Subject: Re: more work in rc.d [was Re: rc, rc.shutdown proposed change]
To: None <tech-userlevel@netbsd.org>
From: Jason R Thorpe <thorpej@zembu.com>
List: tech-userlevel
Date: 03/14/2000 08:41:50
On Mon, Mar 13, 2000 at 10:52:03PM -0800, John Nemeth wrote:

 > On Jun 29, 10:53am, "Simon J. Gerraty" wrote:
 > } 
 > } Oh and yes rc*.conf should die.
 > 
 >      I totally disagree.  I like rc.conf.  It keeps as much information
 > in one place as possible, which makes administration really easy and
 > pain free.  It's also more in tune with the BSD way of doing things.

It's also a giant PITA for package management, etc.

I'm a little annoyed that the current rc.d stuff went in.  It's almost like
Luke simply chose to ignore all feedback he received on it.

A quick rundown of what we used to have:

	- monolithic script where ordering of thing was really obvious
	  and guaranteed.

	- a single way of doing things

	- have to edit rc.conf

What we have now:

	- a bunch of scripts with undocumented and not-always-obvious
	  dependencies, ordering is not obvious or guaranteed

	- additional complexity, magick shell functions and variables

	- multiple, incompatible ways of doing things (making the drop-in
	  package problem even harder to solve)

	- still have to edit rc.conf

To me, this is a big, stinking LOSE.

However, I don't really feel inclined to provide suggestions at this point,
considering that all of the feedback I provided before was simply ignored,
as was the feedback provided by many others.  Someone just wanted to steam
right ahead with their pet project, with pressure to get it in the tree
before 1.5 was ready to branch.  Folks, that's NOT the way you do good
engineering.

...and before you argue that the feedback was conflicting with how the
proposed rc.d system was going to work, well then that should have said
something right there... "The proposal is fundamentally broken!"

Don't get me wrong... I *WANT* rc.d, init.d, etc.  But what we have now
is actually WORSE than what we had before.

And, I VERY STRONGLY feel that the situation is bad enough to delay 1.5.
If we ship a release with this, and then change it all over again for the
next one, we're not exactly providing any semblence of consistency.

-- 
        -- Jason R. Thorpe <thorpej@zembu.com>