71542594e6
Currently it first creates a scorer, then checks if the role bits are sparse, and falls back to the bulk scorer if they are dense. The issue is that creating scorers and bulk scorers is very expensive on some queries such as ranges, prefix and terms queries. So it should rather check whether bits are sparse first in order to decide whether to use the scorer or bulk scorer. Original commit: elastic/x-pack-elasticsearch@067d630099 |
||
---|---|---|
.. | ||
license | ||
qa | ||
x-dev-tools | ||
x-pack | ||
build.gradle | ||
checkstyle_suppressions.xml |