[Pkg-sysvinit-devel] Bug#714039: initscripts: Require-Start problem

[heroxbd]

Dear Petter,

>[Bill Wang]
>> I think that should be #Should-Start, because #Require-Start: $all
>> will have a serious impact on OpenRC(maybe there are other init
>> systems).

> Can you explain a bit more?  The way I understand $all, it is the set
> of all the available scripts without a relationship on $all, thus the
> dependency will always be available and there is no way it can fail to
> have its dependency available.

According to LSBInitScripts page[1], there is a definition of
Should-Start being weak dependence along with a recommendation to be
used with virtual facility names (e.g. $all). Required-Start states that
a facility *must* be available.

$all only means for start ordering, which is a weak
dependence. Therefore using Should-Start instead of Required-Start is
more logical.

> Is OpenRC understanding it differently?  What problem will OpenRC
> users experience?

In OpenRC we developped a method globbing /etc/init.d/* to construct
$all, which results in /etc/init.d/README to be pulled in. A strong
dependence will fail if README is not a init script, while a weak one
will not.

Considering the logical arguments above, we think the most
straightforward fix would be replacing "Required-Start: $all" with
"Should-Start: $all".

Cheers,
Benda

1. http://wiki.debian.org/LSBInitScripts