Bootprocedure again

[Gerhard W. Gruber]

Alexandre Julliard wrote:

> Well, if you discard all objections as "not real" then of course there
> isn't a real objection. But the two mentioned seem very real to

Sorry, I didn't mean them to be not real. It's only that I doubt that
the check for the existence of a registry key and alternatively for a
configurable switch is that performance intensive to hinder such an
implementation. The other one (forcing an entire reboot) is not a real
objection to me as I never wanted to implement it in that way. I don't
like the idea to have all applications to stop just because some program
needs to rename/delete some files for itself. Especially under Unix
where this is no problem anyway. So this was not meant to discard that
objection, rather I never intended this solution anyway. :)

> me. Another one is that you need to be able to control when bootup
> processing happens, and this is a policy decision that doesn't belong
> in the lowest layers. This is why we need a separate app, which can be
> launched when some higher layer decides it's the right time.

I can understand that as a more "real" objection. :) That's why I posted
my first mail, because I wanted to know which reasons could exist to
justify that. I hate to code something without understanding it's needs,
and even more so when I don't agree with it. The disagreement could be
because I don't know about other issues, which I wanted to find out
about here.

So why is it neccessary for this to be in a seperate app and are there
already any plans on how this should have been integrated? Which layer
would that be that decides this? If the decision is done in a higher
app, why not just implement it in a seperate module (which I would have
done anyway) and then execute the code when it is needed?