Information pertaining to particular systems as reported by users and
not checked by me:

=====================

DIGITAL UNIX

Bryan T. Schmidt <bts@profitool.com> reports:

Also, for your information, the FIFO method of getting the log
information does not seem to work correctly under Digital Unix.  This
is not suprising, but it might not be a bad thing to warn about in
your INSTALL file.  The perl method used to read the FIFO returns "no
such file or address", even though the FIFO special file clearly
exists.  Digital is just broken.  (What else is new, eh?).

=====================

FreeBSD SYSLOG

Sigi <sigi@mail2me.de> reports:

>In my syslog.conf I have tried the "... | >/var/smtp-poplock/fifo". I
>don't know if this works. If I put this in an file, then this writes the
>lines in a file.

On FreeBSD this works - that you get an answer   ;-)

/etc/syslog.conf
!popper
*.*       | > /var/smtp-poplock/fifo

=====================

FreeBSD 4.2-STABLE.

Sebastiaan van Erk <sebster@sebster.com> reports:

In your man page you wondered about how to setup syslogd, that nobody
confirmed the working of the pipe configuration.

!qmail-poppw
*.*                                             |exec /bin/cat > /var/qmail/poplock/fifo

This works fine for me.

=====================

VMAILMGR and COURIER-IMAP

Matthew Scott <mscott@omaja.net> reports:

We've successfully integrated smtp-poplock integrated it into our
qmail+vmailmgr+courier-imap setup without much trouble.

We did have to make a new $parse_log_sub that was slightly different
from the default pattern matching. Here is what is working for us:

$parse_log_sub =
sub {
  my $line = shift;
  $line =~ m/(?:pop3|imap).*\Wlogin.*::ffff:(\d+\.\d+.\d+.\d+)\]/i
    && return $1;
  return undef;
};

=====================

???????????????? and ?????????????

Justin R. Miller <justin@voxel.net> reports:

We were experiencing odd dying problems with readlog.  We finally narrowed
it down to a weekly restart of syslogd.  We were using the fifo method,
and for now-obvious reasons, readlog would die when this fifo was cut.
Now we just tail the POP3/IMAP login log directly and use logrotate's
"copytruncate" option and I think it's fixed.  You might want to mention
that in the docs -- that syslog restarting will harm your fifo.