Added Test.Benchmarking.Experiment for benching pure vs effectful fun…

…ctions

test/Benchmark/Experiment.purs

@@ -0,0 +1,88 @@
+module Test.Benchmark.PureComp
+
+  where
+
+import Prelude
+
+import Data.JSDate (getMilliseconds, now)
+import Effect (Effect)
+import Effect.Class.Console (log)
+
+now' :: Effect Number
+now' = now >>= getMilliseconds
+
+-- | Fibonacci as a pure function
+fib :: Int -> Int
+fib 0 = 0
+fib 1 = 1
+fib n = fib (n-1) + fib (n-2)
+
+-- | Benchmarking Fibonacci, where we let-bind its result
+benchFibLetBind :: Int -> Effect Int
+benchFibLetBind n = do
+  t1 <- now'
+  let m = fib n                -- If `m` is not yet used
+  t2 <- now'
+  log (show (t2 - t1) <> "ms") -- prints 0.0 ms
+  pure m
+
+-- | Benchmarking Fibonacci, where we let-bind its result and log it
+benchFibLetBindPrint :: Int -> Effect Int
+benchFibLetBindPrint n = do
+  t1 <- now'
+  let m = fib n                 -- If `m` is subsequently printed
+  log (show m)
+  t2 <- now'
+  log (show (t2 - t1) <> "ms")  -- prints correct ms
+  pure m
+
+-- | Benchmarking Fibonacci, where we monadically bind its result (using `pure`)
+benchFibDoBind :: Int -> Effect Int
+benchFibDoBind n = do
+  t1 <- now'
+  m <- pure $ fib n             -- If `fib` is wrapped in `pure`, and the result `m` is not yet used
+  t2 <- now'
+  log (show (t2 - t1) <> "ms")  -- prints 0.0 ms
+  pure m
+
+-- | Benchmarking Fibonacci, where we monadically bind its result (using `pure`) and log it
+benchFibDoBindPrint :: Int -> Effect Int
+benchFibDoBindPrint n = do
+  t1 <- now'
+  m <- pure $ fib n               -- If `fib` is wrapped in `pure`, and the result `m` is subsequently printed
+  log (show m)
+  t2 <- now'
+  log (show (t2 - t1) <> "ms")   -- prints correct ms
+  pure m
+
+-- | Fibonacci as an effectful function
+fibm :: Int -> Effect Int
+fibm 0 = pure 0
+fibm 1 = pure 1
+fibm n = do
+  n1 <- fibm (n-1)
+  n2 <- fibm (n-2)
+  pure (n1 + n2)
+
+-- | Benchmarking the effectful version of Fibonacci, where we monadically bind its result
+benchFibmBind :: Int -> Effect Int
+benchFibmBind n = do
+  t1 <- now'
+  m <- fibm n                    -- If `fib` itself is defined as an effectful computation `fibm`
+  t2 <- now'
+  log (show (t2 - t1) <> "ms")   -- prints correct ms (without needing the result `m` to be used beforehand)
+  pure m
+
+benchFibs :: Effect Unit
+benchFibs = do
+  log ("-- | Benchmarking Fibonacci, where we let-bind its result")
+  _ <- benchFibLetBind 40
+  log ("-- | Benchmarking Fibonacci, where we let-bind its result and log it")
+  _ <- benchFibLetBindPrint 40
+  log ("-- | Benchmarking Fibonacci, where we monadically bind its result (using `pure`)")
+  _ <- benchFibDoBind 40
+  log ("-- | Benchmarking Fibonacci, where we monadically bind its result (using `pure`) and log it")
+  _ <- benchFibDoBindPrint 40
+  log ("-- | Benchmarking the effectful version of Fibonacci, where we monadically bind its result")
+  _ <- benchFibmBind 40
+  pure unit

test/Main.purs

@@ -18,9 +18,12 @@ import Test.Many (bwdMany, linkMany, many, withDatasetMany)
 import Test.Spec.Mocha (run)
 import Test.Spec.Specs (bwd_cases, desugar_cases, graphics_cases, linking_cases, misc_cases)
 import Util (type (×))
+import Test.Benchmark.PureComp (benchFibs)
 
 main :: Effect Unit
-main = run tests
+main = do
+   if true then benchFibs else pure unit
+   run tests
 
 tests :: Array (String × Aff Unit)
 tests = concat