Whenever I am thinking about columns and fixed-length data files I think of pack and unpack. I not only find it easier to use but, generally a little faster than printf.

print MERGE pack('A20A180A200A350',@vars);
[download]

as for the faster part (I thought it would be a bigger difference):

#!/usr/local/bin/perl

use Benchmark;
use strict;

my $count = 10000;
my @lists = map { [chr($_+65),chr($_+66),chr($_+67),chr($_+68) ] } 0..
+22;

open(FH1,'>/tmp/x') or die "$!\n";
open(FH2,'>/tmp/y') or die "$!\n";

timethese ($count, { 'pack' => '&use_pack', 'printf' => '&use_printf' 
+} );

sub use_pack {
    foreach (@lists) {
        print FH1 pack('A20A180A200A350',@{$_});
    }
}

sub use_printf {
    foreach (@lists) {
        printf FH2 "%-20s%-180s%-200s%-350s", @{$_} ;
    }

}
[download]

Benchmark: timing 10000 iterations of pack, printf...
      pack: 21 wallclock secs (10.10 usr +  9.55 sys = 19.65 CPU) @ 50
+8.91/s (n=10000)
    printf: 24 wallclock secs (10.90 usr +  9.72 sys = 20.62 CPU) @ 48
+4.97/s (n=10000)

Benchmark: timing 10000 iterations of pack, printf...
      pack: 21 wallclock secs (10.03 usr +  9.75 sys = 19.78 CPU) @ 50
+5.56/s (n=10000)
    printf: 23 wallclock secs (11.88 usr + 10.72 sys = 22.60 CPU) @ 44
+2.48/s (n=10000)
[download]

grep

Mynd you, mønk bites Kan be pretti nasti...