Difference between revisions of "Output/Input"

From HaskellWiki
Jump to navigation Jump to search
(Types and instances for more mutable arrays added)
m (Unlinked "IO, partible-style")
(41 intermediate revisions by the same user not shown)
Line 1: Line 1:
[[Category:Code]]
+
[[Category:Theoretical foundations]]
   
  +
=== <u>Clearing away the smoke and mirrors</u> ===
Let me guess...you've read every other guide, tutorial, lesson and introduction and none of them have helped - you still don't understand I/O in Haskell.
 
   
  +
<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
Alright then - have a look at this:
 
  +
The implementation in GHC uses the following one:
   
 
<haskell>
 
<haskell>
  +
type IO a = World -> (a, World)
data OI -- abstract, primitive
 
  +
</haskell>
   
  +
An <code>IO</code> computation is a function that (logically) takes the state of the world, and returns a modified world as well as the return value. Of course, GHC does not actually pass the world around; instead, it passes a dummy “token,” to ensure proper sequencing of actions in the presence of lazy evaluation, and performs input and output as actual side effects!
partOI :: OI -> (OI, OI) --
 
getchar :: OI -> Char -- primitives
 
putchar :: Char -> OI -> () --
 
   
  +
<tt>[https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.168.4008&rep=rep1&type=pdf A History of Haskell: Being Lazy With Class], Paul Hudak, John Hughes, Simon Peyton Jones and Philip Wadler.</tt>
seq :: a -> b -> b -- also primitive
 
  +
</div>
   
  +
...so what starts out as an I/O action of type:
instance Partible OI where ...
 
   
  +
<haskell>
class Partible a where
 
part :: a -> (a, a)
+
World -> (a, World)
parts :: a -> [a]
 
.
 
.
 
.
 
 
</haskell>
 
</haskell>
 
No up-front explanation; I'm guessing you've seen more than enough of those, so I'm trying something different. I will explain it later...
 
   
  +
is changed by GHC to approximately:
Yes, of course there's more to Haskell I/O than <code>getchar</code> and <code>putchar</code>; I've downsized it for convenience. If you want, you can add the rest afterwards...
 
 
Yes, they're somewhat arcane, but they can be used to emulate all the classic approaches to I/O in Haskell, albeit in miniature:
 
   
 
<haskell>
 
<haskell>
  +
() -> (a, ())
module ClassicIO where
 
import qualified Prelude as T
 
import Prelude(Char, String)
 
import Prelude(($), (.))
 
import Data.List(map, foldr, zipWith)
 
import OutputInput
 
import Partible
 
 
-- simple text --
 
 
{- main :: (String -> String) -}
 
 
runMain_text :: (String -> String) -> OI -> ()
 
runMain_text main = \u -> case part u of
 
(u1, u2) ->
 
putchars (main (getchars u1)) u2
 
 
getchars :: OI -> String
 
getchars = map getchar . parts
 
 
putchars :: String -> OI -> ()
 
putchars s = foldr seq () . zipWith putchar s . parts
 
 
 
-- dialogues --
 
 
{- main :: Dialogue -}
 
 
runMain_dial :: Dialogue -> OI -> ()
 
runMain_dial main = \u -> foldr seq () $ yet $
 
\l -> zipWith respond (main l) (parts u)
 
 
type Dialogue = [Response] -> [Request]
 
 
data Request = Getq | Putq Char
 
data Response = Getp Char | Putp
 
 
yet :: (a -> a) -> a
 
yet f = f (yet f)
 
 
respond :: Request -> OI -> Response
 
respond Getq = \u -> case getchar u of c -> Getp c
 
respond (Putq c) = \u -> seq (putchar c u) Putp
 
 
 
-- continuations --
 
 
{- main :: (() -> IOResult) -> IOResult -}
 
 
runMain_cont :: ((() -> IOResult) -> IOResult) -> OI -> ()
 
runMain_cont main = call (main done)
 
 
newtype IOResult = R (OI -> ())
 
 
call :: IOResult -> OI -> ()
 
call (R a) = a
 
 
done :: () -> IOResult
 
done () = R $ \ u -> part u `seq` ()
 
 
getchar_cont :: (Char -> IOResult) -> IOResult
 
getchar_cont k = R $ \u -> case part u of
 
(u1, u2) ->
 
case getchar u1 of
 
c -> seq c (call (k c) u2)
 
 
putchar_cont :: Char -> (() -> IOResult) -> IOResult
 
putchar_cont c k = R $ \u -> case part u of
 
(u1, u2) ->
 
seq (putchar c u1) (call (k ()) u2)
 
