Struct FileSink 

pub struct FileSink { /* private fields */ }

Appends events as JSON Lines to a file. Parent directories are created on demand.

Concurrency and line integrity

Concurrent emit calls on the same FileSink are serialised via a shared tokio::sync::Mutex. This is required for JSONL correctness: on POSIX, O_APPEND only guarantees atomicity for individual write() calls up to PIPE_BUF (4 KiB on Linux). An event whose serialised form exceeds that bound — plausible for events with many attrs — would interleave at the byte level with concurrent writers and produce malformed JSONL.

Cross-process writers (two FileSinks in different processes targeting the same file) remain interleaving-safe only up to PIPE_BUF. Don’t do that; use per-process files and aggregate elsewhere.

Batching lives in a wrapper sink when we need it (v0.2).

Implementations

impl FileSink

pub fn new(path: impl Into<PathBuf>) -> Self

Construct a sink targeting path. The file is not touched until the first emit; parent directories are created on demand at that point.

Trait Implementations

impl Clone for FileSink

fn clone(&self) -> FileSink

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for FileSink

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

Formats the value using the given formatter.

impl TelemetrySink for FileSink

fn emit<'life0, 'life1, 'async_trait>( &'life0 self, event: &'life1 Event, ) -> Pin<Box<dyn Future<Output = Result<(), TelemetryError>> + Send + 'async_trait>>

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

Emit a single event. Sinks SHOULD be idempotent on the event — retries or batching live in wrapper sinks (v0.2).

fn flush<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = Result<(), TelemetryError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Flush any buffered events. Default impl is a no-op.