> They should, in theory, be faster than btrees -- O(1) not O(log N) page
> fetches per lookup. In practice they don't seem to be faster, and
> nobody's bothered to find out exactly why. Again, this isn't a terribly
> encouraging precedent for implementing some other index type that's
> supposed to (sometimes) be faster than btrees.
Yes, I understand. Which is also why I was curious to know if the "claims" those papers (and the databases using them)
makewere real...
Thank you everybody for your replies.
Leonardo