Re: g_str_hash

["B. Scott Michel" <scottm CS UCLA EDU>]

Tim Janik wrote:
> it of course depends on what kind of string you want to hash, for
> strings where you mostly care about the tails, you're certainly willing
> to trade some of the higher bits and thus allow hash collisions for
> strings that only differ in the beginning characters.

One could simply code up a 32-bit CRC generator which works nicely
as a hash function and also takes 5 or so lines (depending on how
you perform the calculation.) If you want a small number of bits
for the hash code instead of all 32, mask what you need. Smaller
bit CRC functions can be had by using invariant polys of smaller
degree from the back of any good error correction book. And the
really cool part is that the prob of a collision is 2^{-L}, where
L is the number of bits in the CRC.

And it works well with strings and incremental hashing.


-scooter