Subject: Re: Assistance/advice in choosing a license for POV-Ray 4.0
From: Chuck Swiger <>
Date: Tue, 15 Nov 2005 14:52:45 -0500

Rod Dixon, J.D., LL.M. wrote:
[ ...odd double-quoting reformatted... :-) ]
>> "If the vendor's code fetches the GPL'ed component without the end-users' 
>> intervention, and if the vendor's code will not run without it, then the 
>> vendor's code is probably a derivative work and the behavior you've 
>> described would be violating the GPL.
>> On the other hand, if the vendor's code works just fine by itself, but it 
>> provides an optional mechanism for the end-user to download additional 
>> components, then the vendor is not violating the GPL.  If the combination 
>> of the vendor code and the GPLed component does not fall within the terms of 
>> the GPL, then the end-user may not redistribute the combination."
> I agree that the quoted answer below makes a lot of intuitive sense and, 
> ultimately, the conclusion may prove to be correct, but that would be 
> fortuitous.  Whether the vendor's code works just fine is not the correct 
> legal analysis.

Agreed.  My understanding for US law is that the legal analysis for copyright 
infringement goes something like:

The Computer Associates v. Altai decision requires an 
"abstraction-filtration-comparison" process to identify copyrightable program 
code.  The "structure, sequence, and organization" of a computer program may 
also merit copyright protection, but are not the focus of the Altai test.

The abstraction step finds expressions of ideas.  The filtration step 
eliminates expressions that are the most reasonably efficient way of doing 
something, expressions that are dictated by external interfaces, and 
expressions that are in the public domain.  The remaining expressions or ideas 
are "nuggets" meriting copyright protection.


Someone who is a lawyer would follow the procedure above to attempt to prove-- 
or disprove, depending on which side they are on-- an allegation of copyright 
infringement.  Although I've given depositions as an expert witness based on 
following the procedure outlined above to compare codebases, I'm not a lawyer.

> To increase the likelihood that [an?] answer to this complex 
> question has a basis in law (and not pure speculation) I think we have to 
> refrain from highly abstract comments about the interaction between an 
> undefined modification of an open source codebase.  It may help if someone 
> prosed a specific (but very brief, please) example of an allegedy 
> permissible or infringing use.

I've provided two examples: using GNU readline with Python, and using a 
proprietary ATI or nVidia driver with a Linux kernel.

If an external component is required, then the combination of the two is 
required for the program to work and thus a derivative work containing the 
external component is clearly and unavoidably created; conversely, if an 
external component is optional and happens only at the behest of the end-user, 
the combination of the two may not happen at all.

If I, personally, was being asked to show that a given program was not a 
derivative work of some other software, I would start by showing that the 
program in question forms a complete, functional, and seperate work without 
using anything from the other software.

If the program works fine without using the external component, my non-lawyer 
perspective would be to claim that the program does not derive from the 
external component.  In the event that my opinion does not convince whoever it 
is and they file suit, well, that's when I would defer to someone qualified to 
hold legal opinions.