I'm working on a project that's currently gpl2+ licensed [1] and we want to include some code from a gpl3 project [2]. This code will be an isolated utility used to generate documentation from data contained in other source files.
[1] https://bitbucket.org/fedoraqa/libtaskotron [2] https://github.com/ansible/ansible
I know that if we went forward with this, the project would need to be distributed as gpl3 but I have some questions around the specifics:
Would all the source in our project need to be re-licensed as gpl3 or is it sufficient to have the project license as gpl3 and the existing source files as gpl2+?
Assuming that it is possible to keep the existing gpl2+ source as gpl2+, would it be possible to change the project license back to gpl2+ in the future if we were to remove any gpl3 code?
Thanks,
Tim
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 05/28/2014 12:22 PM, Tim Flink wrote:
I'm working on a project that's currently gpl2+ licensed [1] and we want to include some code from a gpl3 project [2]. This code will be an isolated utility used to generate documentation from data contained in other source files.
[1] https://bitbucket.org/fedoraqa/libtaskotron [2] https://github.com/ansible/ansible
I know that if we went forward with this, the project would need to be distributed as gpl3 but I have some questions around the specifics:
Would all the source in our project need to be re-licensed as gpl3 or is it sufficient to have the project license as gpl3 and the existing source files as gpl2+?
Assuming that it is possible to keep the existing gpl2+ source as gpl2+, would it be possible to change the project license back to gpl2+ in the future if we were to remove any gpl3 code?
== DISCLAIMER == IANAL, this is not legal advice. == /DISCLAIMER ==
You would not need to re-license existing source from GPLv2+. The effective license on the compiled binary works would be GPLv3 in the scenario you describe. If you removed the GPLv3 code (or the code was relicensed to GPLv2+), then the resulting binary license would be GPLv2+. By adding the GPLv3 code, you're just forcing the "+" to trigger to GPLv3 in the combined binary work. The individual source files are still a mix.
~tom
== ¸.·´¯`·.´¯`·.¸¸.·´¯`·.¸><(((º> OSAS @ Red Hat University Outreach || Fedora Special Projects || Fedora Legal
On Wed, May 28, 2014 at 01:15:06PM -0400, Tom Callaway wrote:
On 05/28/2014 12:22 PM, Tim Flink wrote:
I'm working on a project that's currently gpl2+ licensed [1] and we want to include some code from a gpl3 project [2]. This code will be an isolated utility used to generate documentation from data contained in other source files.
[1] https://bitbucket.org/fedoraqa/libtaskotron [2] https://github.com/ansible/ansible
I know that if we went forward with this, the project would need to be distributed as gpl3 but I have some questions around the specifics:
Would all the source in our project need to be re-licensed as gpl3 or is it sufficient to have the project license as gpl3 and the existing source files as gpl2+?
Assuming that it is possible to keep the existing gpl2+ source as gpl2+, would it be possible to change the project license back to gpl2+ in the future if we were to remove any gpl3 code?
== DISCLAIMER == IANAL, this is not legal advice. == /DISCLAIMER ==
You would not need to re-license existing source from GPLv2+. The effective license on the compiled binary works would be GPLv3 in the scenario you describe. If you removed the GPLv3 code (or the code was relicensed to GPLv2+), then the resulting binary license would be GPLv2+. By adding the GPLv3 code, you're just forcing the "+" to trigger to GPLv3 in the combined binary work. The individual source files are still a mix.
Disclaimer: IAAL but IANYL other than solely to the extent that IAARHL.
I agree with what spot says. I suppose one might argue that a change to a file from libtaskotron in this scenario might be properly considered GPLv3+ only; I would myself consider that absurd if the later-in-time developer retained the original legal notice for the libaskotron file (all the more so if the developer in question is the same as the earlier-in-time developer).
When GPLv3 was released in 2007 the legal assertion we are making here was probably somewhat in question, but it soon came to be established practice to mix individually-licensed GPLv2+ and GPLv3 source files in a single work. It is really not logically different from the situation of including GPL-compatible but noncopyleft-licensed files in a larger GPL-licensed work.
