LSB-ish init script, RFC

[Matthew Farrellee]

Bill Nottingham wrote:
> Matthew Farrellee (mfarrellee+fedora at redhat.com) said: 
>> It can happen whenever a user process shares its name with a service's 
>> process. This just tends to be a common thing with my package.
> 
> Why not generate patches for the functions to support the behavior you
> want, and we can see what issues they may cause on a global level? That
> seems preferable to re-implementing large portions of it.

I can do that. The change is primarily to /etc/init.d/functions:status
to check a pidfile before using pidof. Right now status() uses pidof and
only failing that looks for a pidfile, which probably means if you ever
get to the pidfile check you're going to fall into the "process dead but
pid file exists" state.

This change will only eliminate the need for the pid_status function I
wrote, which is small. It does not eliminate the other changes to be
more LSB-ish.

Best,


matt