 
-- state-passing --
 
 
{- main :: IOState -> ((), IOState) -}
 
 
runMain_stat :: (IOState -> ((), IOState)) -> OI -> ()
 
runMain_stat main = \u -> seq (main (ini_st u)) ()
 
 
newtype IOState = S OI
 
 
ini_st :: OI -> IOState
 
ini_st = S
 
 
getchar_stat :: IOState -> (Char, IOState)
 
getchar_stat (S u) = case part u of
 
(u1, u2) ->
 
case getchar u1 of
 
c -> seq c (c, S u2)
 
 
putchar_stat :: Char -> IOState -> ((), IOState)
 
putchar_stat c (S u) = case part u of
 
(u1, u2) ->
 
seq (putchar c u1) ((), S u2)
 
 
-- and those weird, fickle things ;-)
 
 
{- main :: IO () -}
 
 
runMain_wfth :: IO () -> OI -> ()
 
runMain_wfth main = main
 
 
type IO a = OI -> a
 
 
getchar_wfth :: IO Char
 
getchar_wfth = getchar
 
 
putchar_wfth :: Char -> IO ()
 
putchar_wfth = putchar
 
 
unit :: a -> IO a
 
unit x = \u -> part u `seq` x
 
 
bind :: IO a -> (a -> IO b) -> IO b
 
bind m k = \u -> case part u of
 
(u1, u2) -> (\x -> x `seq` k x u2) (m u1)
 
 
-- supporting definitions --
 
--
 
getchar :: OI -> Char
 
getchar = "getchar" `invokes` T.getChar
 
 
putchar :: Char -> OI -> ()
 
putchar c = "putchar" `invokes` T.putChar c
 
 
</haskell>
 
</haskell>
   
What was that - using <code>Prelude.seq</code> that way won't work in Haskell 2010? You are ''correct!''<br>
+
As the returned unit-value <code>()</code> contains no useful information, that type can be simplified further:
Now look closely at those imports...
 
 
Moving on, here are examples using each of those approaches:
 
   
 
<haskell>
 
<haskell>
  +
() -> a
module Echoes where
 
  +
</haskell>
import Prelude(String, Char(..), Eq(..))
 
import Prelude(($))
 
import ClassicIO
 
import OutputInput(runOI)
 
   
  +
<sub>Why "approximately"? Because "logically" a function in Haskell has no observable effects.</sub>
echo_text :: String -> String
 
echo_text (c:cs) = if c == '\n' then [] else c : echo_text cs
 
   
  +
----
echo_dial :: Dialogue
 
  +
=== <u>Previously seen</u> ===
echo_dial p = Getq :
 
case p of
 
Getp c : p' ->
 
if c == '\n' then
 
[]
 
else
 
Putq c :
 
case p' of
 
Putp : p'' -> echo_dial p''
 
   
  +
The type <code>() -> a</code> (or variations of it) have appeared elsewhere - examples include:
echo_cont :: (() -> IOResult) -> IOResult
 
echo_cont k = getchar_cont $ \c ->
 
if c == '\n' then
 
k ()
 
else
 
putchar_cont c (\_ -> echo_cont k)
 
   
  +
* page 2 of 13 in [https://fi.ort.edu.uy/innovaportal/file/20124/1/22-landin_correspondence-between-algol-60-and-churchs-lambda-notation.pdf A Correspondence Between ALGOL 60 and Church's Lambda-Notation: Part I] by Peter Landin:
echo_stat :: IOState -> ((), IOState)
 
  +
:{|
echo_stat s = case getchar_stat s of
 
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
(c, s') ->
 
if c == '\n' then
 
((), s')
 
else
 
case putchar_stat c s' of
 
(_, s'') -> echo_stat s''
 
   
  +
The use of <code>λ</code>, and in particular (to avoid an irrelevant bound variable) of <code>λ()</code> , to delay and possibly avoid evaluation is exploited repeatedly in our model of ALGOL 60. A function that requires an argument-list of length zero is called a ''none-adic'' function.
echo_wfth :: IO ()
 
  +
</div>
echo_wfth = getchar_wfth `bind` \c ->
 
  +
<sup> </sup>
if c == '\n' then
 
  +
<haskell>
unit ()
 
  +
(\ () -> …) :: () -> a
else
 
putchar_wfth c `bind` \_ -> echo_wfth
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* page 148 of 168 in [[IO Semantics|Functional programming and Input/Output]] by Andrew Gordon:
Regarding <code>seq</code>, this should work as expected[[#refs|[1][2][3]]]:
 
  +
:{|
 
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
  +
<pre>
  +
abstype 'a Job = JOB of unit -> 'a
  +
</pre>
  +
</div>
  +
<sup> </sup>
 
