Safe Haskell | Safe-Inferred |
---|

Provides a convenient and fast alternative to the common
`forM_ [1..n]`

idiom, which in many cases GHC cannot fuse to efficient
code.

Notes on fast iteration:

- For
`Int`

,`(+1)`

is almost twice as fast as`succ`

because`succ`

does an overflow check. - For
`Int`

, you can get around that while still using`Enum`

using`toEnum . (+ 1) . fromEnum`

. - However,
`toEnum . (+ 1) . fromEnum`

is slower than`succ`

for`Word32`

on 64-bit machines since`toEnum`

has to check if the given`Int`

exceeds 32 bits. - Using
`(+1)`

from`Num`

is always the fastest way, but it gives no overflow checking. - Using
`forLoop`

you can flexibly pick the way of increasing the value that best fits your needs. - The currently recommended replacement for
`forM_ [1..n]`

is`forLoop 1 (<= n) (+1)`

.

# Documentation

forLoop :: Monad m => a -> (a -> Bool) -> (a -> a) -> (a -> m ()) -> m ()Source

`forLoop start cond inc f`

: A C-style for loop with starting value,
loop condition and incrementor.

forLoopFold :: a -> (a -> Bool) -> (a -> a) -> acc -> (acc -> a -> acc) -> accSource

`forLoopFold start cond inc acc0 f`

: A pure fold using a for loop
instead of a list for performance.

Care is taken that `acc0`

not be strictly evaluated if unless done so by `f`

.

numLoop :: (Num a, Eq a, Monad m) => a -> a -> (a -> m ()) -> m ()Source

`numLoop start end f`

: Loops over a contiguous numerical range, including
`end`

.

It uses `(+ 1)`

so for most integer types it has no bounds (overflow) check.

numLoopFold :: (Num a, Eq a) => a -> a -> acc -> (acc -> a -> acc) -> accSource

`numLoopFold start end acc0 f`

: A pure fold over a contiguous numerical
range, including `end`

.

It uses `(+ 1)`

so for most integer types it has no bounds (overflow) check.

Care is taken that `acc0`

not be strictly evaluated if unless done so by `f`

.