Subject: Re: Driver hierarchy
To: Chris G. Demetriou <cgd@netbsd.org>
From: Allen Briggs <briggs@ninthwonder.com>
List: tech-kern
Date: 03/20/2000 05:55:49
On Mon, Mar 20, 2000 at 12:04:16AM -0800, Chris G. Demetriou wrote:
> (i.e. mips and pmax logically shouldn't
> live in the same directory of the source tree.)  as reasonable
> alternatives, i could buy cpu/mips and arch/pmax (the least amount of
> stuff moves), arch/mips and machine/pmax (more stuff moves, but the
> naming's more sensible), or maybe even arch/mips and
> arch/mips/machine/pmax (the world is drastically changed, but this may
> make even more sense... but then it doesn't distinguish between mipsel
> and mipseb MACHINEs)...

I don't know how much is shared between, say, luna68k and luna88k--I
know not much is shared between sun3 and sparc or mac68k and macppc,
but some is (arguably, adb should be in dev/adb, though)...  I tend
to think that the most reasonable arrangements don't try to nest cpu
and machine architectures, so I think that

	cpu/mips and machine/pmax,
	arch/mips and machine/pmax, or
	cpu/mips and arch/pmax

make the most sense.  The first makes the most sense to me, and the
last requires probably the least change for most people's navigational
thinking.

-allen