Struct SpectraBuilder 

pub struct SpectraBuilder { /* private fields */ }

Configures and installs Spectra with explicit storage adapter injection.

Both metrics and events backends are required. Storage persistence is enabled by default; an optional SpectraSink can receive the same emits before they are queued for persistence.

Mode	Calls
Direct persist	backends + .build()
Dual-path	.sink(transport).build()
Publisher (distributed)	.sink(transport).persist_disabled().build()

Publisher processes publish through the sink; consumers subscribe on the host bus and write storage. See the spectra crate Getting started → Mode 2.

Examples

use std::sync::Arc;
use spectra_backend_mem::{MemEventsBackend, MemMetricsBackend};
use spectra_core::{RecordingSink, SpectraSink};
use spectra_runtime::Spectra;

let transport = Arc::new(RecordingSink::new());
let spectra = Spectra::builder()
    .metrics_backend(Arc::new(MemMetricsBackend::new()))
    .events_backend(Arc::new(MemEventsBackend::new()))
    .sink(Arc::clone(&transport) as Arc<dyn SpectraSink>)
    .embedded()
    .build()?;

Implementations

impl SpectraBuilder

pub fn new() -> SpectraBuilder

New builder with async storage persist enabled and no backends configured yet.

pub fn metrics_backend( self, backend: Arc<dyn MetricsStorageBackend>, ) -> SpectraBuilder

Register the metrics storage backend (required before Self::build).

pub fn events_backend( self, backend: Arc<dyn EventStorageBackend>, ) -> SpectraBuilder

Register the events storage backend (required before Self::build).

pub fn config(self, config: SpectraConfig) -> SpectraBuilder

Override emit policy and feature flags (defaults to SpectraConfig::from_env).

pub fn persist(self, config: PersistConfig) -> SpectraBuilder

Configure the L2 persist queue and batch settings (ignored when persist is disabled).

Defaults: queue_max=8192, batch_max=32, batch_wait=5ms, batch_enabled=true. Raise batch_max for high-throughput DW ingest.

pub fn sink(self, sink: Arc<dyn SpectraSink>) -> SpectraBuilder

Optional transport or telemetry sink (invoked before async storage persist when enabled).

Use this for the publisher side of distributed ingest: your SpectraSink publishes emits onto a bus (for example Photon). Combine with Self::persist_disabled when the publisher must not write storage.

Without persist_disabled, the runtime installs a persist wrapper that calls your sink first, then queues async storage writes (dual-path).

Implement SpectraSink in your application or use RecordingSink in tests. See the spectra crate Getting started → Mode 2.

pub fn embedded(self) -> SpectraBuilder

Mark in-process embedded topology (in-host storage backends).

This sets a topology flag retained on the returned Spectra handle (Spectra::is_embedded). It does not select or validate backends — you still inject storage via Self::metrics_backend / Self::events_backend.

pub fn persist_disabled(self) -> SpectraBuilder

Disable async storage persist (only the transport sink receives emits).

Publisher / distributed mode. Pair with Self::sink: writers publish through the sink; separate consumer processes subscribe and persist. Requires a sink — building with persist disabled and no sink returns an error.

pub fn telemetry_ndjson( self, dir: impl AsRef<Path>, ) -> Result<SpectraBuilder, Error>

Attach off-thread NDJSON + optional console mirror telemetry (telemetry-console feature).

Writes {dir}/metrics.ndjson and {dir}/events.ndjson.

pub fn build(self) -> Result<Spectra, Error>

Install global config, router, and sink; returns a handle to the running runtime.

Trait Implementations

impl Default for SpectraBuilder

fn default() -> SpectraBuilder

Returns the “default value” for a type.