Queueing fun

Aparently the postgresql people are switching from a patent encumbered caching method to a free one (2Q), I happen to be interested in such things. 2Q seems pretty cool, with a reasonable performance increase over LRU for actual usage. I'm wondering how my squid cache would handle using an LRU/2 or 2Q implementation.

Bitwise Hackery for those of us using languages that allow such operations.