use Socket;
...
setsockopt(SOCK, SOL_SOCKET, SO_REUSEADDR) or
    warn "OOPS, setsockopt failed $!\n";
[download]

I used to solve this problem in C using setsockopt(2), with the option SO_REUSEADDR. It basically tells the system that it's OK to bind to a port as long as no one is listening to the same port.

In Perl you can say:

setsockopt(YOURSOCKET, SOL_SOCKET, SO_REUSEADDR, 1) or die "your messa
+ge\n";
[download]

Okay, thats a fair enough answer. So my question is this, not to sound daft or anything:

How do I setockopt() on a SOCKET when I can't generate the SOCKET because it won't bind to the port I'm telling it to?
Clarify:

my $server = IO::Socket::INET->new(....);
setsockopt($server, SOL_SOCKET, SO_REUSEADDR, 1);
[download]

$server doesn't exist until after the bind, but the bind won't happen (and perl will stop) at the ->new() because the port is already "taken"...
I'm quite sure I'm so far off base you're laughing - which is fine - as long as you can tell me how its _supposed_ to work :)

Thanks,
JP

-- Alexander Widdlemouse undid his bellybutton and his bum dropped off --

setsockopt can be used immediately after calling socket and before bind, if using the functional socket interface.

With perldoc IO::Socket you can easily find out that all you need to do is passing the Reuse option to your constructor:

my $server = IO::Socket::INET->new(...., Reuse => 1);
[download]

Happy recycling!

-- TMTOWTDI