Message from JavaScript discussions
May 2017
— Reduces disk access by a large factor for most queries
Like I was saying about Bloom filters earlier, which are essentialy a set of combined hash algorithms to create one hash/bit map for a set
— It is a bit array, not quite a hash
— If an element is probably in the set, the query will run slightly longer... if it is definitely not in the set, the query will run very fast and avoid computations
— Huh
— This idea was to only split into child objects when there is a difference
— So { "Hello", "Helium" } would become:
{ Hel: { "ium", "lo" } }
— I don't like it since it mixes representation with storage
— I mean, it looks clever, but that doesn't always mean it's a good idea
— This may help speed too https://en.wikipedia.org/wiki/Skip_list
— Since it works with subsets of sets
— I think you can't remove elements from a Bloom filter though