- RF
IANAL, but if the added GPL3 code is, as Richard said, an "isolated utility" that is not linked to any of the GPL2+ code (an assumption about the nature of an "isolated utility"), then isn't this "mere aggregation"? If so, wouldn't it have no effect on the licensing of the GPL2+ code, and the resulting package simply contains a mix of GPL2+ and GPL3 components?
I'm not arguing that there's anything wrong with relicensing the GPL2+ source to GPL3, only that I don't think it automatically occurs in the described scenario. If it was desired to relicense the GPL2+ source to GPL3, I think that should be done explicitly, by actually changing the license notices.
And it's not what Richard was asking, but I think if the isolated utility generated documentation based on the GPL2+ sources, the resulting documentation would be GPL2+, unless the isolated utility included some of its own GPL3 components in that output.
Eric
On Thu, May 29, 2014 at 02:47:31AM -0600, Eric Smith wrote:
IANAL, but if the added GPL3 code is, as Richard said, an "isolated utility" that is not linked to any of the GPL2+ code (an assumption about the nature of an "isolated utility"), then isn't this "mere aggregation"?
I didn't say anything about an isolated utility, but now I see that Tim did. So my original assumption was that there was something more than 'mere' aggregation, but if that's not true then the answer and analysis are different (and easier). Or rather the end result is objectively the same, but the way you look at it might be different.
I'm not arguing that there's anything wrong with relicensing the GPL2+ source to GPL3, only that I don't think it automatically occurs in the described scenario. If it was desired to relicense the GPL2+ source to GPL3, I think that should be done explicitly, by actually changing the license notices.
I basically agree; the idea of implicit relicensing is something that was invented to reconcile what one group of people say about license compatibility and GPL interpretation with what another group of people (which probably overlaps somewhat, maybe even significantly, with the first group) actually do.
- RF
On 05/29/2014 09:18 AM, Richard Fontana wrote:
On Thu, May 29, 2014 at 02:47:31AM -0600, Eric Smith wrote:
IANAL, but if the added GPL3 code is, as Richard said, an "isolated utility" that is not linked to any of the GPL2+ code (an assumption about the nature of an "isolated utility"), then isn't this "mere aggregation"?
I didn't say anything about an isolated utility, but now I see that Tim did. So my original assumption was that there was something more than 'mere' aggregation, but if that's not true then the answer and analysis are different (and easier). Or rather the end result is objectively the same, but the way you look at it might be different.
I concur. If the new code is an isolated utility, then that specific binary would clearly be GPLv3, while the rest of the codebase would remain GPLv2+. In my non-lawyery, non-legal advisey, opinion, of course.
~tom
== ¸.·´¯`·.´¯`·.¸¸.·´¯`·.¸><(((º> OSAS @ Red Hat University Outreach || Fedora Special Projects || Fedora Legal
On Thu, 29 May 2014 09:18:43 -0400 Richard Fontana rfontana@redhat.com wrote:
On Thu, May 29, 2014 at 02:47:31AM -0600, Eric Smith wrote:
IANAL, but if the added GPL3 code is, as Richard said, an "isolated utility" that is not linked to any of the GPL2+ code (an assumption about the nature of an "isolated utility"), then isn't this "mere aggregation"?
I didn't say anything about an isolated utility, but now I see that Tim did. So my original assumption was that there was something more than 'mere' aggregation, but if that's not true then the answer and analysis are different (and easier). Or rather the end result is objectively the same, but the way you look at it might be different.
I'm not sure if this is an important distinction but I figure that it probably wouldn't hurt to make sure:
The utility itself will be isolated in a directory of the main project. However, that utility will read several other gpl2+ licensed source files in order to generate our documentation.
I'm not arguing that there's anything wrong with relicensing the GPL2+ source to GPL3, only that I don't think it automatically occurs in the described scenario. If it was desired to relicense the GPL2+ source to GPL3, I think that should be done explicitly, by actually changing the license notices.
I basically agree; the idea of implicit relicensing is something that was invented to reconcile what one group of people say about license compatibility and GPL interpretation with what another group of people (which probably overlaps somewhat, maybe even significantly, with the first group) actually do.
Thank you all for your input on this. I really appreciate it.
Tim
-
Eric Smith -
Richard Fontana -
Tim Flink -
Tom Callaway