public interface DensityAwareVector
For ListVector, we have been using a multiplier of 5 to compute the initial capacity of the inner data vector. For deeply nested lists and lists with lots of NULL values, this is over-allocation upfront. So density helps to be conservative when computing the value capacity of the inner vector.
For example, a density value of 10 implies each position in the list vector has a list of 10 values. So we will provision an initial capacity of (valuecount * 10) for the inner vector. A density value of 0.1 implies out of 10 positions in the list vector, 1 position has a list of size 1 and remaining positions are null (no lists) or empty lists. This helps in tightly controlling the memory we provision for inner data vector.
Similar analogy is applicable for VarCharVector where the capacity of the data buffer can be controlled using density multiplier instead of default multiplier of 8 (default size of average varchar length).
Also from container vectors, we propagate the density down the inner vectors so that they can use it appropriately.
Copyright © 2024 The Apache Software Foundation. All rights reserved.