Thursday 22 November 2007

GPL3 Questions and Implications

It has been suggested by Shane Martin Coughlan following a gently heated discussion on the mailing list, that a collection of questions relating to the GPL3 be collated and put to the FSF authors.

* Shane Martin Coughlan wrote, On 21/11/07 20:14:

How about we make a list of concerns that people have and I email BrettSmith with them? That way we consolidate the concerns into onesubmission and potentially speed up the process of obtaining answers.MJ, Sam, perhaps you could coordinate the list of questions.

This page is setup for that purpose.

Readers: feel free to add comments or questions relating to GPL3 or AGPL, a compilation of which will be sent to Brett Smith by Shane.

My personal proposal is the creation of a list of implications of choosing a specific license to help prospective licensors choose (or avoid) a license according to their requirements.

My proposed implication is:

The GPL is widely considered a share-alike license where licensors have understood that the same terms will propagate throughout the distribution chain.

With the GPL3 this is not true: at some time in the distribution chain, derived works may have certain additional restrictions added, thus licensing the combined work under the AGPL such that when an original contributor receives the derived work with enhancements to his own work, he may not distribute any combination of his work with any of those enhancements unless he does so with the additional restrictions of the AGPL.
If the licensor finds this disparity objectionable then he may prefer to use the GPL2.

I believe that this implication is not widely understood and because ealier versions of the GPL are widely known to prohibit the addition of extra restrictions, this implication is also unexpected.


  1. I’m not opposed to the AGPL, I’m opposed to the additional restrictions that may be added to derivatives of my work by others that GPL3 permits.
    That means that I may release or enhance AGPL software, but if I release under GPL3 I don’t want to see it going round under the extra restrictions of the AGPL.

  2. Previous GPL licenses have been based on copyright, by granting conditional distribution rights, but section 13 of AGPL the seems not to be such a term, as it restricts use regardless of distribution activity.
    Is this based on copyright permissions required to “install” the software? Or does part 13 not affect non-distributing service providers?

  3. My big question is: Can someone modify a GPL3 web application and make their changes available only under the AGPL3?

  4. edA-qa mort-ora-y22 November 2007 at 19:17

    I intend on releasing code with the Affero GPL so the concern over section 13 of the normal GPL does not affect me too much. Though I can understand the concern that people have over it, particularily those that had licensed under “GPL v2 or later versions”. To some it may seem to be an “extra restriction”.
    I must say however, that the additions that the AGPL add over GPL are in my mind what was intended by the previous GPL versions to begin with. I see the GPL as a way to protect users, primarily against vendor lock-in. In a world of the web and extensive online programs, I just don’t see how the normal GPL can anymore guarantee the user freedom and control of their own data.

  5. To respond to edA-qa mort-ora-y (thanks for your comment):
    I think many people think that AGPL is what was intended by previous GPL versions.
    To many, I think, it extends the definition of user, not beyond the general meaning of the word user, but beyond instances of user that originally existed, and so what was (or would have been) originally intended is what is at dispute.
    Web users are users who usually do not possess the object code, and much of earlier GPL philosphy was based on defining user by those who possess the object code, so it is not surprising that some are surprised when a new new class of user appears and the definition of user changes to include these.
    To my own view, usually web users are not users of the software but users of a service provided by the software. To some this is splitting hairs, but that just defines the disagreement.

  6. My own dispute is not on the grant of source or freedom to the end user, but on the treatment of the author-user, who under GPL2 had expectation of being able to receive back changes under the original terms.
    GPL3->AGPL upgrade (if it exists) breaks that expectation.

  7. I’m trying to explain a fact, not scare-monger.
    Is GPL3/13 a backdoor? Authors who choose not to add “or later”, restricting their software to GPL3 only, can still have it upgraded to AGPL whose text is not yet fixed.
    The AGPL is effectively a later revision of the GPL3.

  8. Previously a service provider - such as a webmail service — could use a GPL webmail system, and modify it without having to release the modifications back to the community. The ISP/Service provide USES the software to provide a service. The user just uses the service.
    With AGPL they can’t.
    If I am providing a webmail service, all service improvements must be implemented as changes to the base webmail software. How can I provide better webmail if I can’t prevent my competitors from immediately copying what I have done?
    All of which gives web service providers a reason not to use the AGPL, and to shun any contributions which use it.

