Struct PlatformNotifyMiddleware 

pub struct PlatformNotifyMiddleware { /* private fields */ }

Platform notification middleware — monitors error rates and latency, emits threshold events with hysteresis.

Emits apcore.health.error_threshold_exceeded when a module’s error rate crosses the configured threshold, apcore.health.latency_threshold_exceeded when p99 latency exceeds the limit, and apcore.health.recovered when a previously alerted module recovers below threshold * 0.5.

Issue #36 — canonical event-name standardization: every threshold event is dual-emitted under both the canonical apcore.health.* name and its legacy bare-name alias (e.g. error_threshold_exceeded) so that existing subscribers continue to fire while consumers migrate to the canonical names. The legacy event payload carries a deprecated: true marker.

Hysteresis prevents repeated alerts until recovery is observed.

Implementations

impl PlatformNotifyMiddleware

pub fn new( emitter: EventEmitter, metrics_collector: Option<MetricsCollector>, error_rate_threshold: f64, latency_p99_threshold_ms: f64, ) -> Self

Create a new platform notify middleware.

Arguments

• emitter — EventEmitter to emit threshold events to.
• metrics_collector — Optional MetricsCollector to read error rates and latency from. If None, all checks return 0.
• error_rate_threshold — Error rate (0.0-1.0) above which to alert.
• latency_p99_threshold_ms — p99 latency in ms above which to alert.

pub fn with_defaults( emitter: EventEmitter, metrics_collector: Option<MetricsCollector>, ) -> Self

Create with default thresholds (10% error rate, 5000ms p99 latency).

Trait Implementations

impl Debug for PlatformNotifyMiddleware

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Middleware for PlatformNotifyMiddleware

fn name(&self) -> &'static str

Name of this middleware for logging/debugging.

fn before<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, _module_id: &'life1 str, _inputs: Value, _ctx: &'life2 Context<Value>, ) -> Pin<Box<dyn Future<Output = Result<Option<Value>, ModuleError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Called before module execution. Can modify input. Return Ok(None) to pass through unchanged, Ok(Some(v)) to modify.

fn after<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, module_id: &'life1 str, _inputs: Value, _output: Value, _ctx: &'life2 Context<Value>, ) -> Pin<Box<dyn Future<Output = Result<Option<Value>, ModuleError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Called after successful module execution. Can modify output. inputs is the original (post-before) input for correlation. Return Ok(None) to pass through unchanged, Ok(Some(v)) to modify.

fn on_error<'life0, 'life1, 'life2, 'life3, 'async_trait>( &'life0 self, module_id: &'life1 str, _inputs: Value, _error: &'life2 ModuleError, _ctx: &'life3 Context<Value>, ) -> Pin<Box<dyn Future<Output = Result<Option<Value>, ModuleError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, 'life3: 'async_trait,

Called when module execution fails. inputs is the original (post-before) input for correlation. Return Ok(Some(v)) to signal a recovery output, or Ok(None) to let the error propagate.

fn priority(&self) -> u16

Priority of this middleware (higher runs first). Default is 100. Valid range: 0-1000 (enforced by MiddlewareManager::add). When two middlewares have the same priority, registration order is preserved.

fn on_error_outcome<'life0, 'life1, 'life2, 'life3, 'async_trait>( &'life0 self, module_id: &'life1 str, inputs: Value, error: &'life2 ModuleError, ctx: &'life3 Context<Value>, ) -> Pin<Box<dyn Future<Output = Result<Option<OnErrorOutcome>, ModuleError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, 'life3: 'async_trait,

Extended on_error hook that can request a pipeline retry via OnErrorOutcome::Retry in addition to producing a recovery output.