Re: cvs commit: src/etc rc rc.shutdown rc.subr src/etc/rc.d localpkg src/sys/sys param.h

Yes, this is an old mail I am replying to ;-)

On Fri, 02-Dec-2005 at 18:42:37 -0800, Doug Barton wrote:
> Andrey Chernov wrote:
> > On Fri, Dec 02, 2005 at 06:16:48PM -0800, Doug Barton wrote:
> >>> Do you mean that scripts without .sh runs in 
> >>> the subshell and not damage main shell? 
> >> Yes, that's what I mean. Once again, sorry I wasn't clear. I've been staring
> >> at this for too long now. :)
> > 
> > Just to clarify it finally. You state that there is a big difference 
> > between system /etc/rc.d scripts (without .sh) which all runs in the 
> > single shell 
> 
> No. The way things stand now, all scripts named foo.sh are sourced into the
> main shell, and everything else (base scripts, local scripts, whatever) are

Today I stumbled over the fact that this statement only
applies to .sh scripts in /etc/rc.d but not to the ones
in /usr/local/etc/rc.d.

I need to source a script.sh and I put it into /usr/local/etc/rc.d
-- only to recognise that it will be run in a subshell.

This patch probably would fix this but maybe there is a
reason not to do so:

--- rc.subr.ORI 2011-11-14 17:18:39.000000000 +0100
+++ rc.subr     2011-11-14 17:19:15.000000000 +0100
_at__at_ -938,7 +938,7 _at__at_
        eval unset ${_arg}_cmd ${_arg}_precmd ${_arg}_postcmd
 
        case "$_file" in
-       /etc/rc.d/*.sh)                 # run in current shell
+       /etc/rc.d/*.sh | /usr/local/etc/rc.d/*.sh)      # run in current shell
                set $_arg; . $_file
                ;;
        *[~#]|*.OLD|*.bak|*.orig|*,v)   # scratch file; skip


Or even use *.sh) 

Thanks,

	-Andre




> all run in subshells. You could answer this for yourself by looking in
> /etc/rc.subr if it's still not clear.
> 
> >> I should have mentioned in my last message that I did take a quick look at
> >> the script itself, and didn't see anything that should be a problem, but as
> > 
> > /usr/bin/limits is the problem there because it change limits for whole 
> > shell, not for command which just invoked. If all scripts runs in the same 
> > shell, all subsequential of them will be affected.
> 
> Assuming you're right about that, then you should do something like what I
> suggested in the patch I sent so that the script gets installed as apache
> instead of apache.sh.
> 
> hth,
> 
> Doug
> 

-- 
Failure is not an option -- it comes bundled with Windows.