[Perlfect-search] idea for a smaller index

[Daniel Naber]

Hi,

I had the idea to replace this (in indexer.pl):

$weights .= pack("ff", $doc_id, $weight);

by:

$weights .= pack("sf", $doc_id, $weight);

this means that one file index gets about 30% or so smaller, as a short 
value is used instead of a float. But it doesn't seem to be possible to 
unpack() these values correctly, or am I missing something? It's not just 
one value in there, but many values after another...

Regards
 Daniel