point taken and taken.

$ ls ../Pidfile
MANIFEST  Makefile.PL   Pidfile-0.0.1.tar.gz  README  foo.pl      t
Makefile  Makefile.old  Pidfile.pm            blib    pm_to_blib
[download]

Pidfile showed promise, but i was trying to be fancy and get by with Pidfile::new returning a simple blessed scalar reference of the pidfilename. i couldn't quite pull it off gracefully =(.

... and furthermore, that top-level names should not be used if you're expecting to put this thing on CPAN and have the world adopt it. (Allowances are made for large, complex systems like application frameworks.)
Be sure to read the appropriate documentation:
• Part 2, section 5, of The Perl 5 Module List
• PAUSE

jdporter
The 6th Rule of Perl Club is -- There is no Rule #6.

knew i should have put a disclaimer in.... woe is it that i can't edit my root node.

Warning: sorta rantish, but only because of the recent threads about Style and Changes in programming habbits.

the original is a patched Proc::Pid::File, i barely dream of CPANing a module, much less with a lower-case-top-level-name.

i'm just too lazy to think of a proper package name when i first need to make sure i'm not borking the whole concept.

my private modules live in namespaces like 'W' and 'Q' because i use them often from one-liners and typing "-MWunderbar::Blat::Bonk -e '$wunderbar = Wunderbar::Blat::Bonk->new or die;..." quickly becomes a pain. i know people who log into a slowish web page with obnoxiously short inactivity timeouts *chuckle* to block an evil-doer when they could just do:

$ blocker --ip <ip> [--case <case>] [--note <note>] [--unblock]

# but this gives me more satisfaction.

$ perl -MBlock -e 'block @ARGV' <ip> <case> <note>
[download]

i usually have to do several ':%s/$scn/$long_descriptive_name/g' before cut-n-pasting a tested section of code here at the monastery.

my daemons still being tested likewise live in /W because i'm to lazy to type /usr/site/app/version/bin/script check. when they're proven for a time and have the bells and whistles to make them usable by anyone but me, then they get big_long_descriptive_to_the_nth_degree_names, get put in rc.local, get their logs rotated, get rdisted ...

now if i could just figure out a good way to document that one script that saves me an hour easily every time i use it and always DWIM, but nobody else can quite get the hang of it...

$ ./cfg d1 I standby 12345678901 foo-6000-1 12 J 15 M12345678V 9876543
+2101 25 baz-6000-1-1 26 baz-6000-2-1 esp foo-6000-2 foo-6000-3
###
# outputs hundreds of lines of configuration commands
# for backend systems that don't have scripting native 
# to their CLI
###

if the others are lucky, they cut-n-paste-n-modify
some snippets:

$ perl -e 'for(1..48){printf"somecmd %s $_ X0000000%02X auto$/","foo-6
+000-1",$_' | backend
[download]

and i can't think of a good way to describe how each argument type is uniqely regexable and a configuration stack is updated when configuration things are given, and actions take place at the last possible moment using the top configuration stack entry when triggered by the next action thing given... but i still like that script more than the ones that give screens of --help information and get used by everybody.

anyways, that was an early era script, and my early era daemons used a ugly hack for quick and dirty daemonization with frills.

use lib qw( /Q/perl );

package Watcher;
use Net::Server;
our (@ISA);
@ISA = qw/Net::Server/;

Watcher->run(
        log_level => 5,
        log_file => 'watcher.log',
        pid_file => 'watcher.pid',
        background => 1,
);

sub post_configure_hook {
  my $s = shift;
  $SIG{INT} = $SIG{TERM} = $SIG{QUIT} = sub { $s->server_close; };
  $SIG{HUP} = sub { $s->sig_hup; };

  my $done = 0;
  while (!done) {
    # work
  }
  $s->server_close;
}
[download]

worked ok, sorta ugly.

the latest version of this uses Config::General for general configuration (overrideable with Getopt::Long command line options), Log::Log4perl (modifying logging while daemon is running is so cool.), Proc::Daemon (but i want to have the parent hang around long enough to verify the daemon is alive, so i'll have to re-wheel or maybe just stick another fork in there somewhere), and a patched Proc::PID::File.

Proc::PID::File gets replaced. by something that DWIM with as short a name as i can manage (not as important when it's stable and you only have to type it once per script), and last but not least Pod2Usage for the docs.

so, review 'package pidfile;' =~ s/pidfile/Some::Longer::Name/; so i can ask about 'package Proc::Daemon::Wrangler::ThingAMaBob;' =p