Re: The Kernel

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

On Tue, 20 Jan 1998, George Bonser wrote:

> .config file.  Chances are the core kernel will appear in exactly 0
> distributions as distributed. All will probably do a make config with
...
> We are not going to help anyone by trying to enforce a particular make
> config on the world.

We need to specify, then where the core ends and distribution liberties
begin.  You say that it's unreasonable to enforce a particular make
config, this seems reasonable.  But extend this line of thinking, to the
thought of adding, say, a FAT32 patch.  Is this within the bounds of the
core, or is it a distribution liberty?  Now suppose someone wants to
distribute a kernel with performance-tweak patches, or the stack-execution
patch, or the patch for modularizing sound, like redhat did, and broke it
altogether?  This moves into the realm of UNreasonable patches.  Or
suppose the core specifies a 2.0.33 kernel, but someone wants to
distributed .32 or .34.  Is this reasonable?  Maybe, but what if they want
to distribute 1.2.18 or 2.1.70?  Is THAT reasonable, probably not.  How
much influence does the Core have over the kernel, and how strongly must a
distribution follow the core in order to be considered compliant?

> Please remember there is a clear distinction between core and SEUL. SEUL
> is built using the core sources.

I am not confusing this distinction, what I am saying is that we need to
have clarified exactly where one ends and the other begins.