IIRC, the correct way to finish a multithreaded program, depends on what the program is doing ...

If your sample code was written like this

sub Main {
    my( $msglimit ) = @_;
    use threads::Q;
    my $qout = threads::Q->new( 50 );
    threads->create( \&message_sender, $qout )->detach();    ## can't 
+join me :)
    my $hear = threads->create( \&message_receiver, $qout, $msglimit )
+;
    my( @ret ) = $hear->join; ## blocking
    print "Bye join @ret\n";
}
[download]

Then its easy to see that the correct way to finish a multithreaded program, is to join all the threads you wanted to wait for, save_state/print_report/flush_logs, and then exit

If you have finished threads, that aren't joined, that means they were taking up memory waiting to be joined, and you simply discarded the results

If you have threads waiting for jobs to process , threads blocking on a queue, these should probably be detached

Threads are like global objects or singletons, at the end of the program they cease to exist, so you have to decide what you're saving, what buffers you're flushing ...

Hi