Date: Mon, 30 Nov 1998 13:29:21 -0500 (EST) From: "John S. Dyson" <dyson@iquest.net> To: eivind@yes.no (Eivind Eklund) Cc: dyson@iquest.net, adrian@ubergeeks.com, rssh@grad.kiev.ua, grog@lemis.com, wes@softweyr.com, tlambert@primenet.com, hackers@FreeBSD.ORG Subject: Re: System V init (was: Linux to be deployed in Mexican schools; Where was FreeBSD?) Message-ID: <199811301829.NAA03045@y.dyson.net> In-Reply-To: <19981130165346.N9226@follo.net> from Eivind Eklund at "Nov 30, 98 04:53:46 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
Eivind Eklund said: > On Mon, Nov 30, 1998 at 10:10:32AM -0500, John S. Dyson wrote: > > With a proper structure, there is flexibility. The current rc > > setup isn't on the surface very flexible and is overly monolithic. > > Additionally, the more that "special tweaks" need to be done for a > > software port, the more there is cost for support of an OS. IMO, it > > is best to look like a predominant market player. Purely technical > > discussions end up being continually tweaked and tuned. I am suggesting > > a normalization to a market "standard." On the technical merits, this > > discussion can last forever, because there are lots and lots of > > technical solutions. > > > > There is the /usr/local/etc/rc.d thing, but that is a superficial > > attempt to look similar to the "standard", but doesn't really perform > > the functions of it. I suspect that a "really nice" rc scheme could > > be layered on top of the "standard" scheme, just like a "really nice" > > rc scheme could be layered on top of the BSD scheme. The question is, > > just why not be compatible? > > I'm in favour of compatible. I'm not sure if this require us to bring > in the bad sides of the SysV system. > IMO, SYSV init isn't one of the "bad" things about SYSV. If anything, it is just not finished (shipped before it was fully thought out.) I suspect that a reasonable compromise is to implement a SYSV style init correctly, providing SYSV compatibility, with the desired flexibility and functionality that we would all prefer. I truly believe that the current BSD scheme is more of a "punt" as opposed to a superior solution. Making a choice to stay with the limited BSD init scheme because the SYSV scheme has it's own problems doesn't solve the problems of either scheme. The SYSV scheme probably mechanically solves the problems better, but can be a mess. The BSD scheme is a problem because of it's monolithic and almost non-layered structure. These technical arguments really beg the issue of compatibility though. Ignoring technical arguments AGAIN, SYSV (and Linux) compatilbility are desirable for packaging and porting reasons. There are lots of things that one might have to do for compatibility reasons (limited 32bit disk file addressing in the Linux emulation), however, if the compatibile choice isn't made, then problems ensue. It would be very bad to say something like "the linux 32bit file offsets are bogus, so we will unilaterally make the offsets 64bit and break compatibility -- but we are RIGHT :-)". > > Eivind, who really believe this discussion belong in freebsd-arch > (which is now open and moderated). > As long as I am CC:'ed when people want to talk to me, I'll reply :-). -- John | Never try to teach a pig to sing, dyson@iquest.net | it makes one look stupid jdyson@nc.com | and it irritates the pig. 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?199811301829.NAA03045>