<haskell>
 
<haskell>
  +
data Job a = JOB (() -> a)
-- for GHC 8.6.5
 
{-# LANGUAGE CPP #-}
 
#define during seq
 
module Sequential(seq) where
 
import qualified Prelude(during)
 
 
{-# NOINLINE seq #-}
 
infixr 0 `seq`
 
seq :: a -> b -> b
 
seq x y = Prelude.during x (case x of _ -> y)
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* page 3 of [https://www.cs.bham.ac.uk/~udr/papers/assign.pdf Assignments for Applicative Languages] by Vipin Swarup, Uday S. Reddy and Evan Ireland:
It didn't work? Try this instead:
 
  +
:{|
 
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
  +
A value of type <code>Obs 𝜏</code> is called an ''observer''. Such a value observes (i.e. views or inspects) a state and returns a value of type <code>𝜏</code>. [...] An observer type <code>Obs 𝜏</code> may be viewed as an implicit function space from the set of states to the type <code>𝜏</code>.
  +
</div>
  +
<sup> </sup>
 
<haskell>
 
<haskell>
  +
type Obs tau = State -> tau
-- for GHC 8.6.5
 
{-# LANGUAGE CPP #-}
 
#define during seq
 
module Sequential(seq) where
 
import qualified Prelude(during)
 
import GHC.Base(lazy)
 
 
infixr 0 `seq`
 
seq :: a -> b -> b
 
seq x y = Prelude.during x (lazy y)
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* [https://image.slidesharecdn.com/lazyio-120422092926-phpapp01/95/lazy-io-15-728.jpg page 15] of ''Non-Imperative Functional Programming'' by Nobuo Yamashita:
As for those extensions - they stay with each definition.
 
 
That still didn't work? Well, give this a try:
 
   
  +
:{|
 
<haskell>
 
<haskell>
yet :: (a -> a) -> a
+
type a :-> b = OI a -> b
yet f = y where y = f y
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* [http://h2.jaguarpaw.co.uk/posts/mtl-style-for-free MTL style for free] by Tom Ellis:
Now that we're firmly on the topic of implementation details, did you notice how easy it was to define that allegedly ''warm, fuzzy''[[#refs|[4]]] <code>IO</code> type using this curious new <code>OI</code> type, and those primitives?
 
 
Sometimes that can be a hint that doing the opposite will be difficult or even impossible while staying within standard Haskell 2010. As it happens, this is one of those cases...
 
 
To define <code>OI</code>, <code>partOI</code>, <code>getchar</code> and <code>putchar</code> will require:
 
 
* modifying your preferred Haskell implementation - lots of work;
 
 
* using some other language for the definitions, with Haskell then calling the foreign code - extra work to deal with two different languages;
 
 
* using unsafe or implementation-specific primitives - work needed to avoid conflicts with Haskell semantics;
 
 
* using implementation-specific extensions - work needed to track relevant extensions, and possible conflicts with Haskell semantics.
 
 
For now, I'll just use the extensions - they're ugly, but at least they'll be contained, as they are in those alternate definitions of <code>seq</code>. But who knows - if this approach to I/O proves useful enough, it might make its way into a future Haskell standard...that's how <code>IO</code> happened[[#refs|[5]]].
 
 
In the meantime, take a very deep breath:
 
   
  +
:{|
 
<haskell>
 
<haskell>
  +
data Time_ a = GetCurrentTime (UTCTime -> a)
-- for GHC 8.6.5
 
{-# LANGUAGE MagicHash, UnboxedTuples #-}
 
module OutputInput(OI, Monomo, runOI, invokes, seq) where
 
import Prelude(Bool, Char, Double, Either, Float, Int, Integer, Maybe)
 
import Prelude(String, Eq(..))
 
import Prelude(($), (++), error, all)
 
import Control.Concurrent(ThreadId, MVar, Chan, QSem, QSemN)
 
import Control.Concurrent.STM(STM, TVar, TMVar, TChan, TQueue, TBQueue)
 
import Control.Concurrent.STM(TArray)
 
import Control.Monad.ST(ST)
 
import Data.Array(Array)
 
import Data.Array.IO(IOArray)
 
import Data.Array.ST(STArray)
 
import Data.Char(isSpace)
 
import Data.IORef(IORef)
 
import Data.STRef(STRef)
 
import Data.Time(UTCTime, NominalDiffTime, Day, TimeOfDay)
 
import Data.Time(LocalTime, TimeZone, ZonedTime)
 
import Data.Time(DiffTime)
 
import Data.Time(UniversalTime)
 
