getdozer / dozer / 5966277235

76%

main: 78%

Pull #1912

github

feat: make probabilistic optimizations optional and tunable in the YAML config

Probabilistic optimization sacrifices accuracy in order to reduce memory consumption. In certain parts of the pipeline, a Bloom Filter is used ([set_processor](https://github.com/getdozer/dozer/blob/<a class=hub.com/getdozer/dozer/commit/<a class="double-link" href="https://git"><a class=hub.com/getdozer/dozer/commit/2e3ba96c3f4bdf9a691747191ab15617564d8ca2">2e3ba96c3/dozer-sql/src/pipeline/product/set/set_processor.rs#L20)), while in other parts, hash tables that store only the hash of the keys instead of the full keys are used ([aggregation_processor](https://github.com/getdozer/dozer/blob/2e3ba96c3f4bdf9a691747191ab15617564d8ca2/dozer-sql/src/pipeline/aggregation/processor.rs#L59) and [join_processor](https://github.com/getdozer/dozer/blob/2e3ba96c3f4bdf9a691747191ab15617564d8ca2/dozer-sql/src/pipeline/product/join/operator.rs#L57-L58)).

This commit makes these optimizations disabled by default and offers user-configurable flags to enable each of these optimizations separately.

This is an example of how to turn on probabilistic optimizations for each processor in the Dozer configuration.

```
flags:
  enable_probabilistic_optimizations:
    in_sets: true # enable probabilistic optimizations in set operations (UNION, EXCEPT, INTERSECT); Default: false
    in_joins: true # enable probabilistic optimizations in JOIN operations; Default: false
    in_aggregations: true # enable probabilistic optimizations in aggregations (SUM, COUNT, MIN, etc.); Default: false
```

542 of 542 new or added lines in 26 files covered. (100.0%)

47321 of 62989 relevant lines covered (75.13%)

47456.64 hits per line