Thomas Hood <[EMAIL PROTECTED]> writes:
> This is a second call for objections to #156161 as restated
> in my recent posting to this list.
> http://marc.theaimsgroup.com/?l=debian-devel&m=106150585805521&w=2
>
> To recap, the idea is that invoke-rc.d be modified such that
> it _not_change_ the state of a service (i.e., neither start it
> nor stop it) on entering a runlevel if there is neither a start
> link nor a stop link for the service in that runlevel. This
The jobs running after entering a runlevel should allways be the same
irrelevant from where one comes.
Say runlevel 1 starts foo.
Change to runlevel 2 which has no foo scripts, foo keeps running.
Change to runlevel 3 which has a kill script, foo is stoped.
Change to runlevel 2, foo is not running.
The existence of starts scripts alone is sufficient to get the right
behaviour:
old level | new level | affect
----------+-----------+--------
| | no change (if it runs don't kill it)
has start | | stop service
| has start | start service
has start | has start | no change (if it doesn't run, don't start it)
For scripts that just run once like rcS.d/checkroot.sh all runlevels
would need a start script or a new prefix "R" for just run on enter
would be needed.
But I think that behaviour would be most consistent. Note that it
includes yours.
MfG
Goswin
