> On a related note, I have in general concerns about latency spikes due to rehashing. Is there a good, general, incremental rehashing scheme? One option is to

> On a related note, I have in general concerns about latency spikes due to rehashing. Is there a good, general, incremental rehashing scheme? One option is to move a few elements over from the old hash table on each operation[0]. A second option is to use linear hashing[1] (different from linear probing). Linear hashing let's you increase the size of your hash table by one bucket at a time. The general idea is once your hash table is full, you allocate a new bucket (generally from a preallocated array) and move half of the elements from an old bucket to the new bucket. I know Postgres used to use linear hashing, but I'm not sure about the current state. I heard they started using Robin Hood hashing, but I haven't looked at whether that is unrelated to the code that uses linear hashing.