I often run commands, then verify the return code, stdout and stderr. Generally, I like to put as much logic as I can in support functions (such as run_cmd_capture below), so the test script itself is short and reads easily. Here is a simple example.

use strict;
use warnings;
use Test::More tests => 6;

sub read_file_contents {
    my $fname = shift;
    local $/;
    open(my $fh, '<', $fname) or die "error: open '$fname': $!\n";
    <$fh>;
}

sub run_cmd_capture {
    my $cmd = shift;
    my $tmpfile = "klink-$$.tmp";
    my $outstr = `$cmd 2>$tmpfile`;
    my $rc = $? >> 8;
    my $errstr = read_file_contents($tmpfile);
    unlink($tmpfile) or die "error: unlink '$tmpfile': $!\n";
    return ($rc, $outstr, $errstr);
}

my $outcmd = qq|$^X -le "print 'Klink'"|;
my ($rc, $outstr, $errstr) = run_cmd_capture($outcmd);
cmp_ok( $rc, '==', 0,   "command '$outcmd' rc is zero" );
is( $outstr, "Klink\n", "command '$outcmd' stdout is Klink" );
is( $errstr, "",        "command '$outcmd' stderr is empty" );

my $errcmd = qq|$^X -le "print STDERR 'Klink'"|;
($rc, $outstr, $errstr) = run_cmd_capture($errcmd);
cmp_ok( $rc, '==', 0,   "command '$errcmd' rc is zero" );
is( $outstr, "",        "command '$errcmd' stdout is empty" );
is( $errstr, "Klink\n", "command '$errcmd' stderr is Klink" );
[download]