Re: rcrun: savecore and swap1 patch

[Peter Kadau <peter.kadau@xxxxxxxxxxxxxxxx>]

Hi !

>     I wouldn't worry about one-offs like savecore.  For something like
>     savecore the only appropriate return values are RC_CONFIGURED or
>     RC_IRRELEVANT (maybe I should rename IRRELEVANT to INAPLICABLE :-)).
Well, I'd opt for RC_CONFIGURED in the RCNG framework as it is now.
Just because irrelevant means "I don't need this at all here", but
if I said savecore in rc.conf I'd be rather confused to see it in state
"irrelevant". And it would be even more confusing for "localswap".

>     This does bring up the fact that we need to differentiate between
>     boot-time ordering issues and run-time dependancies.  
Hmm, I see two different approaches towards that.
First one: Let the rcng scripts that are only relevant to boot
return RC_CONFIGURED or RC_IRRELEVANT as appropriate.
Second one: A new rcorder header word like ESTABLISHES and a new
rcrun state "established" which can't change.

The problem being that those ordering issues (boot vs. runtime)
are kind of orthogonal.

Cheers
Peter

--
<peter.kadau@xxxxxxxxxxxxxxxx>

Campus der Max-Planck-Institute Tübingen
Netzwerk- und Systemadministration

Tel: +49 7071 601 598
Fax: +49 7071 601 616