Suggestion for default arch logging behavior

Question

Okay let me throw this out there and see what comes back...
Every Linux distribution I have used (apart from arch) has a boot log file enabled by default, or provides a simple Yes/No flag to turn the feature on.
Arch does not.  Moreover, I have searched the forums and posted a question or two myself about the topic.  No one seems to know the answer, and a lot of responses seem to indicate that, amongst the arch community, the possibility of a boot log file is treated like: a) something that no one in their right mind would want, or b) something that maybe, might be useful, but that no one can figure out how to implement.
What is this?  What is going on?  Arch has a bootlogd binary in /sbin/.  Presumably there is no reason why it won't actually work, but no one seems to know where to put the call to the binary, or why they would put it there.
This seems like an easy-to-add feature that a developer or moderator (anyone who really knows the system layout well) could make the default for future installs.  It just makes a text file that is a) not large, and b) can be very useful at times.
So how about it?  Why not make this part of the default arch logging set up?  (and of course explain what you did that worked.)

Answer 1

tomk:
Sorry to disclose my ignorance, but I'm not sure how to do that.
What I have been doing though is trying to see in more detail how it is done on a debian system which I have access to.  Here is what I've learned so far:
There are not that many references to bootlogd on the system, so it might be possible (for me) to track down what is going on:
root@wave32p:/etc# locate bootlogd
/etc/default/bootlogd
/etc/init.d/bootlogd
/etc/init.d/stop-bootlogd
/etc/init.d/stop-bootlogd-single
/sbin/bootlogd
/usr/share/man/man8/bootlogd.8.gz
1) "/etc/default/bootlogd" must be edited (trivially) such that a "No" becomes a "Yes"  -- this seems like just a master switch.
2) the man files are exactly the same on the two systems, and the output of "/sbin/bootlogd -v" is the same on the two systems, however the size of the bootlogd binary itself is not the same on both systems (larger on debian64 system).  not sure what to make of that, but it is not what I was hoping to see.
3) the script "/etc/init.d/bootlogd" runs with (eg.) a "start/stop" flag, the same as most "functions" under arch that have scripts associated with them.
4) it would seem that I have to grind my way through the above script if I'm going to make any progress.  I'm doing that in my spare time at the moment, though it's a challenge since it's been a few years since I've written bash scripts on a regular basis.  FYI, here is the /etc/init.d/bootlogd script verbatim (additional note:  the option -r (below) is supported on the arch version of the bootlogd binary, but the -c option does not seem to be...interesting?  Here is the man page entry for -c: "Attempt  to  write to the logfile even if it does not yet exist.  Without this option, bootlogd will wait for the logfile to appear before attempting to write to it.  This behavior prevents
              bootlogd from creating logfiles under mount points."):
#! /bin/sh
### BEGIN INIT INFO
# Provides: bootlogd
# Required-Start: mountdevsubfs
# X-Start-Before: hostname keymap keyboard-setup procps pcmcia hwclock hwclockfirst hdparm hibernate-clean
# Required-Stop:
# Default-Start: S
# Default-Stop:
# Short-Description: Start or stop bootlogd.
# Description: Starts or stops the bootlogd log program
# which logs boot messages.
### END INIT INFO
PATH=/sbin:/bin # No remote fs at start
DAEMON=/sbin/bootlogd
[ -x "$DAEMON" ] || exit 0
NAME=bootlogd
DESC="boot logger"
BOOTLOGD_OPTS="-r -c"
[ -r /etc/default/bootlogd ] && . /etc/default/bootlogd
. /lib/init/vars.sh
. /lib/lsb/init-functions
# Because bootlogd is broken on some systems, we take the special measure
# of requiring it to be enabled by setting an environment variable.
case "$BOOTLOGD_ENABLE" in
[Nn]*)
exit 0
esac
# Previously this script was symlinked as "stop-bootlogd" which, when run
# with the "start" argument, should stop bootlogd. Now stop-bootlogd is
# a distinct script, but for backward compatibility this script continues
# to implement the old behavior.
SCRIPTNAME=${0##*/}
SCRIPTNAME=${SCRIPTNAME#[SK]??}
ACTION="$1"
case "$0" in
*stop-bootlog*)
[ "$ACTION" = start ] && ACTION=stop
esac
case "$ACTION" in
start)
# PATH is set above
log_daemon_msg "Starting $DESC" "$NAME"
if [ -d /proc/1/. ]
then
umask 027
start-stop-daemon --start --quiet --exec $DAEMON -- \
$BOOTLOGD_OPTS
ES=$?
else
$DAEMON $BOOTLOGD_OPTS
ES=$?
fi
log_end_msg $ES
stop)
PATH=/bin:/sbin:/usr/bin:/usr/sbin
log_daemon_msg "Stopping $DESC" "$NAME"
start-stop-daemon --oknodo --stop --quiet --exec $DAEMON
ES=$?
sleep 1
log_end_msg $ES
if [ -f /var/log/boot ] && [ -f /var/log/boot~ ]
then
[ "$VERBOSE" = no ] || log_action_begin_msg "Moving boot log file"
# bootlogd writes to boot, making backup at boot~
cd /var/log && {
chgrp adm boot || :
savelog -q -p -c 5 boot \
&& mv boot.0 boot \
&& mv boot~ boot.0
ES=$?
[ "$VERBOSE" = no ] || log_action_end_msg $ES
fi
restart|force-reload)
/etc/init.d/bootlogd stop
/etc/init.d/bootlogd start
status)
status_of_proc "$DAEMON" "$NAME" && exit 0 || exit $?
echo "Usage: $SCRIPTNAME {start|stop|restart|force-reload|status}" >&2
exit 3
esac