Awesome, sorted with one minor adjustment to your code; I had to dereference the conditions of the sort because it was sorting by the scalar reference address. I changed it to:

foreach my $key (
    sort {
        ${$self->{_counts}{$a}} <=> ${$self->{_counts}{$b}}
    } keys %{ $self->{_counts} }
[download]

Thanks so much for the help.

Try this:

foreach my $key (
    sort {
        $self->{_counts}->{$a} <=> $self->{_counts}->{$b}
    } keys %{ $self->{_counts} }
[download]

If you do $self->{_counts}{$b}, Perl thinks $self->{_counts} is a real Hash, not an HashRef. If you use the arrow to dereference, it is easier to read.

Take my advice. I don't use it anyway.

-- greengaroo

What you say is wrong. Quoting Arrow Rule in perlreftut:

In between two subscripts, the arrow is optional.

That code collides with your explanation of the data structure in 987336. Only one can be correct.

In the constructor I see you initialise the field _counts with a hash reference. Where do you now get a scalar reference from?

It would help if you dump $self before sorting, so we can see what the data structure really looks like.

use Data::Dumper; print Dumper $self;
[download]

You're absolutely right and I've found the problem.

I made a mistake whenever I was creating/adding to values in the hash. I had this line:

${$self->{_counts}{$key}} += $value;
[download]

This was where the scalar ref was coming from. I should note, key and value are parameter passed in, representing the obvious. This line explains why I had to do some extra, dodgy dereferencing later on and, of course, should have read:

$self->{_counts}{$key} += $value;
[download]

All seems to be sorted now. Code looks a lot tidier too. First time I've really used references much but it's been a good learning experience.

Thanks to all for the help and suggestions.

Thanks! It looks kind similar to what I have. Maybe I'm screwing up somewhere with the dereferencing. I'll give that a shot.