[Incubator] Transfering Code Copyright - restart

[rich at richsteele.org]

Quoting Frank Warmerdam <warmerdam at pobox.com>:

> My understanding is that neither licensing (via a contributor agreement
> nor by assignment) is strictly necessary as long as the code is all
> properly under an OSI license.  That license already gives the
> foundation the right to redistribute the code as long as it does not
> alter the copyright or license conditions in place.
>
> Furthermore that the contributor agreement was primarily seen as being
> useful in that it would grant the foundation to alter licenses for code,
> and perhaps to give it some sort of standing in code to protect the
> project if needed.

This is mostly accurate.  The "pitch" for and against CLAs was made on  
the maillists early on in a debate between ESR and me.  Archivists can  
probably locate it quicker than I can dig up a link.  The wiki FAQ  
also sets out much of the argument for CLAs:
http://wiki.osgeo.org/index.php/Contributor_Agreement

> Earlier in the year it was decided to not pursue contributor agreements
> because they give a rather dubious benefit and require a fair amount
> of paperwork to chase people down to sign the agreements.  Some projects
> also felt uncomfortable giving the foundation the power to alter the
> license on a project - especially I think a concern that GPLed projects
> might be changed to a weaker license against the wishes of the original
> contributors.
>
> But while we decided we wouldn't require contributor agreements across
> the board, it was still considered ok for specific projects to require
> them.  Rich felt that MapGuide Open Source would still likely require them
> from it's contributors.  So at that point we are already lacking
> consistency. :-)

This is an accurate history.  I believed at the time, and still do,  
that obtaining CLAs is necessary for a well-administered open source  
project.  Others disagreed.  With my (then) Autodesk hat on, I stated  
that Autodesk would only be willing to put MG OS into the foundation  
if committers on the project signed CLAs to ensure the pedigree of the  
code going forward.  Out of this, we decided that the CLA requirement  
would be project specific.  So the inconsistency we now have on the IP  
policy was the price to be paid for keeping OSGeo together (such  
inconsistencies were foreseen in Chicago, so IMO should not be viewed  
negatively).  The decision of whether CLAs will still be a requirement  
for MG OS to be formally contributed to OSGeo is probably best left to  
Jennifer, Gary and other Autodeskers at this point, but I continue to  
believe it is the prudent thing to do.  By the way, Linus, the Apache  
Foundation and most others also think the benefit of CLAs is more than  
dubious and worth the administrative work to obtain.

> We also were not intending to require copyright assignment, but to
> allow it where desired.  For projects like Geotools, GRASS and MapBuilder
> (I think) the code has been setup with a copyright message attributing
> the copyright to non-legal entities.  Ie. GeoTools Development team,
> GRASS development team, etc.  In this case assignment was seen as a
> way of cleaning up the situation and assigning the code to an actual
> legal entity.

As Jennifer and I have stated, this type of attempted copyright  
assignment doesn't work.  The copyright remains with the original  
creator of the work.  To properly effect an assignment, you'll need  
the copyright holder to sign a written assignment document.

> Alternatively, in theory, we could try and track through fix the headers
> to name the specific contributors as copyright holders on a module by
> module basis.  But this seems contrary to the original intent of placing
> the code into the hands of the "project", and also very difficult to do
> accurately after the fact.

It may seem contrary to the intent of placing code into the hands of  
the project, but this is precisely what would need to be done in order  
achieve this intent.  Sometimes the law doesn't allow easy shortcuts  
that would make sense to most people.

> For those file copyrighted by me, I was also thinking of assigning them
> to the foundation as a statement of faith, without requiring that it be
> done for all GDAL files.  But given the complexity, I'm not sure I want to
> pursue that for the time being.

It isn't complex at all if that is what you want to do.  A form of  
copyright assignment is attached.  Sign away!

> All that said the next step for GeoTools seems to be for them to take
> a copyright assignment letter to as many contributors as they can
> identify.  Can you provide some sort of template copyright assignment
> letter we should use?  Since it is hard to identify specific files I
> am hoping that each contributor could sign a letter indicating that
> copyright for all their contributions that are in the GeoTools SVN
> repository as of a particular date (perhaps also with an url
> identifying the SVN repository in question) are to be assigned to the
> foundation.  Is that a reasonable approach?

The only question I have is why one would assign copyright of  
contributions only as of a particular date?  What about contributions  
made the very next day?  Are those simply licensed to OSGeo?  Under  
what license?  The attached copyright assignment would assign past and  
future contributions to a particular project, if the contributor so  
desired to assign the copyright.

> Once they have such a template letter, they can get people to start
> printing them, signing them, and mailing them to OSGeo for storage
> somewhere.

Ah yes, "mailing them to OSGeo".  Do we have a PO box and an admin yet? :)

-------------- next part --------------
A non-text attachment was scrubbed...
Name: OSGeo_Assignment.pdf
Type: application/pdf
Size: 85091 bytes
Desc: not available
Url : http://lists.osgeo.org/pipermail/incubator/attachments/20060907/4e50e06c/OSGeo_Assignment.pdf