> So there will be at most one hash implementation? One per binary I expect. Modern hash functions are pretty good, so it's unlikely that someone needs to come up with special purpose hashes.
I found Bob Jenkins' spooky is rather good for this case (very large incremential keys), but it is only efficient on 64bit hosts. So it would need a fallback (like mumurhash2a or the existing one) on 32bit. Another alternative would be to use CityHash, but that also has multiple variants, including some machine specific ones (e.g. use the CRC instructions on SSE2 hosts) > Maybe use a namespace instead of a hash then? I don't understand the suggestion. > So other places can extend it? Not sure that is needed. > Why didn't you replace the tree.c uses BTW? Patches were already quite big, but I'll add it. -Andi
