Struct dataflow::logging::BatchLogger

pub struct BatchLogger<T, E, P> where
    P: EventPusher<Timestamp, (Duration, E, T)>,  {
    time_ms: Timestamp,
    event_pusher: P,
    _phantom: PhantomData<(E, T)>,
    granularity_ms: u64,
    buffer: Vec<(Duration, E, T)>,
}

Logs events as a timely stream, with progress statements.

Fields

time_ms: Timestamp

Time in milliseconds of the current expressed capability.

event_pusher: P

_phantom: PhantomData<(E, T)>

granularity_ms: u64

Each time is advanced to the strictly next millisecond that is a multiple of this granularity. This means we should be able to perform the same action on timestamp capabilities, and only flush buffers when this timestamp advances.

buffer: Vec<(Duration, E, T)>

A stash for data that does not yet need to be sent.

Implementations

impl<T, E, P> BatchLogger<T, E, P> where
    P: EventPusher<Timestamp, (Duration, E, T)>, 

const BATCH_SIZE_BYTES: usize

Batch size in bytes for batches

fn buffer_capacity() -> usize

Calculate the default buffer size based on (Duration, E, T) tuples.

pub fn new(event_pusher: P, granularity_ms: u64) -> Self

Creates a new batch logger.

pub fn publish_batch(
    &mut self,
    time: &Duration,
    data: &mut Vec<(Duration, E, T)>
)

Publishes a batch of logged events and advances the capability.

Trait Implementations

impl<T, E, P> Drop for BatchLogger<T, E, P> where
    P: EventPusher<Timestamp, (Duration, E, T)>, 

fn drop(&mut self)

Executes the destructor for this type.