On Sun, Aug 26, 2001 at 01:23:32AM -0700, Simon J. Gerraty wrote:
> As I mentioned - I like kre's "associated data" model - which could be
> used to implement (the storage of) ACL's but is much more generic - so
> hopefully each cool thing that comes along after ACL's could be
> implemented without the need to introduce new filesystem types each
> time.  I'm sure there are lots of details that need to be worked out,
> but on principal - I like it :-)

Um, a side topic: what would it take to get an {,Open}AFS server running on
NetBSD? 

I think I remember someone once saying that AFS stores data in extra
fields in the inode, but NetBSD doesn't have enough spare fields. Could
we use the "associated data" model (or the one-large-file-indexed-by-inode
model) to hold this AFS server data?

In either case, could we have a push-down layer that hides the extra
file or files from the upper filesystem layers? We would need a new
push-down layer for each new set of additions, but that wouldn't be
terrible.

(BTW, has anyone released the AFS patches to OpenAFS in order to run
an OpenAFS client on NetBSD?)
-- 
Kevin P. Neal                                http://www.pobox.com/~kpn/
           On the community of supercomputer fans:
"But what we lack in size we make up for in eccentricity." 
  from Steve Gombosi, comp.sys.super, 31 Jul 2000 11:22:43 -0600