Re: Re: A (memory) poor man's <strike>hash</strike> lookup table.

[BrowserUk]

in reply to Re: A (memory) poor man's <strike>hash</strike> lookup table.
in thread A (memory) poor man's <strike>hash</strike> lookup table.

The timing comes out to 10 minutes for the insertion and 9.6 minutes for iteration of the million keys.

use Tie::SubstrHash;

tie %hash, 'Tie::SubstrHash', 7, 1, 1_000_000;

print time; 
$hash{ pack 'A7', $_ } = ' ' for 1 .. 1_000_000; 
print time;
1069489311
1069489911 

$n = 0; 
print time; 
$n += defined $hash{ pack 'A7', $_ } for 1 .. 1_000_000; print time, $
+/, $n;

1069490448
1069491023 
 1000000

I think that it may be possible to improve this a little by updating the module. It uses local several places where my would probably be quicker. Though, I suspect that the main performance hit is the cost of tie rather than the implementation. Part of my reasoning for liking the mechanism I described was that it was simple enough in it's implementation that it could be coded directly avoiding the need for tie.

I also think that it would be possible to improve the user interface by having the module perform any padding required rather than the caller,