Okay so I am an ass yet again. I did try it on my system before I rambled and the tr did not work, so I must have had a dumb little typo.

Just for kicks I benchmarked the s/// vs tr/// since they seem to do the same in this reguard:

use Benchmark;
$str = "\n#\n#\nSTUFF\n#\n#\n";
timethese(1000000, {
        's' => sub { $a = $str; $a=~s/\n/ /g; },
        'tr' => sub { $a = $str; $a=~tr/\n/ /; }
});

RESULTS:

Benchmark: timing 1000000 iterations of s, tr...
         s: 11 wallclock secs (11.22 usr +  0.00 sys = 11.22 CPU)
        tr:  3 wallclock secs ( 3.61 usr +  0.00 sys =  3.61 CPU)
[download]

The moral of the story: Use tr/// where you can.