Sort multi-dimensional array based on dereferenced value

lbmp has asked for the wisdom of the Perl Monks concerning the following question:

Esteemed Monks, I would like to sort a multi-dimensional array (LoL), based on the value of a particular subscript in the second-level array.

Here is some test code:

#!/usr/bin/perl

my $aref = [[(1..10)],
            [(30..40)],
            [qw(the quick brown fox)],
            [qw(jumps over the lazy dog)]
    ];

for (@$aref) {
    print "$_: @$_\n";
}

print "\n";

# print some values, just to test whether dereferencing is correct
print ${$$aref[0]}[0], "\n";
print $aref->[0]->[0], "\n";
print $aref->[1]->[0], "\n";
print $aref->[2]->[0], "\n";
print $aref->[3]->[0], "\n";

print "\n";

for (sort { $aref->[$a]->[0] cmp $aref->[$b]->[0] } @$aref) {
    print "$_: @$_\n";
}
[download]

This returns:


ARRAY(0x8153c28): 1 2 3 4 5 6 7 8 9 10
ARRAY(0x8154630): 30 31 32 33 34 35 36 37 38 39 40
ARRAY(0x81548a0): the quick brown fox
ARRAY(0x8154a50): jumps over the lazy dog

1
1
30
the
jumps

Out of memory!
[download]

An easy work-around (that I can think of) would be to use an intermediate hash (turning the structure into a HoL), then sorting on the hash keys. But I would like to understand why the above code does not work.

Thanks!

-lbmp

Comment on Sort multi-dimensional array based on dereferenced value - out of memory error Select or Download Code

Replies are listed 'Best First'.
Re: Sort multi-dimensional array based on dereferenced value - out of memory error by moritz (Cardinal) on Sep 05, 2011 at 14:44 UTC
The problem is that `$a` and `$b` inside the sort block are array references (and not indexes to them), and you use them as indexes, so as numbers. When you use a reference as a number, it evaluates to the memory address of the referencee. This address can be a rather large number, and your code autovivifies an array big enough to hold that many elements. `$ perl -wE 'say 0 + []' 10034664` [download] So it tries to create an array with about 10Mio elements on my machine. The fix is to use just `$a` in place of `$aref->[$a]` in your code Perl 6 - second systems done right	[reply] [d/l] [select]
Re^2: Sort multi-dimensional array based on dereferenced value - out of memory error by lbmp (Initiate) on Sep 05, 2011 at 15:32 UTC
Aha! Thanks to both of you (Anonymous Monk and moritz) for the clear explanation. Indeed, it works fine when not using $a and $b as indexes. Additionally, I noticed that when running with -w, perl tells me exactly what I was doing wrong: $ perl -w foo.pl Use of reference "ARRAY(0x8153c28)" as array index at foo.pl line 21. Out of memory! Lesson learned!	[reply]
Re: Sort multi-dimensional array based on dereferenced value - out of memory error by Anonymous Monk on Sep 05, 2011 at 14:41 UTC
$a and $b are references not indexes, when treated as a number, they're a big number, use `$a->[0] cmp $b->[0]` [download]	[reply] [d/l]