Re: Rumpification (was Re: CVS commit: src/usr.sbin/envstat)

[Antti Kantee <pooka%netbsd.org@localhost>]

On Tue Dec 14 2010 at 22:19:15 +0000, David Holland wrote:
> If my explanation is wrong, please provide a correct one, as many
> people want to understand why you're making a mess. You will never be
> able to convince them (or me) that this mess should be accepted if you
> don't/can't explain it; nor is there any chance that someone will be
> able to suggest a less messy alternative.

Did you read the webpage and/or manpages?  Unless people do that *and*
throw away any invalid based on guesses, we're not going to get anywhere.
Luckily, we'll at least be left with working code.

> While you're at it, as far as I can remember you never explained why
> filesystem-level access to a rump kernel should be via dd instead of
> via e.g. nfs or 9p, either.

Device access on nfs does not DTRT, whereas with dd you want to dd
devices (it's not so much about file system access as it is about file
namespace access).  Besides, nfs uses networking, which I thought I just
said is not available always.

9p might be applicable as soon as our kernel and base system supports
it and we can transport it without requiring kernel networking, but I'll
evaluate it when that time comes.

-- 
älä karot toivorikkauttas, kyl rätei ja lumpui piisaa