Mechanics of Contributing with a New License

[Chris Zumbrunn]

On Jul 24, 2005, at 11:44 AM, David Barrett wrote:

> Chris Zumbrunn wrote:
>> On Jul 23, 2005, at 9:22 PM, David Barrett wrote:
>>>
>>> Actually, you bring up a good point.  Given that each file is 
>>> typically licensed as a whole, but modifications are done in part, 
>>> if you submit changes under a new license, how is this represented?
>> A contributor cannot relicense code covered under the OVPL. This is 
>> the case with any license. Only the copyright holder can relicense 
>> code. The idea is that when the ID receives a modification from a 
>> contributor, it receives this modification under a BSD style license, 
>> which allows for reuse inside the OVPL licensed project.
>
> Well, yes I get this.  But my questions is at what granularity do 
> licenses apply?

At any granularity.

> So far as I can tell, virtually all code is licensed on a per-file 
> basis.  Perhaps there are exceptions, but I've never encountered them.
>
> Based on this observation, I see a reasonable argument that making any 
> modification to a file constitutes modifying the whole file, and thus 
> justifies -- indeed requires (according to the OVPL/BSD proposal) 
> releasing the entire file under the BSD license.

No. If the contributor releases the entire file, the original code in 
that file is OVPL and only the modifications are BSD.

> Furthermore, were this simply done for every file in the entire 
> project, then this would justfy releasing the entire project under the 
> BSD license, because the entire thing has has been "modified".  And 
> once released under BSD, I (and anyone else who wants one) now have a 
> proprietary copy of the entire project, at no cost, effectively 
> nullifying any potential commercial value the ID might have been able 
> to create.

Again. Nobody other than the copyright holder can relicense. The 
original code stays OVPL even if it is modified. This changes only if 
the code is modified to the extent that it no longer resembles a 
derivative work of the original, hence the original contributor has no 
copyright on it anymore and the "contributor" becomes the ID. This is 
the case no matter what license the original work was under or even if 
it was not licensed at all.

> Personally, I wouldn't be eager to defend myself against the argument:
>
> Person A: "Your honor, the OVPL mandates I release contributions under 
> the BSD.  In order to comply with this requirement, I followed the 
> industry standard (see exhibits A through ZZZZ) and recorded the new 
> license at the top of the modified file, and posted it on my website."
>
> Person B: "Your honor, I downloaded a copy of a BSD-licensed project 
> from some website and made changes, and now this man says something 
> about an 'OVPL' license?  Look for yourself, there is no mention of 
> this 'OVPL' anywhere in the entire codebase.  Because this project is 
> plainly licensed under BSD (just check the tops of the files), and 
> because the BSD does not require me to reveal my changes to anyone, I 
> am clearly under no obligation to share future modifications with this 
> man."

Again. Nobody other than the copyright holder can relicense.

> Whether or not the court would uphold this decision is impossible to 
> tell, but I wouldn't want to be the guinea pig to find out.
>
> But to bring me back to the topic I really want to discuss, I still 
> haven't heard why making 3.3 opt-out on a per-file basis doesn't 
> neatly solve this whole dillema.

Because making 3.3 opt-out is not an option for Alex. It might work for 
you, but it does not work for the stewards of the OVPL license.

Chris