Subject: [Fwd: Re: I'm leaning toward going with gplv3 but...]
From: Aaron Aichlmayr <waterfoul@gmail.com>
Date: Fri, 07 Aug 2009 09:58:35 -0600
Fri, 07 Aug 2009 09:58:35 -0600
Thanks for all the info! after reading through it I'm probbly going to 
do what you guys suggested and since Its php (which is an interpreted 
language) I don't think the compiling thing is an issue but would being 
interpreted instead of compiled would that change things? I also have a 
few other questions. First, if I do this licence buffer approach will 
this make it so no libraries can be released gpl but would allow lgpl? 
Also could I do this with the agpl since it will be displayed on the 
web? (by the way thanks for clarifying what the other clause meant, the 
way everyone else worded it kind of confused me)

Ben Tilly wrote:
> 2009/8/7 Dag-Erling Smørgrav <des@des.no>:
>   
>> Ben Tilly <btilly@gmail.com> writes:
>>     
>>> It should be noted that while that is the legal opinion of the FSF,
>>> there is no guarantee that a court will rule that way.  For an
>>> interesting alternative, the opinions of Linus on binary kernel
>>> modules is interesting.  See
>>> http://linuxmafia.com/faq/Kernel/proprietary-kernel-modules.html.
>>> While his answers to the question of binary kernel modules changes
>>> over time, he consistently holds to several principles:
>>>
>>> 1. Whether you're bound by the GPL depends on whether you're a
>>> derivative work of the Linux kernel.  Which will be true depending on
>>> the facts.
>>>
>>> 2. When the plugin API was a generic limited subset of standard Unix
>>> interfaces, there was no question that things which went against it
>>> were not derivative of Linux.
>>>
>>> 3. When the plugin API was later expanded to expose a lot of Linux
>>> specific details, things that use the full API become derivative of
>>> Linux.
>>>       
>> I call bullshit.  You can't say that with one breath, and with the next
>> rejoice over winning a lawsuit on the grounds that interfaces are not
>> copyrightable [to oversimplify].
>>     
>
> I don't know what lawsuit you're bringing up.  That I've summarized
> Linus' stated opinion over the years can be verified by reading his
> various statements on the topic over time.  It should be obvious that
> while Linus is an intelligent and important person, he is not a
> lawyer.  Salt what he says accordingly.
>
>   
>> I would simply publish function prototypes and struct definitions for
>> the plugin interface under a permissive license (MIT or Simplified BSD).
>>     
>
> Isn't this exactly the approach that I wound up suggesting?  I called
> it "a separate library", but that is exactly what an independent file
> of function prototypes and struct definitions is.  However, as I
> noted, if you do that then you should be careful about copying random
> useful-looking bits from the GPLed code to this file, because once you
> start copying things that other people contributed under a GPL, you
> don't have a clear right to say that that file is still under a
> permissive license.
>
> Ben
>
> Ben
>   


Thanks for all the info! after reading through it I'm probbly going to do what you guys suggested and since Its php (which is an interpreted language) I don't think the compiling thing is an issue but would being interpreted instead of compiled would that change things? I also have a few other questions. First, if I do this licence buffer approach will this make it so no libraries can be released gpl but would allow lgpl? Also could I do this with the agpl since it will be displayed on the web? (by the way thanks for clarifying what the other clause meant, the way everyone else worded it kind of confused me)

Ben Tilly wrote:
2009/8/7 Dag-Erling Smørgrav <des@des.no>:
  
Ben Tilly <btilly@gmail.com> writes:
    
It should be noted that while that is the legal opinion of the FSF,
there is no guarantee that a court will rule that way.  For an
interesting alternative, the opinions of Linus on binary kernel
modules is interesting.  See
http://linuxmafia.com/faq/Kernel/proprietary-kernel-modules.html.
While his answers to the question of binary kernel modules changes
over time, he consistently holds to several principles:

1. Whether you're bound by the GPL depends on whether you're a
derivative work of the Linux kernel.  Which will be true depending on
the facts.

2. When the plugin API was a generic limited subset of standard Unix
interfaces, there was no question that things which went against it
were not derivative of Linux.

3. When the plugin API was later expanded to expose a lot of Linux
specific details, things that use the full API become derivative of
Linux.
      
I call bullshit.  You can't say that with one breath, and with the next
rejoice over winning a lawsuit on the grounds that interfaces are not
copyrightable [to oversimplify].
    

I don't know what lawsuit you're bringing up.  That I've summarized
Linus' stated opinion over the years can be verified by reading his
various statements on the topic over time.  It should be obvious that
while Linus is an intelligent and important person, he is not a
lawyer.  Salt what he says accordingly.

  
I would simply publish function prototypes and struct definitions for
the plugin interface under a permissive license (MIT or Simplified BSD).
    

Isn't this exactly the approach that I wound up suggesting?  I called
it "a separate library", but that is exactly what an independent file
of function prototypes and struct definitions is.  However, as I
noted, if you do that then you should be careful about copying random
useful-looking bits from the GPLed code to this file, because once you
start copying things that other people contributed under a GPL, you
don't have a clear right to say that that file is still under a
permissive license.

Ben

Ben