gnu-misc-discuss
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: GPL 2(b) HUH?


From: Tim Smith
Subject: Re: GPL 2(b) HUH?
Date: Thu, 18 Sep 2008 01:14:20 -0700
User-agent: MT-NewsWatcher/3.5.3b2 (Intel Mac OS X)

In article <address@hidden>,
 Ciaran O'Riordan <address@hidden> wrote:
> But if you looked at Linux, decided the scheduler was crap, and then wrote a
> completely new scheduler for Linux, then that would be a derivative work,
> AFAICT.  Whether you link or compile it all together, or whether you output
> one file or multiple, wouldn't be a deciding factor, AFAIK.

OK, but how about this.  Suppose *I* look at the schedular, decide that 
it is crap, and so I modify Linux so that the scheduler can be loaded at 
run time.  To make a scheduler using my system, you write a function in 
C named my_scheduler, with an interface that I define.  You compile that 
to an object file using "gcc -o my_scheduler.c".  To load the scheduler, 
you run a new command I provide, "load_scheduler", giving it the .o file 
containing your scheduler code.

I have designed my interface to be not Linux specific.  It represents a 
general abstraction of scheduling, and maps Linux scheduling parameters 
and settings to more abstract things.

HYPOTHETICAL #1.  I also write my new scheduler, and distribute it, 
along with my patches that modify the kernel to add the schedule loader.

Is my scheduler a derivative work of the kernel?

HYPOTHETICAL #2.  I do not write a scheduler.  I limit my contribution 
to modifying the kernel to work with schedulers following my 
specification.  You write a scheduler, following my specification.

Is your scheduler a derivative work of Linux?

HYPOTHETICAL #3.  Someone over at Microsoft sees my work, and decides 
that loadable, generic schedulers are cool.  Microsoft puts this feature 
in the next version of Windows.  They do not take any code from my 
implementation.  All they do is use the same interface as mine.  They do 
such a good job that binary schedular files for Linux work out of the 
box with Windows.

Does this change the answer to #1 or #2?  Does it matter whether or not 
your scheduler or mine was written after Microsoft makes the 
aforementioned version of Windows available?

-- 
--Tim Smith


reply via email to

[Prev in Thread] Current Thread [Next in Thread]