*BSD News Article 30561


Return to BSD News archive

Xref: sserve comp.os.linux.misc:15119 comp.os.386bsd.misc:2466 comp.unix.unixware:5298 comp.unix.solaris:16348
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!news.mtu.edu!sol.ctr.columbia.edu!howland.reston.ans.net!europa.eng.gtefsd.com!MathWorks.Com!news.kei.com!news.byu.edu!news.mtholyoke.edu!nic.umass.edu!opine.cs.umass.edu!opine!hagan
From: hagan@opine.cs.umass.edu (Craig I. Hagan)
Newsgroups: comp.os.linux.misc,comp.os.386bsd.misc,comp.unix.unixware,comp.unix.solaris
Subject: Re: A good NFS server ?
Date: 20 May 1994 15:54:36 GMT
Organization: University of Massachussetts Computer Science Department
Lines: 38
Distribution: inet
Message-ID: <2rimfs$43t@opine.cs.umass.edu>
References: <CpC9Fq.I2n@acsu.buffalo.edu> <2rcdee$c8@wea.eel.ufl.edu> <Cq1HGz.8w1@hippo.ru.ac.za> <1994May19.115606.11109@uk.ac.swan.pyr> <Cq27xK.9Ix@novell.co.uk>
NNTP-Posting-Host: opine
X-Newsreader: TIN [version 1.2 PL0]

Martin Sohnius (msohnius@novell.co.uk) wrote:
> Alan Cox (iiitac@uk.ac.swan.pyr) wrote:
> : In article <Cq1HGz.8w1@hippo.ru.ac.za> csgr@cs.ru.ac.za writes:
> : >[.. something about Linux and freeBDS...]

> : [.. something else about Linux and something about BSD in general...]

> Hey, you guys,  this whole thread is degenerating into a little
> NFS-on-BSD discussion!

the point is to get people talking about what is a good nfs
server. since linux isn't the best, people who need the
best now should be aware of their options. people developing
for linux should know where to look for implementation ideas. 
If i remember right, *BSD got a lot of nifty ideas
from linux already (and vice versa), isn't that the point
of free software and competition (even if it is friendly)?
I like having multiple free os's for the pc. it allows
people to explore in different directions. I believe
that anthropoligists would call this a diversified gene pool,
and with all of this stuff kicking around, we can potentially
introduce some hybrid vigor into the linux nfs server/client.
I look forwards to this -- and if i have some more time,
i intend upon finishing some work on nfs' caching 
that i started for linux. If not, changing the current
(useable, but slow) implementation is pretty simple,
change the linear search to a cache -- i was working with
a MxN hash table, hash into M, linear search through N (e.g. 256x4 table)
and allow M and N to be parameters in a .h file.

i would be happy to provide my code (which doesn't yet compile)
to anyone who wants to work, as i haven't been able to look at
it for 2 months -- nfs hasn't changed since then,
so it is still drop in for any kernel.



-- craig