in reply to Re: Sorting Hash By Large Number Values
in thread Sorting Hash By Large Number Values

The "Any" type is probably more natural for Heap::Simple here. The code then becomes something like:
#!/usr/bin/perl -w use strict; use Heap::Simple; my %Stats = map { $_ => int rand 1000 } "AA".."FF"; my $NumOfPortsToGet = 10; my $heap = Heap::Simple->new (order => '<', elements => 'Any', max_count => $NumOfPortsToGet, dirty => 1, ); for my $key (keys %Stats ) { $heap->key_insert($Stats{$key}, $key); } print "$_ => $Stats{$_}\n" while defined($_ = $heap->extract_first);
If you have the perl version of Heap::Simple, this will be slower than simple non-heap solutions upto a max_count of up to about 30 or so. After that the heap will win (measured without accounting for the time of the first new())

Heap::Simple::Perl does partial evaluation with the arguments to new() to build a (relatively) efficient set of methods for exactly the type of heap you want to use. That makes the (first) new() slower, but actual runtime faster than you might expect.

If you have the XS version of Heap::Simple it will beat the naive version for almost any max_count.

And I think its easier to understand in all cases.