Re: main or contrib?

[Michael Poole <mdpoole@troilus.org>]

Francesco Poli writes:

> On Mon, 23 Oct 2006 08:50:00 -0400 Michael Poole wrote:
>
> [...]
>> I personally disagree -- on the grounds that the software works as it
>> should without the blobs, and the hardware is what fails to provide
>> the necessary interface -- but mine is a minority viewpoint.
>
> By that line of reasoning, a non-free-kernel could be seen as a means to
> cause the hardware to provide the necessary interface, and any
> user-space DFSG-compliant program could go in main, even if it somehow
> requires the non-free-kernel in order to be useful.
> 
> Or you could conceive the kernel + non-free-interpreter combination as a
> means to cause the hardware to provide the necessary interface, and any
> DFSG-compliant script could go in main, even when it requires the
> non-free-interpreter in order to be useful.
>
>
> IOW, I'm not convinced by your argument.

If many people were, mine would probably not be a minority viewpoint.

It is commonly held that software in main might only be useful when
used with non-main (and non-free) software over a network.  I think
the same rule should apply when talking to some reasonably discrete
device (such as the DSL modems of this thread).  You clearly do not.

I doubt either of us will change the other's mind.  I also doubt
either of us has a bright line test that distinguishes the two types
of dependency.

(My perspective is born of close work with many FPGA developers.  I
find it natural to distinguish between the CPU-side software I write
and the firmware they write.  I also understand why others disagree.)

Michael Poole