From owner-freebsd-current@FreeBSD.ORG Tue Aug 17 05:45:24 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 8A7B816A4CF for ; Tue, 17 Aug 2004 05:45:24 +0000 (GMT) Received: from jeffenstein.dyndns.org (Lsne-CaTV-dhcp-15-72.urbanet.ch [80.238.15.72]) by mx1.FreeBSD.org (Postfix) with ESMTP id B6F0343D68 for ; Tue, 17 Aug 2004 05:45:23 +0000 (GMT) (envelope-from jeff@jeffenstein.dyndns.org) Received: by jeffenstein.dyndns.org (Postfix, from userid 1001) id 1A388155D6; Tue, 17 Aug 2004 07:45:03 +0200 (CEST) Date: Tue, 17 Aug 2004 07:45:01 +0200 From: Jeff Fisher To: current@freebsd.org Message-ID: <20040817054501.GB942@frogger.jeffnet> Mail-Followup-To: current@freebsd.org References: <20040816174010.GA82600@prophecy.dyndns.org> <20040816181501.38B785D09@ptavv.es.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; x-action=pgp-signed Content-Disposition: inline In-Reply-To: <20040816181501.38B785D09@ptavv.es.net> User-Agent: Mutt/1.5.6i Subject: Re: RFC: Alternate patch to have true new-style rc.d scripts in ports (without touching localpkg) X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 17 Aug 2004 05:45:24 -0000 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Mon, Aug 16, 2004 at 11:15:01AM -0700, Kevin Oberman wrote: > > Date: Mon, 16 Aug 2004 13:40:10 -0400 > > From: Christopher Nehren > > Sender: owner-freebsd-current@freebsd.org > > > > --/04w6evG8XlLl3ft > > Content-Type: text/plain; charset=us-ascii > > Content-Disposition: inline > > Content-Transfer-Encoding: quoted-printable > > > > On Mon, Aug 16, 2004 at 11:56:53 EDT, Mike Makonnen scribbled these > > curious markings: > > > I have thought about this considerably, and I think the best solution > > > is to have ports rc.d scripts installed to /etc/rc.d. One of the problems > > > > Please, no. This is in direct violation of hier(8), POLA, the concept > > of separating third-party packages from the base system, and it also > > pollutes the concept of a lean, clean, vendor-provided / file > > system. One of the things that I love about FreeBSD is that it doesn't > > make a mess of the base system like Linux does. If I wanted the mess > > that putting port scripts in /etc/rc.d would cause, I'd use Linux. I'd have to agree. Startup scripts are generally the exception on other unixy o.s., but this does not look like a good direction. First, startup scripts are sourced, and now they are moved into the root filesystem. Soon, they'll be installed under /usr... Someone else has already asked what benefit this sourcing gives, but they went unanswered, so I'll try again. :-( > There are many local system mods that require configuration files and/or > scripts be available prior to mounting of any file system. > > I do like a separation of local stuff, but it really, really should be > in the physical root partition. (And I oppose the use of symlinks, as > well. That's really ugly!) The problem is that all of the ports require /usr/local (or /usr/X11R6) to be available before they can do anything. Putting their startup scripts in / won't add Jack over the existing system. Just to go off onto a bit of a rant, I originally chose FreeBSD over Linux for just a couple important reasons: it was stable, and it had this great ports tree. One of the great things about the ports tree was that it installed stuff seperately from the OS. With linux, everything gets put into /usr, and you don't know what you've added afterwards. With FreeBSD, it's added in a seperate package, and in a seperate directory; all very clean, and much closer to commercial unix versions. Commercial unix versions are like this for a reason: manageability on the large scale. Also, stability could be threatened with the recent change to source ports scripts instead of run them. Now a mistake in a ports script can cause all of the following ports to not start, leaving a half-running system. With the old system of running ports scripts, if there is an error or a problem introduced by a local modification of a port startup script, it affects only that port. If the objective is to simplify writing startup scripts, could the sourcing not have been in the opposite direction? Why not just supply a script with common functions, and have the ports scripts source this common script? This accomplishes the same task without sacrificing stability. And now, I see that the seperation of ports from the base o.s. is also being threatened. Not greatly, but moving the startup scripts now, and later some config files, and then why not move the binaries? It'll only break a few sites stupid enough to run nfs-mounted or rsynced /usr/local... Pardon me if I've gone too far, but I love the O.S., and only want it to remain the same O.S. I chose to use and advocate so long ago. - -- jeff@jeffenstein.dyndns.org http://jeffenstein.dyndns.org/ PGP encrypted mail preferred. Key id 0x19C987F5 === Hanlon's Razor: One should never attribute to malice that which can be adequately explained by stupidity. -----BEGIN PGP SIGNATURE----- iQIVAwUBQSGbXRwPMBUZyYf1AQIMww/9Grk6m8bxEQ9ZtbOl9PzYh5L9YRnLfUNz wPEf9bFwB7aRkcOGx9VgDnMQVGtcIXpVUc7jkx/oJHlUwLtlG5pn/+iodIfGRQjw tLrRz+NThhyjnvljozHPzmCX0ptNKoiXq+ChBF7O0vSpqe8TIVb3sCquTt1e2d0j UX+1rM5rILKuDHPf0ipE+iBwdeC3Po8T88gSb3O07bloH0FGu/8lfIaRNH0h8MDa e2jZ6gGOfa8zRr9bD3styashHulq0GWOfvijhUoBjaPb9wOOMpz9Ws7cm7mhDj8j 2pF/BGYJt0R40Q/U2YaBNImXMVAQM/WgJJhJo/tk7ELZ/u4xHDKNp4NSTkYa+gCx 31gzXyaj/h8t/I37DcEKiPKWrMFqya/WOKXNF3s+vHy/RRxgFVJFlDulgAGlXSfs wQ/DUijiK/nPFdAwMUHSIVcyN7EKAdyJayVgQwkrECvA43Me4seOEYnk1hMFs1B2 wHs364qEr1wzC37QLZZKoH/tisYrTfo0Tn1I799PjjdVeMneT3pHoBbagYqoFb5j pUBPvx/gM6l4cp8/s2TqXN/dZIai3Z0bgMqIZmxJ1Q+KkkO+VeTd2sS4Ha+lKrwR LM/RfyrFDR7OUMdmF5/AM/ThiFC21SPl9fGHLCS2t/LZDk8TKhGxyn7BXjFRMJKI G3qEIUxATvE= =JRTm -----END PGP SIGNATURE-----