From mboxrd@z Thu Jan 1 00:00:00 1970 Message-Id: <200312171632.hBHGWFfq019532@math.Princeton.EDU> To: 9fans@cse.psu.edu Subject: Re: [9fans] fortune-worthy In-reply-to: References: From: John Stalker Date: Wed, 17 Dec 2003 11:32:15 -0500 Topicbox-Message-UUID: a78561b0-eacc-11e9-9e20-41e7f4b1d025 > I think however that its main design goal---easy distribution of services > and resources and the shared environment it presents to the users---has > been 'superseded? replaced? ignored?(pick your favorite)' in the real > world by the web, and that is another discipline. I disagree somewhat. Per machine namespaces are really a great nuisance. I have three and a half machines at home and one at work and I already feel the strain of keeping everything in sync. Web interfaces are not much good for this, and the standard solutions--NSF, rsync, etc.--feel like the hacks that they are. I can only imagine what administering a large network is like. The problem is, I think, a bit more subtle. The per process namespace idea and largely transparent networking are still as good an idea now as they were originally. Plan9 doesn't work as seemlessly as I would like, however, in a network whose topology is continually changing. My laptop, for example, may be part of my LAN, the internet, my employer's LAN, or operating stand-alone. Other OSes are no better at this, but it seems that this is the sort of thing to which plan9 should be well suited. Plan B has some interesting ideas in this area. Plan9 also has some problems at small scales. It really wants a dedicated file server, though not quite as badly as in the old days. -- John Stalker Department of Mathematics Princeton University (609)258-6469