Re: The Kernel

[William T Wilson <fluffy@dunadan.com>]

On Tue, 20 Jan 1998, Nick Mathewson wrote:

> The concept of the Linux Core, as I understand it, would not forbid
> developers from including kernel patches, but only from including
> kernel patches that would break the functionality of the Core.

This is the crux of the disagreement here.  Is the core a strong policy to
be adhered to, or is it just a weaker set of guidelines that you can
disregard when you think you have a "good enough" reason? 

I believe in a strong core, because that is necessary if the purpose of
the core is to be preserved.  A system which is compliant with Core
version x.y should be bit-for-bit compatible with it.

While it seems reasonable to say "We must not change any userland kernel
structures, nor system calls" I wonder what kernel functionality it is
possible to add without making such changes?  The only things I can think
of are additional device drivers, which, usually, go into the regular
kernel as soon as they are "ready for prime time."  Consider the FAT32
filesystem, it is only available as a patch.  I think it is stable, well
tested, has little or no impact on userland kernel functionality and is
one of the better patches out there.  So why is it not part of the kernel?
Probably because the policy of 2.0 is not to add features, only to fix
bugs.

Whether such modifying should be done is widely open to debate, and goes
to the heart of what the Core really is.