in reply to Re: How to create a compact data structure?
in thread How to create a compact data structure?

Inspired by some of the other replies to the OP (especially samtregar's reply mentioning Devel::Size) here is a comparison of some of the techniques suggested:

use warnings; use strict; use Devel::Size qw(total_size); use warnings; use strict; my @large_list = ( {key => 1, cond => 1}, {key => 2, cond => 0}, {key => 3, cond => 0}, {key => 3, cond => 0}, {key => 1, cond => 0}, ); my $recordU = bless {}; my $recordA = bless {}; my $recordI = bless {}; my %record; print "Empty sizes:\n"; print 'Packed: ', total_size ($recordU), "\n"; print 'Array: ', total_size ($recordA), "\n"; print 'int: ', total_size ($recordI), "\n"; print 'Hash: ', total_size (\%record), "\n"; for my $item (@large_list) { $recordU->addU ($item); $recordA->addA ($item); $recordI->addI ($item); $record{$item->{key}}{count}++; $record{$item->{key}}{flag} ||= $item->{cond}; } print "\nPopulated sizes:\n"; print 'Packed: ', total_size ($recordU), "\n"; print 'Array: ', total_size ($recordA), "\n"; print 'int: ', total_size ($recordI), "\n"; print 'Hash: ', total_size (\%record), "\n"; for (sort keys %$recordU) { print "Item $_ count ", $recordU->getCountU ($_), ', flag ', $reco +rdU->getFlagU ($_), "\n"; print "Item $_ count ", $recordA->getCountA ($_), ', flag ', $reco +rdA->getFlagA ($_), "\n"; print "Item $_ count ", $recordI->getCountI ($_), ', flag ', $reco +rdI->getFlagI ($_), "\n"; print "Item $_ count ", $record{$_}{count}, ', flag ', $record{$_} +{flag}, "\n"; }
sub addU { my ($self, $item) = @_; my $key = $item->{key}; if (! exists $self->{$key}) { $self->{$key} = pack ('NN', 1, $item->{cond}); } else { my ($count, $flag) = unpack ('NN', $self->{$key}); $self->{$key} = pack ('NN', $count + 1, $flag || $item->{cond} +); } } sub getFlagU { return (unpack ('NN', $_[0]->{$_[1]}))[1]; } sub getCountU { return (unpack ('NN', $_[0]->{$_[1]}))[0]; } sub addA { my ($self, $item) = @_; my $key = $item->{key}; if (! exists $self->{$key}) { $self->{$key} = [1, $item->{cond}] } else { $self->{$key}[0]++; $self->{$key}[1] ||= $item->{cond}; } } sub getFlagA { return $_[0]->{$_[1]}[1]; } sub getCountA { return $_[0]->{$_[1]}[0]; } sub addI { my ($self, $item) = @_; my $key = $item->{key}; if (! exists $self->{$key}) { $self->{$key} = 2 | !!$item->{cond}; } else { $self->{$key} += 2; $self->{$key} |= !!$item->{cond}; } } sub getFlagI { return int ($_[0]->{$_[1]} & 1); } sub getCountI { return int ($_[0]->{$_[1]} / 2); }

Prints:

Empty sizes: Packed: 92 Array: 92 int: 92 Hash: 92 Populated sizes: Packed: 260 Array: 497 int: 209 Hash: 682 Item 1 count 2, flag 1 Item 1 count 2, flag 1 Item 1 count 2, flag 1 Item 1 count 2, flag 1 Item 2 count 1, flag 0 Item 2 count 1, flag 0 Item 2 count 1, flag 0 Item 2 count 1, flag 0 Item 3 count 2, flag 0 Item 3 count 2, flag 0 Item 3 count 2, flag 0 Item 3 count 2, flag 0

DWIM is Perl's answer to Gödel