G'day cavac,

"Your solution is probably a lot faster, ..."

I've always found for to be faster than map so I benchmarked. I ran the following with: 0 .. 5 (from the two nodes on which you commented); 0 .. 40 (the OP's original range); and, 0 .. 5000 (just to see if a large range made any significant different).

#!/usr/bin/env perl

use strict;
use warnings;

use Benchmark 'cmpthese';

cmpthese 0 => {
    with_for => \&with_for,
    with_map => \&with_map,
};

sub with_for {
    my @f;
    push @f, [$_+2, $_+1] for 0 .. 5;
}

sub with_map {
    my @f = map [$_+2, $_+1], 0 .. 5;
}
[download]

I ran each three times; there was little difference in the results; I've just posted the middle results below.

# 0 .. 5
             Rate with_map with_for
with_map 602818/s       --     -23%
with_for 780657/s      30%       --

# 0 .. 40
             Rate with_map with_for
with_map  92655/s       --     -23%
with_for 120756/s      30%       --

# 0 .. 5000
          Rate with_map with_for
with_map 745/s       --     -23%
with_for 968/s      30%       --
[download]

The smallest difference was -21% vs. 27%; the largest was -25% vs. 33%. As the range was increased, all values tended towards -23% vs. 30% (the middle values for all runs).

In this instance, for is faster than map (i.e. what I've always found in the past).

Thanks for taking the time to benchmark this.

perl -e 'use Crypt::Digest::SHA256 qw[sha256_hex]; print substr(sha256_hex("the Answer To Life, The Universe And Everything"), 6, 2), "\n";'