Date: Wed, 14 Jul 2004 09:06:30 +0100 From: Steve O'Hara-Smith <steve@sohara.org> To: Garance A Drosihn <drosih@rpi.edu> Cc: current@freebsd.org Subject: Re: CVSUP and 5.2.1 RELEASE Message-ID: <20040714090630.4e11a48a.steve@sohara.org> In-Reply-To: <p06110419bd1a42ac1bda@[128.113.24.47]> References: <26481.1089755361@critter.freebsd.dk> <p06002065bd1a0a3116a6@[10.0.1.3]> <40F45DAC.5070800@linuxpowered.com> <40F48F6B.5070201@uiowa.edu> <p06110419bd1a42ac1bda@[128.113.24.47]>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 13 Jul 2004 21:56:49 -0400 Garance A Drosihn <drosih@rpi.edu> wrote: > And at least for now, a reliable set of steps must include a > "reboot and check kernel" before installing the world. There is > no way for `make' to do that. I'm not so sure about that - the existing world target could be renamed jailworld and run by the world target whenever / is not the DESTDIR. For installs to / staging with flag files (like the ports build system) should enable the world target to start with a mergemaster -p, buildkernel, installkernel and reboot if the kernel hasn't been built. Then if the kernel stage has passed it could go on with installworld and mergemaster. It would mean that make world has to be typed twice, although a bit of rc magic could avoid even that by testing for an incomplete make world run just after mounting the filesystems and before too much else has happened. Even without this getting users to make world twice is probably more reliable than getting them to use the sequence described in UPDATING which seems to be repeated wrongly a *lot*. -- C:>WIN | Solar Thermal Systems The computer obeys and wins. | http://www.soleire.com/ You lose and Bill collects. | Directable Mirror Arrays | http://www.sohara.org/
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040714090630.4e11a48a.steve>