I agree with Laurent. Using both is unnecessary.

Here's an adjusted version which only uses a hash and accounts for possible missing line numbers, which prevents the uninitialized warnings.

use 5.010;
use strict;
use warnings;
use List::Util qw( max );

my %hash = (
  4 => 'The future is unknown.',
  1 => 'The day before yesterday sucked.',
  5 => 'Today is a good day.',
  3 => 'Not much hope for tomorrow.',
);

my $last = max keys %hash;

for my $line ( 1 .. $last ) {
    say exists $hash{$line} ? $hash{$line} : '';
}
[download]

If, as I suspect, the OP does not want the numbers to print:

use strict;
use warnings;
my (%hash);
while (<>) {
  chomp;
  my ($text,$line) = $_ =~ m/(.+)(\d+)$/;
  $hash{$line} = $text;
}
print $hash{$_} . "\n" for (sort keys %hash)
[download]

Since I use only the keys, no worries about missing line numbers.

Usage:

perl pm1.pl < input.txt > output.txt
[download]

Update: I along with other respondents missed the fact that there could potentially be non-sequential line numbers in the data, and there might need to be blank lines created accordingly. Updated again to be a bit more perlish.

use strict;
use warnings;
my (@array);
while (<>) {
  $array[$2] = 1 if ($_ =~ m/(.+)(\d+)$/)
}
print defined $_ ? $_ . "\n" : "\n" for @array[1..$#array];
[download]

Dum Spiro Spero

But even if such is the case, still, there is no need for duplicating the data.

My view is that using an array to store the data is the simplest choice. But I would still add code to prevent spurious warnings for undefined values, for example something like:

for my $i (0..$#array) {
    print defined $array[$i] ? "$array[$i]\n" : "\n";
}
[download]

for my $i (0..$max_hash_key) {
    print defined $hash{$i} ? "$hash{$i}\n" : "\n";
}
[download]