For server side, one cares whether each connection can read or write, not the listening server socket. The listening socket does not read or write.

Let's do a simple testing. With the following client and server:

#server
use strict;
use warnings;

use IO::Socket;
use IO::Select;

my $sock = IO::Socket::INET->new(Proto=>"tcp", LocalHost=>"localhost",
+ 
Listen=>16, Reuse=>1, LocalPort=>3000) 
    || die("Could not create socket!\n");
my $conn = $sock->accept();
my $sel = new IO::Select($sock);
if ($sel->can_read()) {
    print "can read\n";
}
#client
use IO::Socket;

my $sock = IO::Socket::INET->new(
                                Proto=>"tcp",
                                PeerHost=>"Localhost",
                                PeerPort=>3000
                                ) or die("Could not create socket!\n")
+;


print $sock "abcd\n";
[download]

Run perl -w server.pl in one window, then run perl -w client.pl in another window, nothing gets printed in the server window! run perl -w client.pl again, "abcd" gets printed. This is not what one will want.

change server code:

use strict;
use warnings;

use IO::Socket;
use IO::Select;

my $sock = IO::Socket::INET->new(Proto=>"tcp", LocalHost=>"localhost",
+ 
Listen=>16, Reuse=>1, LocalPort=>3000) 
    || die("Could not create socket!\n");
my $conn = $sock->accept();
my $sel = new IO::Select($conn);
if ($sel->can_read()) {
    print "can read\n";
}
[download]

The first time you run perl -w client.pl, server side knows that it can read, and prints "abcd". That's what one wants.