Date: Thu, 9 Jun 2005 13:10:17 -0700 From: Brooks Davis <brooks@one-eyed-alien.net> To: Doug Barton <dougb@FreeBSD.org> Cc: freebsd-rc@FreeBSD.org, "J.R. Oldroyd" <fbsd@opal.com> Subject: Re: Use of rcorder for local rc.d/*.sh scripts Message-ID: <20050609201017.GA10248@odin.ac.hmc.edu> In-Reply-To: <42A89CD7.1080404@FreeBSD.org> References: <20050607001447.GG37208@linwhf.opal.com> <20050607003142.GD10526@odin.ac.hmc.edu> <20050607033536.GH37208@linwhf.opal.com> <20050607160855.GO37208@linwhf.opal.com> <20050607173741.GI11758@odin.ac.hmc.edu> <20050607191109.GU37208@linwhf.opal.com> <20050608233802.GA29707@odin.ac.hmc.edu> <20050609003459.GK37208@linwhf.opal.com> <20050609050517.GA28710@odin.ac.hmc.edu> <42A89CD7.1080404@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--OXfL5xGRrasGEqWY Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Thu, Jun 09, 2005 at 12:47:35PM -0700, Doug Barton wrote: > I've been following this discussion with interest, and I think that Brooks > is right about not having a flag day. On the other hand, I have seen many > people say that they want to get the ports rc.d scripts integrated into t= he > main rcorder, since some of their software "needs" this functionality. Th= is > will become increasingly important as we move to modularize the base more > and more (ala BIND and openssl). > > So, what about this for a compromise? We continue running localpkg to han= dle > the 000.*.sh and such, with JR's hack to incorporate those better. But we > ALSO build in the functionality to the main rcorder implementation to han= dle > scripts that do not have *.sh, and we require port authors to include a > special keyword in the script so that rc.subr will know that it is "safe"= to > handle them this way. Once the transition is "complete," we drop processi= ng > of the keyword, and the port authors can remove it at their leisure. This sounds mostly workable. The generation of the rcorder commandline gets a bit more complicated because we'll need to make two passes through /usr/local/etc/rc.d in rc, but that's probably acceptable. I'm a bit unhappy with doubling the size of code in /etc/rc to do this, but I suppose it's probably worth it. One thing to worry about is that this does actually require a flagday. Currently, the RC_ORDER bsd.port.mk variable allows scripts to be included in rcorder by installing them directly in /etc/rc.d. Any thing that used the new flag and moved back to /usr/local/etc/rc.d would stop working on 5.x in this case. I suppose we could modify the RC_ORDER support to install the files in /usr/local/etc/rc.d on systems with the rc changes. That would require a couple of port build runs though. -- Brooks --=20 Any statement of the form "X is the one, true Y" is FALSE. PGP fingerprint 655D 519C 26A7 82E7 2529 9BF0 5D8E 8BE9 F238 1AD4 --OXfL5xGRrasGEqWY Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (GNU/Linux) iD8DBQFCqKIpXY6L6fI4GtQRAsQMAKDEtFlN0DY3dfjNTeSo5IWfiJfL3gCeLtKm 5NzcOP1+MkVR0tcw9EEhWbM= =0Dcp -----END PGP SIGNATURE----- --OXfL5xGRrasGEqWY--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050609201017.GA10248>