>>> [...], I could completely configure a system from a bare prototype
>>> by editing *one* file.  (Well, two, [...]).
>> Wow.  I found I need to edit seven files and fix a symlink:
> That depends on how much configuring you want to do before
> considering a system done.

Well, of course.  To consider a system *truly* ready to go, I also need
to set up /home/mouse, including building and installing the stuff in
.local/src/bin, and set up /local - a minimal /local setup involves
building and installing a whole pile of things (c-publish, fsm, halign,
libarc4, libidea, libmd5, libsearchstr, libsha, makefiles, makewrapper,
mcgrep, mcsh, md5, nc, see, sha, skipcat, wait-for...if tight for space
(and depending on what the system will be used for) I may add to or
delete from that list).

> Nevertheless, even with everything on your list it's still better
> than editing `ls /etc/rc.d|wc -l` files and _much_ easier to see what
> is actually going to start.

Oh, certainly.  This sort of thing is what I meant when I wrote, of the
rc.d way, that while it's admittedly easier on programs that want to
frob the startup sequence, it's hell for humans that want to have
anything to do with it.

> I like the rc.d stuff as a way to easily drop in new things to start
> and as a great way to make sure things run when they're supposed to
> without having to worry about the precise order.

That's not the impression I got from that person who was going all over
hell's back acre trying to get the control lines set so that things
would start correctly.  Or that person who made atalkd depend on xdm
merely to get a login prompt without waiting ages (the same person? I
forget).

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B