Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site sjuvax.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!princeton!astrovax!sjuvax!jss
From: jss@sjuvax.UUCP (J. Shapiro)
Newsgroups: net.arch
Subject: RMS v/s UNIX (non-religious)
Message-ID: <917@sjuvax.UUCP>
Date: Wed, 27-Feb-85 22:49:36 EST
Article-I.D.: sjuvax.917
Posted: Wed Feb 27 22:49:36 1985
Date-Received: Fri, 1-Mar-85 20:42:35 EST
Organization: Saint Josephs Univ. Phila., Pa.
Lines: 23

[Pacman's revenge...?]

	I agree that this conversation has run overlong, but one thing that I
am sure all of us have noticed by now is the fact that neither the UNIX file
system nor the RMS services  is perfect for everything. There are other
regards in which UNIX is better than VMS or vice versa. (E.g. pipes in
favor of UNIX, file and record locking in favor of VMS).

	One poster (regrettably I forget who) suggested that an operating
system should be built using a bottom level file model with libraries to
manipulate that bottom level at successively higher levels of abstraction.
This seems on initial consideration to be a good idea - it avoids always
paying for things you don't really want.  The problem lies in picking an
underlying model sufficiently robust to do all of the things you want to do
with record and file locking, and in noting which level of abstraction a
file shuold be evaluated under.

	Has anyone out there given some thought to what might make a
sufficiently robust underlying system?? There are things UNIX can't provide
under its current incarnation even with library routines....

Jon Shapiro
Haverford College