effect-io-ai

Package: effect
Module: Channel

Channel.bufferArray

Buffers array output elements in a queue with the configured capacity so a faster producer can progress independently of a slower consumer.

When to use

Use when emitted arrays are batches of elements and it is acceptable for buffering to flatten and rebuild those batches.

Details

Finite queues use the strategy option. The default "suspend" strategy applies backpressure, while "dropping" and "sliding" can discard output elements when the queue is full. "unbounded" capacity does not use a finite capacity strategy.

Gotchas

Input arrays are offered to the queue element-by-element and outputs are rebuilt from the currently available queued elements, so upstream array boundaries are not preserved.

See

• buffer for buffering output elements without flattening arrays

Signature

declare const bufferArray: { (options: { readonly capacity: "unbounded"; } | { readonly capacity: number; readonly strategy?: "dropping" | "sliding" | "suspend" | undefined; }): <OutElem, OutErr, OutDone, InElem, InErr, InDone, Env>(self: Channel<Arr.NonEmptyReadonlyArray<OutElem>, OutErr, OutDone, InElem, InErr, InDone, Env>) => Channel<Arr.NonEmptyReadonlyArray<OutElem>, OutErr, OutDone, InElem, InErr, InDone, Env>; <OutElem, OutErr, OutDone, InElem, InErr, InDone, Env>(self: Channel<Arr.NonEmptyReadonlyArray<OutElem>, OutErr, OutDone, InElem, InErr, InDone, Env>, options: { readonly capacity: "unbounded"; } | { readonly capacity: number; readonly strategy?: "dropping" | "sliding" | "suspend" | undefined; }): Channel<Arr.NonEmptyReadonlyArray<OutElem>, OutErr, OutDone, InElem, InErr, InDone, Env>; }

Source

Since v4.0.0