Google Calling: Inside Android, the gPhone SDK (O'ReillyNet) [LWN.net]

Google Calling: Inside Android, the gPhone SDK (O'ReillyNet)

Posted Nov 13, 2007 18:01 UTC (Tue) by MattPerry (guest, #46341)
In reply to: Google Calling: Inside Android, the gPhone SDK (O'ReillyNet) by xav
Parent article: Google Calling: Inside Android, the gPhone SDK (O'ReillyNet)

> Now you can argue however you like

Indeed I can, but in this case my argument is backed up by the fact that companies do
successfully distribute non-GPL kernel modules.

(Log in to post comments)

Google Calling: Inside Android, the gPhone SDK (O'ReillyNet)

Posted Nov 13, 2007 22:38 UTC (Tue) by rfunk (subscriber, #4054) [Link]

Define "successfully". Just because kernel people have not stopped a company from 
distributing proprietary kernel modules doesn't mean they won't.  You're basically 
arguing, "It must be OK because they haven't been caught."

Anyway, it's important to understand the difference between distributing your own 
proprietary code and distributing Linux and your own proprietary code.  When you're not 
distributing someone else's code then nobody else can have a copyright case against 
you.  When you're distributing someone else's code then their license kicks in, and if you 
read the GPL you'll find it has a lot to say about how you can distribute GPL'd code and 
non-GPL code that is derived from it.

Of course, the definition of "derived" in this context (and whether any given kernel 
module is derived from Linux) is the key controversy, but I suggest you look toward 
sources related to copyright law rather than toward whether the kernel provides an API.  
Your API talk just demonstrates that you're new here and don't begin to understand the 
real issues.

Google Calling: Inside Android, the gPhone SDK (O'ReillyNet)

Posted Nov 14, 2007 1:35 UTC (Wed) by MattPerry (guest, #46341) [Link]

> Define "successfully". Just because kernel people have not stopped a
> company from distributing proprietary kernel modules doesn't mean they
> won't.  You're basically arguing, "It must be OK because they haven't
> been caught."

I define successfully as distributing the module for a minimum length of time (say, six to
twelve months) without legal challenges being brought against you.  That's my personal
definition.

> if you read the GPL you'll find it has a lot to say about how you can
> distribute GPL'd code and non-GPL code that is derived from it.

Then you get into the quagmire of defining derived.  Even Linus Torvalds stated that,
"including one header file in order to compile against something does not automatically make
something a 'derived work'."[1]  There is also the fact that the comments in
include/linux/module.h specifically mention the ability to define the license used for the
module, including using "Proprietary" as an example of a non-free module license identifier.
See my other post at http://lwn.net/Articles/258419/ which has a copy of the comments from
module.h.

These public statements combined with the comments in the module.h header file show explicit
intent to allow non-free modules to be used with the kernel.

> I suggest you look toward sources related to copyright law rather than
> toward whether the kernel provides an API.  

If you know of any case law that I can read that addresses this type of issue I would be happy
to read it.

> Your API talk just demonstrates that you're new here and don't begin to
> understand the real issues.

I will admit that I misused the term API; however, I have a strong understanding of the real
issues even if I'm not a kernel programmer.  I have also cited primary sources to support my
statements.

1. http://lkml.org/lkml/2006/12/14/218