On Thu, Mar 23, 2000 at 04:39:47PM -0600, Manoj Srivastava wrote: > >>"Radim" == Radim Kolar <hsn@cybermail.net> writes: > Radim> In section 3.3.2. is not clear what 'restart' does. > Radim> 1. restart=stop and start > Radim> 2. if service is running, stop it and start it, when is not > Radim> running, do not start it. > [...] > I do agree that these two behaviours are valid, and are > required in some situations. What I do not have a handle on yet is > which is the more common case, and thus which should be the default > behavior. > > Either we needd restart+force-restart, or we need a > restart+maybe-restart. The main case (IMHO) where we need a `maybe-restart' is in postinst's so that an upgrade doesn't restart a service that doesn't need to be restarted. I personally think a more useful thing would be a "start-rc.d" that you use a la: $ cat debian/postinst #!/bin/sh start-rc.d inetd restart which will then call /etc/init.d/inetd restart iff inetd ought to be started in whatever runlevel we're currently in. This copes with people who disable a service by stopping it and deleting the rc.d links (which isn't coped with atm, really), but not with people who just stop a service. Note, though, that even with a `maybe-restart', this wouldn't be coped with: portmap, for example, gets stopped in the prerm and started in the postinst rather than just restarted in the postinst. start-rc.d could also `cd /' and clear the PATH and any other weird environment variables, which would be a nice side benefit. Note that start-rc.d would need to be provided by both dpkg and file-rc. Sample implementations were posted to -devel last year or so. Ummm, check: http://www.debian.org/Lists-Archives/debian-devel-9910/msg00427.html maybe. The weblogs seem to be missing the mail I was actually thinking of (although it's in the logs on master). Bizarre. Cheers, aj -- Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/> I don't speak for anyone save myself. GPG encrypted mail preferred. ``The thing is: trying to be too generic is EVIL. It's stupid, it results in slower code, and it results in more bugs.'' -- Linus Torvalds
Attachment:
pgpSz8Z0qO800.pgp
Description: PGP signature