Date: Tue, 1 Dec 1998 04:39:56 -0500 From: Garance A Drosihn <drosih@rpi.edu> To: Adrian Filipi-Martin <adrian@ubergeeks.com>, Matthew Dillon <dillon@apollo.backplane.com> Cc: Dan Busarow <dan@dpcsys.com>, Eivind Eklund <eivind@yes.no>, hackers@FreeBSD.ORG Subject: Re: System V init (was: Linux to be deployed in Mexican schools; Where was FreeBSD?) Message-ID: <v04011701b28964562781@[128.113.24.47]> In-Reply-To: <Pine.BSF.3.96.981201033655.4238Q-100000@lorax.ubergeeks.com> References: <199812010743.XAA03898@apollo.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
At 3:43 AM -0500 12/1/98, ADRIAN Filipi-Martin wrote: >On Mon, 30 Nov 1998, Matthew Dillon wrote: > >> Not to detract from the other threads I'm trying to push commits >> through, but... oh damn, I just can't pass it up: Why not organize >> an RC system based on the specification of a set of dependancies? >> >> For example, require a '#DEPENDS/RETURNS' line (or equivalent) in >> the first 2 lines of each rc file. >> >> ---- some rc.xxx file ---- >> #!/bin/sh >> #DEPENDS(network,sendmail,...) RETURNS(atm) >> # >> ... >> -------------------------- > > Hmmm, not compatable with anything else, and more complicated. > Were order dependencies ever that hard to begin with? Sure someone > has to read the manpages to know that portmapper needs to run before > nfsd, but are we really solving any problems by having the system > dynamiacally determine that "yes, we still need to start portmapper > before nfsd"? I think the DEPENDS/RETURNS idea is pretty interesting. The initial goal is just to break up the "more-monolithic" BSD startup scripts into smaller pieces. I do think that's a good thing (and yes I've worked with both BSD and SysV setups). This DEPENDS/RETURNS idea is not meant to solve some current problem, it's meant to solve the problems which crop up once you do break up the "more-monolithic" startup scripts into lots of little scripts. I, for one, would like to see the "more-monolithic" (for lack of a better phrase) scripts broken up. I'd be happy enough with some SysV-ish result, but I'd be much happier if we happened to add in some idea such as this to solve the ordering-of-little-files problem. > I'm not trying to pick on you in the above, but a lot of the > justification for svr4 init/statt/stop has shifted towards being > compatable with other systems like linux, solaris, etc. If I'm thinking about this clearly (and that is not at all certain given the hour...), I think this idea can be compatable with sysV setups, or at least "compatable" in the way I want it. All this requires is some comments added in the little RC file. >From the standpoint of a person writing a script, the contents of the script could be exactly the same for freebsd and SysV. (they do have to add a few lines, but adding those lines does not break anything under SysV). The *name* of the file may have to be different too --- but then SysV by itself can get you into the situation where you need different names for the same startup script anyway, depending on what else is installed. The specifics of what those comment lines should be, and how they would be processed at startup under freebsd may need some more thought, but I like the general direction of this idea. --- Garance Alistair Drosehn = gad@eclipse.its.rpi.edu Senior Systems Programmer or drosih@rpi.edu Rensselaer Polytechnic Institute To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?v04011701b28964562781>