  9. Questions on the workability of the AGPL web-source requirements.
    The AGPL can be used as a technical lockout or DDOS, preventing small-scale providers with small network capacity from making effective use of AGPL software.
    How many times does the service provider have to provide source to each user?
    How much use counts as “use”?
    Can users be required to make accounts with valid email addresses first?
    What sort of network speeds does the source have to be made available at? What sort of slow-random-access-tape-based-backing-store must the source be fetched from?

  10. Reply to Ben:
    This is true, and although the AGPL may represent a change in expectation of service providers, I think the GPL was never designed to convenience the business of users who want to with-hold code; i.e. their inconvenience is incidental to the rights of users.
    (for a given definition of user).

  11. Is the AGPL’s clause on GPL3 ineffective without GPL3 clause 13?
    As far as I can see it can still be effective without GPL3 clause 13.
    If this is true, it sheds light on the possibility of once license restricting another.
    I admit that GPL3 effectively warns of this so I’m not suggesting unfairness on the part of the FSF.

  12. * MJ Ray wrote, On 25/11/07 17:23:
    > Carsten Agger wrote:
    >> […] e.g. I create a
    >> realtime online GCC compiler using a CMS under the Affero GPL (not that
    >> weird an idea: The compiler might create binaries for portable devices
    >> on which it’s not easy to install a compiler yourself), then the Affero
    >> GPL will apply to the CMS as well to the work as a whole - HOWEVER, the
    >> GCC compiler will still be pure GPLv3. […]
    > And what licence(s) can be used for any modifications to the compiler
    > done as part of the same project?
    > The work as a whole seems to be able to be under Affero GPL, so can
    > the modifcations to the previously-GPL part be under Affero GPL?
    > I think this may be another question for the collation.

  13. From - The term “user” is not clearly defined. If I get an “access denied”
    error page through a browser, am I a user of the web application? When
    I visit a portal, am I a user of the browser? Of the portal
    application, as well?

  14. From - > Ultimately I am still left without a meaningful definition of
    > interaction.
    I agree that it would be helpful to have further guidance from the FSF. It appears though that the key elements are (i) accepting user requests and sending them over a network, and (ii) this being in some way inherent to the application’s intended functionality

  15. Scratch that last one - just found which isn’t what I’d wish (looks like a lawyerbomb to me), but hey, it’s additional guidance.

  16. However, if the modifier of the GPL3 portion of an GPl3/AGPL combined work can’t release an AGPL patch to the GPL3 portion they may choose to not release a patch, but to merely continue to distribute the combined work as permitted by the AGPL. It seems like the AGPL then would prohibit the separation of the patch from the combined work (which may not be what the license authors intended)
    Can the original GPL3 author can extract this modification to the GPL3 portion as a GPL3 patch, (the patch not being his work and only distributed as part of the AGPL combination) and combine this patch with his own work and distribute it according to the GPL3?
    An example case: if a glue-class is written in php, subclassing a class from the GPL3 module, but instantiated by the AGPL3 module (thus also performing a link operation), and it be this subclass that has some implementation that “ought” to be incorporated into the GPL3 module, will the GPL3 author, receiving the source by virtue of being a “user” of the AGPL3 work, be able to combine parts of this subclass into his own work, without receiving additional permission from the subclass author?

  17. MJR around the web...18 April 2008 at 12:43

    […] Sam Liddicott » GPL3 Questions and Implications […]

  18. Sam I think Gnus should definitely be licensed. We don’t want any old Tom, Dick or Harry to think they can just go and buy one with no restrictions, who knows what that would lead to?!? Then there are grazing rights to consider and fines for gnus fouling the public highway. Look at the trouble Norway found itself in when restrictions were removed from reindeer! And what about the length of quarantine time for African imports, that should continue to be monitered.
    If restrictions are lifted all kinds of madness may ensue. Whilst being unable to turn their yard into the hanging gardens of Babylon, some Gnu users may be tempted to throw caution to the wind and try to recreate Africa on their lawns by having herds of Wildebeasts (Gnus) galloping majestically across the Serengeti (the rockery). It’s not British!