[distcc] Re: Re: Finding peers: zeroconf/rendezvous?

[Bret Mogilefsky]

On Wed, 26 Feb 2003 15:33:09 +1100, Martin Pool wrote:
> Suppose person A wants to combine GPL'd and APSL'd code and distribute
> that work, and that person B downstream wants to modify and redistribute
> the work.  The GPL says that A *must not* impose additional conditions
> on B.  The APSL says B *must* publish their source.  Therefore they are
> incompatible; A cannot distribute this work.  I think A would be allowed
> to write this but not distribute it, but that's not very useful.

This pretty much addresses what I was trying to describe below:

>> I guess the question is, what happens when a business then takes distcc
>> and makes modifications to the zeroconf portions of it for their
>> private use, not realizing that that code is under the more restrictive
>> APSL (restrictive in that you're required to make your changes known)
>> and not the GPL?  In this case, distcc isn't violating the license, but
>> that business would be.

> Do you mean violating the APSL Rendezvous licence?

Yes.  As you've pointed out, though, distcc would be violating the GPL by
restricting the rights of the business granted in the GPL before the
business ever got around to violating the APSL. =)

> Unfortunately even though the intentions are good, it is still
> incompatible.  Distributing the patched code would (I think) violate the
> distcc licence.
> 
> Hopefully an updated APSL 1.3 will resolve these problems, or somebody
> will write a Rendezvous implementation under a free licence.  In the
> meantime using a separate program to do mDNS avoids the issue.

Yup, that seems to be the case.  I will cast about for a GPL'd mDNS
implementation in case one already exists.

Thanks for the thoughtful responses,

Bret