On Tue, 4 Apr 2000, Jay Maynard wrote:

> Put simply, do we want the system to join the 90s? Or do we want to
> keep looing like v7? People are using BSD to provide real production
> services to real people, and in that kind of environment, hardening
> the system against admin screwups is a Seriously Good Thing.

What kind of argument is this? If your system screws up, there's always
single user mode. Frankly, I'm tired of all these systems of initalization
that think modularity and abstraction are ways of achieving godliness,
because they aren't. Sure, isn't it fun to start and stop processes using
cute little scripts? But don't we have pid files?

I think a call for any system to join the 90s is kind of dubious,
especially for NetBSD, since on a system based upon correctness, I find
this to be a very messy solution.

If I wanted a system designed for the 90s, I'd use Linux or NT, which are
both rather recent additions. But I'm not.

> Fine. Just don't make it the standard config. The split rc.d is a step
> - or sefveral steps - forward.

Forward? To what? To runlevels? You're scaring me.

> You may consider this a strong vote for making and keeping the system
> as maintainable and robust as possible, and rc.d is the Right Way to
> get there.

Ow! I've been hit by the "Right Way" arrow!

Besides your original idea that "the existence of admin screwups requires
the split rc.d implementation" I haven't heard any more reasons why this
new system makes a system more maintainable /or/ more robust. If a vital
script doesn't run in the grand sequence, you're still left with a
crippled box, no matter what system you use.

I'm shrugging my shoulders.

How are we going to do this cleanly?

--
Christopher Kyin-hwa Chen <cchen@nougat.org>
<http://www.nougat.org/~cchen/>
"2:00, 23 March 2000. Nope, no food in the house.
Even the ice tray is empty, and that's automatic."