columna API reference - v0.5.0
    Preparing search index...

    Interface ExecutionReport

    What actually ran, as opposed to what explain() planned.

    interface ExecutionReport {
        backendsUsed: EngineKind[];
        cachedFrom?: {
            backendsUsed: EngineKind[];
            dispatched: EngineKind;
            requested: EngineKind;
            strict: boolean;
        };
        cacheHit?: boolean;
        cacheSkipped?: string;
        dispatched: EngineKind;
        events: ExecutionEvent[];
        fallbacks: { from: EngineKind; reason: string; to: EngineKind }[];
        peakBytes?: number;
        requested: EngineKind;
        spilledBytes?: number;
        strict: boolean;
        totalMs: number;
    }
    Index
    backendsUsed: EngineKind[]

    Distinct backends that executed at least one node.

    cachedFrom?: {
        backendsUsed: EngineKind[];
        dispatched: EngineKind;
        requested: EngineKind;
        strict: boolean;
    }

    On a cache hit: the request and engines of the execution that produced the cached table.

    cacheHit?: boolean

    True when persist() served the result from the LRU cache. dispatched / backendsUsed then describe the execution that produced the cached table (also in cachedFrom) — nothing executed during this call.

    cacheSkipped?: string

    The plan was persist()-marked but not cacheable (caller-owned buffers, an untrusted UDF); why.

    dispatched: EngineKind

    Backend the runtime dispatched the plan to.

    events: ExecutionEvent[]
    fallbacks: { from: EngineKind; reason: string; to: EngineKind }[]

    Whole-plan fallbacks performed by the runtime (backend threw or declined).

    peakBytes?: number

    Peak estimated live table bytes observed during this execution.

    requested: EngineKind
    spilledBytes?: number

    Bytes written to spill files during this execution (Node, when MemoryPolicy.spill is on).

    strict: boolean
    totalMs: number