why not have rcvars for all rc.d scripts? -- and downinterfaces and bootconf.sh not complete rc.d sc

(I am currently writing a chapter about the NetBSD startup
and rc.d 
configurations...)

1) Why not have rcvars for all rc.d scripts?

Several scripts can't be disabled (i.e. they don't have any
rcvar to 
enable): bootconf.sh, downinterfaces, fsck, local, mountall,

mountcritlocal, mountcritremote, network, ppp, pwcheck,
raidframeparity, 
root, securelevel, staticroute, swap1, swap2, sysdb (on
older systems), 
sysctl, and ttys.

If I am overlooking something about that please let me
know.

Should we consider offering an rcvar and setting to YES for
all or some of 
these?

What were the arguments against this?

2) downinterfaces and bootconf.sh not complete rc.d scripts

I noticed that bootconf.sh and downinterfaces aren't real
rc.d scripts.
bootconf.sh doesn't use rc.subr and downinterfaces only uses
rc.subr when 
ran with the "stop" argument.

Is that okay?

  Jeremy C. Reed

p.s. And why does ttys have a different name (ttyflags)?

> 1) Why not have rcvars for all rc.d scripts?

> Several scripts can't be disabled (i.e. they don't have
any rcvar to
> enable): [...]

Probably someone thought they should always run, and nobody
objected.
(It wasn't until I squawked that fsck_flags appeared, I
think.)

Many of the ones you list have other controls that are
functionally
equivalent; for example, a missing sysctl.conf effectively
disables
sysctl.  (I am not defending the current paradigm, just
trying to
describe it; in fact, I rather dislike the way it pushes
configuration
information out into a bunch of little files instead of
keeping it in
rc.conf - the aspect of that that bothers me most,
personally, is
network interface configuration.)

/~ The ASCII				der Mouse
 / Ribbon Campaign
 X  Against HTML	       mouserodents.montreal.qc.ca
/  Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3
27 4B

this dredges up some old memories 
   
   1) Why not have rcvars for all rc.d scripts?
   
   Several scripts can't be disabled (i.e. they don't have
any rcvar to 
   enable): bootconf.sh, downinterfaces, fsck, local,
mountall, 
   mountcritlocal, mountcritremote, network, ppp, pwcheck,
raidframeparity, 
   root, securelevel, staticroute, swap1, swap2, sysdb (on
older systems), 
   sysctl, and ttys.
   
   If I am overlooking something about that please let me
know.

   Should we consider offering an rcvar and setting to YES
for all or some of 
   these?

probably.  having some like mountall or root turned off may
be
confusing to people who do that accidentally.  (also
include
mountcritlocal, mountcritremote and sysdb in this list..)
   
   What were the arguments against this?

were?  when i split up /etc/rc into the scripts lukem
turned
into what first appeared in netbsd, there were only a few
rc.conf variables in use, and most of /etc/rc ran.  so for
a lot of them, i did not create rc.conf variables.

   2) downinterfaces and bootconf.sh not complete rc.d
scripts
   
   I noticed that bootconf.sh and downinterfaces aren't real
rc.d scripts.
   bootconf.sh doesn't use rc.subr and downinterfaces only
uses rc.subr when 
   ran with the "stop" argument.

these seem broken.  they should be fixed.

   p.s. And why does ttys have a different name (ttyflags)?

it sets up ttys, not just runs ttyflags?


.mrg.