[3209] in Athena Bugs
Re: /tmp/Xbin (actually deactivate in general)
daemon@ATHENA.MIT.EDU (probe@ATHENA.MIT.EDU)
Fri Sep 15 10:35:17 1989
From: <probe@ATHENA.MIT.EDU>
Date: Fri, 15 Sep 89 10:34:57 -0400
To: hkbirke@ATHENA.MIT.EDU
Cc: bugs@ATHENA.MIT.EDU, quality@ATHENA.MIT.EDU, danapple@ATHENA.MIT.EDU,
In-Reply-To: Hal Birkeland's message of Fri, 15 Sep 89 03:16:20 -0400,
Reply-To: Richard Basch <probe@ATHENA.MIT.EDU>
/etc/athena/deactivate does not get run until there are no entries in
/etc/utmp. No claims have been made if it is run manually as root.
The cleanup turns out to be essential.
1) it should not remove /tmp/* blindly on deactivation! this stomp
on many things such as X servers (not usually a problem), galatea servers
(a major problem) and /tmp/Xbin (a major problem... emacs bounces through it
regardless of terminal type!). instead it should remove all files and leave
directories and unix-style directories alone (ie, remove the contents thereof
but not the directories themselves and leae .*/* alone)
Since /tmp/Xbin gets created during /etc/athena/activate and
/etc/athena/deactivate does not normally get run while people are logged
in, there shouldn't be a problem here. /tmp is considered VERY
temporary space, and anything requiring more than a login session should
not be stored here. /usr/tmp is considered temporary, but will last for
a few days.
2) CellServDB.public should be grabbed locally if a local
one exists (and from the packs if the local one isn't there).
Supported customization:
Break the link from CellServDB to CellServDB.public. If you want to
maintain a reasonably up-to-date one, may I suggest the following in
deactivate.local:
(cd /usr/vice/etc; cat CellServDB.public CellServDB.local > CellServDB)
-Richard
