Question

Say I want to compute two long running processes in two separate threads in Haskell. However, I only care about the result from the first one done. How would I do this?

Example (pseudo code):

thread1 = spark $ long_running some_arg1
thread2 = spark $ long_running some_arg2
result = first_done thread1 thread2 -- Maybe even first_done [thread1, thread2]?

Answer 1

The async package does this, and is now part of the Haskell Platform.

import Control.Concurrent.Async
import Control.Concurrent (threadDelay)

main :: IO ()
main = do
  x <- async (threadDelay 2000000 >> return 1)
  y <- async (threadDelay 1000000 >> return 2)
  (_, res) <- waitAnyCancel [x, y]
  print (res :: Int)