Questions to OSI Board quorum

[David Ryan]

David Barrett wrote:

> Chris Zumbrunn wrote:
>
>> On Nov 14, 2005, at 8:56 PM, Russell Nelson wrote:
>>
>>>> Question 2) Would a license which requires all contributions to be
>>>> licensed uner a BSD style license still be deemed descriminatory?
>>>
>>>
>>> Probably not, because anybody could then do what they wanted with them.
>>> The Apache license does something like this, but it's not a reciprocal
>>> license, requiring all modifications to be contributed.
>>
>>
>> This would not eliminate the asymmetry between the initial contribution
>> and any subsequent contributions. OSD#3 says that the contributors
>> must have the right to distribute their modifications under the same
>> terms as the initial contributor. An open source license can not require
>> subsequent contributions to be granted under terms that the original
>> contribution was not granted under.
>
>
> What about an 'opt-out' license -- contributors have the option of 
> releasing code under the original license (with grant-back in place), 
> or a modified license with no grant-back.
>
> Thus the only "discrimination" in this license would be that it's 
> disproportionally easy (but in no way required) to grant-back to the 
> initial developer.
>
> I favor this approach over the BSD approach, not only because it feels 
> "cleaner" (having one license wholesale switch to another feels 
> strange), but also because it helps the initial developer maintain 
> license consistency in the main branch.  With the BSD approach, the 
> main branch will gradually get taken over with BSD code, creating an 
> auditing nightmare in the event there's question as to whether a block 
> of code is BSD or OVPL.  With the opt-out approach, the initial 
> developer rejects everything without the grant-back in place, and 
> therefore there's no question that the main branch is pure OVPL.

I agree with Chris on this.  I was surprised that Russell actually said 
it would be ok.  The main reason I brought it up was to make sure I 
fully understood where the OSI stood on these issues.  However, given 
Russell's response, I'm completely confused.

I also agree with you, that the danger of the BSD approach is that it 
will be slowly taken over by BSD code.  Even if the OVPL was accepted, 
there is still value in asking contributors that commit to the main 
branch of code that they sign a joint copyright assignment.  This 
ensures the cleanest source code tree; free from copyright or licensing 
issues. 

The OVPL protects the initial developer from incompatible forks from 
happenning in the wild.  I wouldn't rely on the OVPL by itself for 
project's licensing and copyright.  Having the license back to initial 
developer would allow portions of code to be brought back from other 
releases to ensure compatibility.  I don't think it would be wise to be 
taking large portions of code as it erodes the copyright in the project.

For these reasons I don't think the opt-out versus opt-in provides much 
value.  If a developer wishes to opt-in you are much better off with a 
joint copyright assignment that the developer signs.  In regards to the 
opt-out, I don't think it provides enough protection to be worth the 
additional work in drafting a license.

Of course, these are all issues which create descriminatory business 
models and ensures an unlevel playing field when taking software 
products to market.  Just aswell companies can be safe in the knowledge 
that the OSI won't be making any public statements making that clear 
anytime soon. ;)

I'm still pondering how to proceed with my software.  Given the OVPL is 
now *not* compliant with the OSD, I have a lot more freedom in making 
other incompatible changes.  I'm starting to look around at other 
licenses to see if there is any value in them.  I've found  
http://www.collaborativesource.org/  and http://www.softdevelcoop.org/   
If you know of any other active groups providing alternatives to OSD 
licenses please let me know.  The licenses around Java seem to be worth 
a look for my purposes.  Java Research License and a few others.  I'm 
not sure if Sun would allow other companies to use modified versions of 
their licenses though.

Regards,
David.

--
David Ryan. aka Oobles.
http://www.livemedia.com.au/Blog