Combining GPL and non-GPL code

Chris Travers chris at
Sun Aug 19 06:50:44 UTC 2007

Rick Moen wrote:
> Quoting John Cowan (cowan at
>> Quite so.  But the copyright on a derivative work (as opposed to a mere
>> collective work) belongs to the deriver, provided the derivation was
>> lawfully made, which means obeying any licensing requirements attached
>> to the original.  So this is not "relicensing" or "sublicensing", it's
>> applying the license permissions to make a derivative work.
> Sure, I'll go along with that.  The work entailed in creating the
> derivative (the set of incremental changes), to the extent it gives rise
> to copyright title, is required to be available under GPL terms in order
> to qualify for creation and distribution of that derivative, rights
> otherwise reserved.
Agreed this far.
> If anyone wants access to the BSD-licensed code in your hypothetical
> under its owner's terms, that access does remain available -- anywhere
> such separate code can be found.
There is an exception that is worth noting.  IANAL though.

If the GPL work author *is also* the author of the code or has had 
copyright asigned to him/her, it may be possible to release the code in 
terms other than the BSD terms.  But then the issue here is not 
derivation or work as a whole but outright ownership of the relevant 
copyrights.  So while path *may* matter, it only matters where a person 
has the legal right to further restrict distribution of the relevant 
source.  The BSD license does not automatically grant this.
> Note:  It might be argued that the changes required to Exim to hack in
> OpenSSL calls raise derivative work questions, as opposed to merely
> creating a collection.
I don't now.  I think it would depend on the nature of the changes.  Are 
the changes derivative of the Exim code?  Or are they separate?  If they 
are derivative, then who owns the copyrights to the sections of code in 
Exim and the changes in OpenSSL?  If they are both owned by the same 
person, I wouldn't see a problem.  This is likely to depend on an 
analysis of the actual nature of the changes and the interaction of the 
related sections of code.

Now you see the problem with this sort of discussion.
>   When my friend Marc Merlin wrote the bridge code 
> patches in around 2000, arguably they used significant copyrightable
> elements of OpenSSL, as GNU TLS wasn't yet available.  (At that time,
> Phil Hazel had Exim under pure GPLv2:  Marc released the patched Exim
> tarball, I pointed out the licensing problem, Marc swore a blue streak,
> and an e-mail to Phil resulted in him amiably issuing the needed licence
> exception.)
Cool.  Glad it worked out. :-)

Best Wishes,
Chris Travers

-------------- next part --------------
A non-text attachment was scrubbed...
Name: chris.vcf
Type: text/x-vcard
Size: 171 bytes
Desc: not available
URL: <>

More information about the License-discuss mailing list