Subject: Re: kernel & libkvm [was IIci success]
To: None <current-users@NetBSD.ORG>
From: Perry E. Metzger <>
List: current-users
Date: 01/11/1996 22:07:06
Open Carefully -- Contents Under Pressure writes:
> I won't even begin to dispute the claims that procfs and kernfs are useful
> in their own contexts.  "Recommended" is a good label for them.  "Required"
> is not.  I wouldn't "require" kernfs or procfs any more than I would
> "require" NFS.  The only FS which I could see justifying as "required"
> is UFS, which is already there.*

One can take another view.

In some sense, all user level programs on Unix-like systems are
"optional", but who could survive without, say, syslogd, or ls?

As it stands, we have three mechanisms to get information on kernel
state -- kernfs, sysctl, and kernel grovelling. This doesn't feel
particularly clean. If it were thought that kernfs was the best way to
do stuff, the mere fact that it had not been heretofore required
wouldn't be reason enough not to require it in the future.