singleflight

Request deduplication for Gleam.

Concurrent requests for the same key are only execute once, all calls share the same result without repeated work.

Blog

I wrote about how this works here

Install

gleam add singleflight

Usage

import gleam/io
import gleam/erlang/process
import gleam/otp/actor
import singleflight

pub fn main() -> Nil {
  let name = process.new_name("singleflight")
  let config = singleflight.config(1_000, 1_000)

  let assert Ok(actor.Started(data: server, ..)) =
    singleflight.start(config, name)

  let value =
    case singleflight.fetch(server, "key", fn(key) { key <> "-value" }) {
      Ok(value) -> value
      Error(singleflight.Crashed) -> panic as "singleflight worker crashed"
      Error(singleflight.TimedOut) -> panic as "singleflight request timed out"
    }

  io.debug(value)
}

Errors

fetch returns Error(singleflight.Crashed) if the singleflight actor or the worker process exits before producing a value.

fetch returns Error(singleflight.TimedOut) if no reply arrives within fetch_timeout_ms.