The search function of the ordinary index can probably get by with less allocations. We shall conduct a micro-benchmark in the style of the compact-index micro-benchmark and use its results to reduce the allocation count by turning specific patterns into bang patterns. See https://github.com/IntersectMBO/lsm-tree/pull/244#discussion_r1632935316 and follow-ups.
We shall extend the above-mentioned optimization to the binarySearchL function for immutable vectors, which is used by search, in case we haven’t ended up implementing that function by reusing binarySearchL from vector-algorithms.
The
searchfunction of the ordinary index can probably get by with less allocations. We shall conduct a micro-benchmark in the style of the compact-index micro-benchmark and use its results to reduce the allocation count by turning specific patterns into bang patterns. See https://github.com/IntersectMBO/lsm-tree/pull/244#discussion_r1632935316 and follow-ups.We shall extend the above-mentioned optimization to the
binarySearchLfunction for immutable vectors, which is used bysearch, in case we haven’t ended up implementing that function by reusingbinarySearchLfromvector-algorithms.