Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/3/84; site wjh12.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!harvard!wjh12!kendall
From: kendall@wjh12.UUCP (Sam Kendall)
Newsgroups: net.unix-wizards
Subject: /tmp vs. /usr/tmp (System V)
Message-ID: <528@wjh12.UUCP>
Date: Sun, 14-Oct-84 22:03:11 EDT
Article-I.D.: wjh12.528
Posted: Sun Oct 14 22:03:11 1984
Date-Received: Tue, 16-Oct-84 06:08:43 EDT
References: <5279@brl-tgr.ARPA>
Organization: Delft Consulting Corp., New York
Lines: 22

> No utility should use /tmp for large files.  That is what /usr/tmp is for.

(I think this is only supposed to be the case for USG UNIX.  Do non-USG
utilities use /usr/tmp much or at all?)  But I have seen a System V
system with a small /usr/tmp and a very large /tmp; conversely, I have
seen a 3B2 which came configured with a small /tmp and large /usr/tmp--but
the system utilities (compiler, assembler) on the 3B2 used /tmp by default,
leading to overflows!

   There is an undocumented (?) way around this--the TMPDIR environment
variable, which many utilities on System V use to determine their choice
of /tmp directory, if TMPDIR is available.  It would be nice if there
was a standard program which would look at the output of df and output
the proper TMPDIR value, perhaps based on a given amount of disk space
required--shell scripts or .profile's could use it.  I have a use for
such things, and I think anyone else who uses a lot of tmp space (e.g.,
anyone who uses the wonderful but space-greedy System V C cross-
referencer "cxref") would also.  Does anyone think this is a good idea?
Any volunteers to write it?  I'm not on a System V system.

	Sam Kendall	  {allegra,ihnp4,ima,amd}!wjh12!kendall
	Delft Consulting Corp.	    decvax!genrad!wjh12!kendall