I, for one, would like to see your code for building an 5-byte index?
Storing a million 5-byte integers using 5 million bytes for data, and 44 bytes additional overhead:
my $buffer = "";
my $BYTES = 5;
my $BITS_IN_BYTE = 8;
my $FULL_BYTE = (1 << $BITS_IN_BYTE) - 1;
sub store {
my ($pos, $value) = @_;
for (0 .. $BYTES - 1) {
vec($buffer, $pos * $BYTES + $_, $BITS_IN_BYTE) =
($value >> ($BITS_IN_BYTE * ($BYTES - 1 - $_))) & $FULL_BYT
+E;
}
}
sub fetch {
my ($pos) = @_;
my $sum = 0;
for (0 .. $BYTES - 1) {
$sum <<= $BITS_IN_BYTE;
$sum += vec($buffer, $pos * $BYTES + $_, $BITS_IN_BYTE);
}
$sum;
}
#
# Testing
#
my $TEST_SIZE = 1_000_000;
my @offsets = map {int rand 1 << ($BYTES * $BITS_IN_BYTE)} 1 .. $TEST_
+SIZE;
for (my $i = 0; $i < @offsets; $i++) {
store $i, $offsets[$i];
}
for (my $i = 0; $i < @offsets; $i++) {
my $val = fetch $i;
die unless $val == $offsets[$i];
}
use Devel::Size 'size';
use 5.010;
say "Index size: ", size $buffer;
__END__
Index size: 5000044