Re: The SFLC dismissals should be coming soon

[Hyman Rosen]

On 2/16/2010 12:06 PM, RJack wrote:
> Obviously, If you author a derivative work you *must* be able to
> identify the owners of the "one or more preexisting works *that you
> modified*.

No, that's not true, and does not follow from anything
you quoted. To create a derivative work you must have
permission from the rights holders. In the case of GPLed
code, that permission is expressed through the license,
and no specific knowledge of the identity of the owners
is required.

> If you modify say "basename.c (* Copyright (C) 1999-2004 by
> Erik Andersen)" you don't create a derivative work from the code owned
> by the fifty or more distinct authors of some of the other 680 source
> code modules -- you didn't even touch their work.So you didn't create a
> derivative work of "BusyBox" -- you crated a derivative work of
> "basename.c" owned by Erik Andersen.

Correct. You create a derivative work of all the code
that is in basename.c, which itself might have many
distinct authors.

> You can't simultaneously claim
> BusyBox is a single computer program for purposes of defining a
> derivative work *and* that BusyBox is a compilationspaghetti of code
> representing each of fifty disparate authors' "own interests".

Of course you can. When you build a new BusyBox executable
using the module you have modified, and possibly even by
simply checking the new module into the BusyBox source code
repository, you create a derivative work of the collective
work which is the creative arrangement of modules which form
BusyBox.

> A copyright owner cannot file an infringement claim over another's code
> and he can't grant permission to form a derivative work for code he
> doesn't own.

True, but irrelevant. Every contributor to BusyBox owns
copyright on the code he contributed, and on the entire
collective work from his version onward.

> The binary ownership nodes of *individually owned* derivative works
> comprising the 680 C modules and their patches probably runs into
> *millions* of nodes. BusyBox is actually a compilation of thousands of
> tiny derivatives works "spaghetti code". Using combinatorial statistics
> how many patches and modifications of original source code modules have
> occurred in the last ten years of BusyBox development?

Why does that matter?