For Approval: GPLv3

[Chris Travers]

Hi Donovan;

I am not sure this holds.  IANAL though.


> So whatever license that the downstream user receives through my
> conveyance, it will originate with you. But what license is that? We
> already had a lengthy discussion about BSDL, the gist of which is that
> licensing under license A is equivalent to also licensing under license B
> if the rights granted by B are a subset of the rights granted by A. Since
> the rights granted by bare GPL v3 is a subset of the rights granted by GPL
> v3 + your Additional Permissions, you have already released the program
> under bare GPL v3 as well. GPL v3 section 7 explicitly allows me to select
> this license when conveying downstream.



But if that is the case, you have created a new set of permissions for that
code.  In essence, there is no longer an offer for a specific license
between *me* and the downstream recipient.   If you are doing this on my
behalf, this would seem to be sublicensing and the rights would originate
from you.


BSDL allows you to sublicense any subset of the rights of BSDL downstream.


And certain BSDL varients are explicit on this point.  Hence they allow one
to enter into copyright agreements with others who may not have received the
code and offer copyright licenses  under conditions of the BSDL (plus
possible other restrictions, since those restrictions have to remain).  The
GPL3 Plus additional permissions does not do this.

GPL v3 allows you to EFFECTIVELY sublicense any subset which contains the
> entire bare GPL v3. It achieves this not through sublicensing but through
>
removal of Additional Permissions, which is not really sublicensing if you
> define that both licenses are still "GPL v3".



Isn't this false advertising though?  I mean, if I grant downstream users
rights to use the software a certain way, at most someone can hide my
additional permissions.  Hence this is advertising that the code is subject
to copyright restrictions which it is not subject to.  They can't enforce
those changes because they aren't a party to the license.  They are just
advertising restrictions on the code that nobody can enforce.  This seems
dangerous to me but IANAL.

The selection of a subset
> did not entail using a different license, in the sense that it remained
> the GPL v3 license.



Yes it does.  I license is a permissive grant.  CHanging what is granted
means a different license.  Failure to change the name does not mitigate the
fact that the license is different.


That distinction makes a little more sense if you consider the lay
> description of the GPL: downstream releases of GPL software must remain
> under the GPL.



But past versions of the GPL did not mandate source licenses for other
components, which is what makes this tricky.  The presumption was that the
corresponding source would be one single work, but this need not be the
case.  Hence different licenses could apply provided that the code was made
available.  Furthermore prior versions of the GPL explicitly disclaimed an
attempt to claim control over the source code of others.


In the face of Additional Permissions and various
> disclaimers and Additional Legal Notices, one has to wonder how many
> "GPLs" there are in the world. It is much easier to understand the GPL if
> you think of there being only one GPL v3 with various "extras" tacked on,
> rather than N different licenses.



Of course the disclaimers, legal notices, etc can all be tacked on in
various ways, and they are harder to remove.

Making this distinction also allows them to be absolutely clear that
> ARBITRARY sublicensing is not permitted (since it would destroy the
> copyleft aspects of GPL). Allowing anything called "sublicensing" might
> leave the door open just a hair for a judge to interpret it overly
> broadly.



The problem is that sublicense vs license seems to imply specific
relationships between the parties.  In this case you argue for an effect of
a sublicense (which would indicate that the person removed the additional
permissions could then enforce such a contract), but wihtout the legal
relationship, which makes no sense.  IANAL, but this seems very strange.  It
might be interesting to see more lawyers chime in here.  This seems very
strongly suspect in my opinion.


Disallowing all sublicensing (and then effectively letting it
> back in through very limited means) could make it harder for a judge to
> make that mistake.



 It still leaves the following questions:
1)  Who were the parties in the license (or sublicense)?
2)  What rights were granted in that license or sublicense?  Answer depends
entirely on #1 above.
3)  Who distributed and advertised the terms of that license?
4)  Were the advertised terms accurate?  Answer depends entirely on #1
above.


Would you want to put those questions before a judge?

Best Wishes,
Chris Travers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensource.org/pipermail/license-discuss_lists.opensource.org/attachments/20070830/43760e25/attachment.html>