Re: How long is it acceptable to leave *undistributable* files in the kernel package?
Raul Miller writes:
> Ok, this is good -- I did not know that.
> However -- by this definition, the linux kernel is very definitely a
> derivative work, and the firmware is content which has been incorporated
> into the kernel.
> According to what you just cited, the concept of a collective work
> doesn't enter into the picture at all.
The kernel (I assume as a whole) is a derivative work of what?
I would argue that the kernel is a compilation of what executes on the
host CPU with other parts (boot logos, fonts, and firmware data). The
executable part may be a derivative in part of Adam Richter's code,
but that does not necessarily make the kernel as a whole a derivative
of his work.
The GPL is clear how derivative works must be treated, but it is not
clear how it applies to a compilation (given, among other things, what
it says about "mere aggregation" works).
The only court case I know of that dealt with the issue of "derivative
work" with respect to the GPL was MySQL AB vs Progress Software
Corporation, which settled. While the FSF claims a broad definition,
Lawrence Rosen (general council for OSI and author of the Open
Software License) disagrees at http://www.rosenlaw.com/html/GL18.pdf.
Even if the static or dynamic library question is resolved in
agreement with the FSF, embedding of firmware may not: I would guess
the question of derivative work (in terms of the kernel's host
executable code) would turn on the nature or detail of the interfaces,
rather than the distribution method of the firmware.