From: chr_news@... (chr_news@...)
Date: 2001-12-12T18:44:32+09:00
Subject: [ruby-talk:28330] Re: Hash.keys Q

ts <decoux@moulon.inra.fr> wrote in message 

>  I suspect that you have just seen the effect of modulo with num_bins

Yeap this is the complete explanation. My point was that the 
hash value alone does not determine the order of the keys if
collision occurs. The up shoot is that the apparent determinism
of the key ordering for small hashes is just a implementation fluke
related to the of the initial bucket size and the (non-equiv)
distribution of hash values.

/Christoph