2 ms·
Obacht. They don't even mention the original Bloom filter problem, the constant overhead of calculating 3-5 hashes per key, which is problematic with overly slo
by rurban 5y ago
Obacht. They don't even mention the original Bloom filter problem, the constant overhead of calculating 3-5 hashes per key, which is problematic with overly slow hash functions (like siphash) or overlong keys. Then it's faster to go without bloom, xor or ribbon, which just cancels searching for not-included keys.
Or if you know that your key is always included in the dataset, it is pure overhead.