Interface QueryPipeline
public interface QueryPipeline
Runs queries through the modern pipeline and streams the assembled rows into a
RowSink.-
Method Summary
Modifier and TypeMethodDescriptionrun(DataPointQueryOptions<?> options, ResultWriterOptions writerOptions, RowSink sink) Runs a query and streams its rows intosink.run(MultiEngineQueryOptions options, ResultWriterOptions writerOptions, RowSink sink) Runs amulti-enginequery: each entry's keys run against their declared engine with that engine's options, and the results stream into oneRowSink.
-
Method Details
-
run
Runs a query and streams its rows intosink.Row ordering
Aggregatedqueries emit rows in timestamp order across keys.Rawqueries emit rows in the order engines produce them, so with several engines a sink may see one key's T=200 before another's T=100. Use an aggregated query, or sort downstream, for a timestamp-ordered view.
Termination
On success,Flow.Subscriber.onComplete()fires after the last row and the handle'sresult()completes with aGoodQueryResult. On error,Flow.Subscriber.onError(Throwable)fires exactly once,onCompleteis not called, andresult()completes exceptionally. -
run
Runs amulti-enginequery: each entry's keys run against their declared engine with that engine's options, and the results stream into oneRowSink.Ordering per engine follows the single-engine overload; across engines it is arrival order. Each
Rowcarries theengineIdof the engine that produced it (seeRow.engineId()); the sharedRowSchemadoes not.Flow.Subscriber.onComplete()fires after the last engine's last row;Flow.Subscriber.onError(Throwable)fires at most once for any engine's failure. Engines run concurrently, and one engine's failure does not cancel the others unlesscancel()is called on the returned handle.- Returns:
- a handle summarizing all engines' progress and completion
-