Martin Schlemmer wrote:

> for instance.  Sure, it may need some improvements, but why don't we
> *rather* work on that, than reinvent the wheel ?
>
I might be a little off-topic, but I have a small suggestion which isn't 
covered by any of the drafts I have seen until now...

I've seen it so many times: The people start or stop a service on a 
*production* machine and they also want to change the boot settings 
accordingly, but of course without booting. Next time when the machine 
boots (maybe months later), it doesn't get up anymore.

Hit list of those fatal changes: typos in configuration, broken 
dependencies, broken links, untested firewall settings, lost kernel 
versions.

One of the small problems I see with the init mechanism as it is now 
that the dependencies are resolved during boot. I would like to see 
something like a test mode, or, the init mechanism could be modified so 
that it the final boot script is generated by compilation.

Only an idea...

Kind regards
k.j.




--
gentoo-dev@g.o mailing list