NetBSD /etc/rc.d/ port for OpenBSD

http://orbital.wiretapped.net/cgi-bin/cvsweb.cgi/rcd/

I love the simplicity of BSD's rc.boot/rc/rc.local system. It's clean and obvious, and it leaves room for administrator customization.

By contrast, the rc.d system is a bitch to customize or maintain, so much in fact that I despise Linux for that (although Red Hat has an rc.local - or whatever it's called - but Debian definitely does not).

Hearing that some other BSDs had started to adopt that system worried me but, for Christ's sake, DON'T mess up OpenBSD with this stuff! Hopefully, Theo will refuse to adopt it for the real distribution...

I think it's a good idea, and if the core doesn't like it, or others, then would be kew to have it for those who do like and want it!


Would be something kewl to see in /usr/ports..

I have been reading the first posts... I think I understand why the core do not want it. And can understand why others would. I myself do not really care which way OpenBSD goes. But I am sure what ever the core decides, will be best. I trust them the most in these matters.

But lets not have another ipfilter, netfilter, etc, etc, debate. That got really silly towards the end.

As far as I have read at the misc@openbsd
the OpenBSD developpers don't want to have the
rc.d structure. I won't argue with that ..
but if someone wants he can have the scripts
and change the system on his own.
Choosing your way and doing it your way
is one of the good things about Open Source
Software, but if the system with the patches
doesn't word don't blame the developers ..
after all they gave us a clean robust system
that DOES work!

Aristotelis

While some folks accept that change must happen and that time moves on, others seem intent on resisting and sticking with what they know and love because they're afraid to change their ways or learn new things.

If you do not understand how BAD the current rc system is (with /etc/rc and /etc/rc.local) then I doubt you would be in a position to criticize the rc.d architecture currently in use on NetBSD. It is totally unlike that used by SVR4 or even Linux.

IMHO, the prize for the best rc.d still goes to HP-UX, where tunable parameters for rc scripts live in a different place to the scripts themselves.

So how is all this of benefit you ask ?

If you install a port and it wants to add a daemon, how does it do it in a seemless fashion? What about when you remove that port? Can you ever gaurantee that any scripted editing of a file such as /etc/rc.local will work, especially if it depends on other things starting first? And so on.

The old BSD rc system has served its time, now it's time to move on. Maybe the crusty old core developers need to be replaced with more progressive thinkers or will OpenBSD become "OldBSD"? Of course, it may just be that the NetBSD rc.d will be rejected because it came from NetBSD...but there are moves afoot for it to get into FreeBSD too, so hopefully the dreaded NIH syndrome won't apply.

It seems that all those who scream with anger right now, do not know what they are talking about ... It happens all the time ... The /etc/rc.d that was proposed is NOT like the one that Linux and System V style systems are using ... The core has rejected it, that's ok so if we like OBSD we go with the core, but folks see what's everything about first and then start bitching ...

After reading both side´s comments I feel I am at a lost, I thought NetBSD had adopted SVR4 rc.d

Could anyone explain:

• how is it different to BSD rc?
• how is it different to SVR4 rd.d?
• why this rc.d is better/worst than the "traditional" BSD rc?

I would really appreciate