Re^5: Better Hash Tables?

[Jenda]

If you make the engine 1 % more efficient while at the same time it weights ten times as much and costs twenty times as much, I say, please make the tank 1 % bigger. Thank you.

The algorithm suggested by the paper adds complexity and slows all adds and the advantage is better asymptotic behavior in case of hugely overfilled tables. Using this algorithm would be optimizing for situation that doesn't happen. Any even just remotely sane implementation of hashes increases the size of the table long long before this algorithm might provide any improvement.

It may make sense in some special cases when the memory is the scarce resource and a slowing down all adds/searches by a constant factor is not the important thing. Possibly when the hash table is a physical processor cache or something, but that's not what Perl is for and it's not where Perl is used.