Use of rcorder for local rc.d/*.sh scripts

[Doug Barton]

J.R. Oldroyd wrote:
> I posted this to -current yesterday; it was suggested I resend to
> this list.

Thanks for bringing this up, and yes, we prefer discussion here on rc.d topics.

> ----- Forwarded message from "J.R. Oldroyd" <fbsd at opal.com> -----
> 
> From: "J.R. Oldroyd" <fbsd at opal.com>
> Date: Thu, 2 Jun 2005 17:10:23 -0400
> Subject: Use of rcorder for local rc.d/*.sh scripts
> To: freebsd-current at freebsd.org
> 
> Is there a reason that rcorder is not used in /etc/rc.d/localpkg
> so that the scripts there can be ordered using their tags?
> 
> If not, could we make the following change to /etc/rc.d/localpkg:
> 
> --- localpkg.orig	Fri Oct  8 13:52:43 2004
> +++ localpkg	Thu Jun  2 16:53:56 2005
> @@ -36,7 +36,7 @@
>  		done
>  		script_save_sep="$IFS"
>  		IFS="${script_name_sep}"
> -		for script in ${slist}; do
> +		for script in `rcorder -s nostart ${slist} 2>/dev/null`; do

This is an interesting approach. I'm in favor of incorporating rcorder for 
local scripts (whether ports, or homegrown stuff), but I think I prefer an 
approach that brings all the scripts into the fray, so that local scripts 
that need to execute before parts of the base can do so. Have you put any 
thought into such an approach?

> If we do, there may be some scripts that ports install in the
> local/etc/rc.d dirs that need rcorder tags adding, notably those
> that currently use "000.foo.sh" names to ensure they're started
> first.

Yeah, if we're going to do something like this, it would be nice to do it 
before the 6.0 code slush, so if we can at least agree in principle that it 
should be done, and roughly on how, then we can send a message to the -ports 
list with a heads up on this topic. The ports code freeze comes after the 
base, so there will be time to fix ports that are broken by this change.

Doug

-- 

     This .signature sanitized for your protection