Date: Sun, 24 Sep 1995 09:13:56 -0400 (EDT) From: "Jonathan M. Bresler" <jmb@kryten.Atinc.COM> To: Peter da Silva <peter@taronga.com> Cc: hackers@freebsd.org Subject: Re: ports startup scripts Message-ID: <Pine.3.89.9509240822.E27857-0100000@kryten.atinc.com> In-Reply-To: <199509241241.HAA06911@bonkers.taronga.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 24 Sep 1995, Peter da Silva wrote: > > > > with the K* and S* files in different directories, one for each > > > >run level, ascertaining the differences is needlessly harder. > > > > There are no differences unless you're using a broken editor. > > > i dont understand this. what editor are you suggesting? > > Some editors break links by renaming the original file to .bak. VI doesn't. solves the rename problem. not the 'is it the same of is it not, i have to did deeper' problem. startup and shutdown are inherently complicated with strict order dependencies (eg mount, nfs, ifconfig, routing, ......) a single file (that may well invoke others to do the work) serves as a clear outline of the startup/shutdown process. > > a check for what? for identical files? > > For non-identical files. Files with the same tail should be the same file. should, but you question consistency and ability of those writing install scripts. same question applies to files, contents, and filenames. with all of them in one directory, one can NOT have a single filename performing two non-identical sets of commands. removes a source of possible error > > copy the file to /tmp. edit the version in tmp. display the > > diff. either prompt for accepting the change (all ports that add > > services become interactive) or report the new file's location and ask > > the user to verify and install. > > And you expect everyone to do that consistently? > > See, the problem is you have lots of people with varying levels of > competance writing install scripts, often trying to make the same script > work on BSD and System V and OSF/1 and NextStep and... i expect the ports-meister and packages-master to enforce that level of consistency. the newbie sysadmin expects/wants clear unequivocal directions. the experienced one will do what he knows is 'the right thing' ;) > > > I'm not that averse to having a unified directory, but each component should > > > have its own startup and config file. > > > okay, sounds good. one file, one service. one master file that > > coordinates them all. > > NO. No master file. There is *no* justification for one. > > (If I had my druthers I'd have an /etc/inetd.d as well) no master file?? what executes the individual files and determines the order of execution....ascii sort order in a shell script? thats a master file which is generated on the fly. and must rely on numerics to determine execution order rather than names. (eg 67 vs named, we could use 67named, at least. preserve sort order and provide information regarding action taken) master file could be a shell script, but in place of ascii sort order, a shell variable and a 'for x in $daemons' loop. (are we much closer to agreement than i realize?) > > > Like I said in my original response, just having /etc/rc.d with S and K > > > scripts run by /etc/rc and shutdown would be a massive improvement. > > > isnt this the one file that each port has to modify. an > > objection you raised above? > > Nope. You only move files into and out of the directory. > > I even ported this scheme to my Amiga. > ;) Jonathan M. Bresler jmb@kryten.atinc.com | Analysis & Technology, Inc. FreeBSD Postmaster jmb@FreeBSD.Org | 2341 Jeff Davis Hwy play go. | Arlington, VA 22202 ride bike. hack FreeBSD.--ah the good life | 703-418-2800 x346
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.3.89.9509240822.E27857-0100000>