Re: Questions concerning S/390 OCO-modules
> > Actually, the requirement is a little different from that. We have to
> > ensure that anyone who uses the program will comply with the license.
> > They don't have to have read the license, so we don't have to have a
> > post install pop-up (which would be incredibly annoying). But they do
> > have to follow it. So if someone gets it from Debian and then reverse
> > engineers it, Debian might no longer be able to distribute it (and
> > must destroy all copies).
> I assume you are referring to this part of the license agreement:
> "You will ensure that anyone who uses the Program does so only in
> compliance with the terms of this Agreement."
> If we force the user to read the license before installing the modules (as
> I suggested in a preinstall script, or in a script that is launched during
> bootstrap time, in case (s)he needs one of the modules during
> installation) and force him to agree to the license (by not installing
> the modules or not bootstrapping Debian if he answers "no"), wouldn't
> this mean that from this time on he would be the "You" who has to ensure
> that he will use the modules in compliance with the license and that
> Debian couldn't be made responsible any longer for any violations of the
I don't think that is how the license reads. There is nothing in it
about transferring the license (like section 6 of the GPL). It also
says, in particular, that we can not sublicense the code. You can
only make a deal with IBM, not with an intermediary.
> Actually that is what SuSE does in their S/390 distribution at install
Well, I think the naked truth here is that IBM is unlikely to ever go
after anyone who is just distributing the binaries. That is probably
why SuSE decided to cover themselves this way.
> > I'm not sure whether that injunction will then apply to all of the
> > people that we gave it to. If it does, then people would have to
> > check the debian website at regular interval to make sure that we
> > still had the ability to distribute. I think this would be too much
> > even for non-free.
> This would mean that IBM would potentially loose customers by
> forcing them to make an installed Linux distribution de facto unusable
> (as I wrote the modules are an essential part for Linux on S/390) so I
> don't think that that's what they intended with this license.
I didn't say that it was a good license. Unfortunately, I'm now a bit
more convinced that this is the way that the license has to be read.
The people receiving a copy from debian only do so under debian's
license. The license is revokable under violation of any of its
terms. This is one of the reasons that revokable license's suck.
> > There is some verbiage saying that no one is responsible for events
> > beyond their control. That might save this agreement, but I'm not
> > sure what "events beyond its control" means. People could certainly
> > argue that we knew, or should have known, that someone would reverse
> > engineer it (or violate export law, or make a copy without the
> > license, etc.).
> Yes we should have known, but if we force the user to agree to the license,
> then (s)he should have known as well, so wouldn't (s)he be the first to be
> made responsible?
No, because we are the ones who agreed to the license. We have to
make sure that no one that we give the code to violates the license.
That is why this license is so annoying. Normally, if we agree to the
license, we can distribute. If one of the people that we give it to
violates the license, it doesn't affect our ability to distribute. In
this case, it does.
> > > What would be the alternative, if the package could not go into
> > > non-free (i.e. not be part of the distribution at all)?
> > Make an installer, if that makes sense here.
> I assume by installer you mean a package that asks the user to download
> the modules from the IBM site to a certain place in the filesystem and
> then goes on with installation? This makes only sense if the user
> already has a network connection. But he won't have a connection without the
> modules (s)he is just about to install (at least in most if the cases).
> So at least at bootstrap time (where we need the modules on the ramdisk)
> this is not really an alternative...
I had a feeling that it might not work. Sorry.