2005 at 07:06:23PM +1300, Nick Phillips wrote:
> > You must be able to remove the exception--that is, I must be able to
> > distribute the software under the plain GPL, without additional
> > exceptions. If I can't do so, then your exception itself is making
> > the license GPL-incompatible.
> No. I think whoever you were quoting there was correct. Remember that
> the permissions granted to your users/downstream by the GPL when you
> distribute a work come directly from the copyright holder, not from
> you. You have no place in the permission-chain in that situation, so
> cannot modify the permissions granted.
Sure. The copyright holder is free to prohibit me from removing the
exception (that's not what I meant)--but if he does so, the result is
GPL-incompatible, and that's almost never intended. To implement an
exception in a way that doesn't do this, the copyright holder must
grant some way for me to remove it when redistributing (eg. dual-licensing).
> > You *can* grant exceptions to the GPL, but if you prohibit the removal
> > of those exceptions, you've created something which is GPL-incompatible.
> If you were to prohibit the use of your software in a combined or derived
> work which was unable to grant those extra permissions, then yes, that
> would be an extra restriction.
When I see is software under a license that looks a lot like the GPL,
except it has two differences: one restriction is removed (due to the
exception), and one restriction is added ("you can't add the restriction
I waived back in"). I see that first, and a direct effect of that is
that you're prohibiting the use of your sftware in a derived work which
is unable to grant those extra permissions.
However, we're just looking at it from different angles; I'm pretty
sure we're agreeing.
This is why I've tried to be sure that all GPL exception clauses d-legal
recommends explicitly permit removal of the exception (eg. dual-licensing
with the original, untouched GPL, or something to that effect). Having
to deal with adding exceptions for OpenSSL, etc. is bad enough--we don't
want to do it wrong and cause further headaches down the line.