17 Jan
2005
17 Jan
'05
4:16 p.m.
At 04:51 AM 1/17/2005, Matt Hurd wrote:
Perfect hashing and statistically reasonable hashing are quite different but practically the same ;-)
Isn't that stretching it a bit? Collision handling code can be eliminated if you know the hashing is perfect, for example. Although I do understand your point that for many practical applications, perfect hashing and statistically reasonable hashing would yield very similar performance. Seems like both could have a place in Boost. Statistically reasonable hashing of particular sets of strings has been a need that I've often run into. --Beman