import System.Directory(XdgDirectory, XdgDirectoryList, Permissions)
 
import System.IO(Handle, IOMode, BufferMode, SeekMode, HandlePosn)
 
import System.IO(TextEncoding, Newline, NewlineMode)
 
import Partible
 
import Sequential
 
import GHC.Base(IO(..), State#, MutVar#, RealWorld)
 
import GHC.Base(seq#, newMutVar#, atomicModifyMutVar#, noDuplicate#)
 
 
data OI = OI OI#
 
 
instance Partible OI where
 
part = partOI
 
 
partOI :: OI -> (OI, OI)
 
partOI (OI h) = case part# h of (# h1, h2 #) -> (OI h1, OI h2)
 
 
runOI :: (OI -> a) -> IO a
 
runOI g = IO $ \s -> case dispense# s of
 
(# s', h #) -> seq# (g (OI h)) s'
 
 
invokes :: Monomo a => String -> IO a -> OI -> a
 
(name `invokes` IO act) (OI h)
 
= (name `invokes#` act) h
 
 
class Monomo a
 
 
-- local definitions --
 
--
 
type OI# = String -> State# RealWorld
 
 
part# :: OI# -> (# OI#, OI# #)
 
part# h = case h "partOI" of
 
s -> case dispense# s of
 
(# s', h1 #) ->
 
case dispense# s' of
 
(# _, h2 #) -> (# h1, h2 #)
 
 
dispense# :: IO# OI#
 
dispense# s = case newMutVar# () s of
 
(# s', r #) -> (# s', expire# s' r #)
 
 
expire# :: State# s -> MutVar# s () -> String -> State# s
 
expire# s r name = case atomicModifyMutVar# r use s of
 
(# s', () #) -> s'
 
where
 
use x = (error nowUsed, x)
 
nowUsed = name' ++ ": already expired"
 
name' = if all isSpace name then "(unknown)"
 
else name
 
 
invokes# :: Monomo a => String -> IO# a -> OI# -> a
 
(name `invokes#` act) h = case act (noDuplicate# (h name)) of (# _, t #) -> t
 
 
type IO# a = State# RealWorld -> (# State# RealWorld, a #)
 
 
-- supplemental instances --
 
--
 
instance (Monomo a, Monomo b) => Monomo (Array a b)
 
instance Monomo Bool
 
instance Monomo BufferMode
 
instance Monomo Char
 
instance Monomo a => Monomo (Chan a)
 
instance Monomo Day
 
instance Monomo DiffTime
 
instance Monomo Double
 
instance (Monomo a, Monomo b) => Monomo (Either a b)
 
instance Monomo Float
 
instance Monomo Handle
 
instance Monomo HandlePosn
 
instance Monomo Int
 
instance Monomo Integer
 
instance Monomo (IO a)
 
instance (Monomo a, Monomo b) => Monomo (IOArray a b)
 
instance Monomo IOMode
 
instance Monomo a => Monomo (IORef a)
 
instance Monomo a => Monomo [a]
 
instance Monomo LocalTime
 
instance Monomo a => Monomo (Maybe a)
 
instance Monomo a => Monomo (MVar a)
 
instance Monomo Newline
 
instance Monomo NewlineMode
 
instance Monomo NominalDiffTime
 
instance Monomo Permissions
 
instance Monomo QSem
 
instance Monomo QSemN
 
instance Monomo SeekMode
 
instance Monomo (ST s a)
 
instance (Monomo a, Monomo b) => Monomo (STArray s a b)
 
instance Monomo (STM a)
 
instance Monomo a => Monomo (STRef s a)
 
instance Monomo TextEncoding
 
instance Monomo ThreadId
 
instance Monomo TimeOfDay
 
instance Monomo TimeZone
 
instance (Monomo a, Monomo b) => Monomo (TArray a b)
 
instance Monomo a => Monomo (TBQueue a)
 
instance Monomo a => Monomo (TChan a)
 
instance Monomo a => Monomo (TMVar a)
 
instance Monomo a => Monomo (TQueue a)
 
instance (Monomo a, Monomo b, Monomo c, Monomo d, Monomo e, Monomo f) => Monomo (a, b, c, d, e, f)
 
instance (Monomo a, Monomo b, Monomo c, Monomo d, Monomo e) => Monomo (a, b, c, d, e)
 
instance (Monomo a, Monomo b, Monomo c, Monomo d) => Monomo (a, b, c, d)
 
instance (Monomo a, Monomo b, Monomo c) => Monomo (a, b, c)
 
instance (Monomo a, Monomo b) => Monomo (a, b)
 
instance Monomo a => Monomo (TVar a)
 
instance Monomo ()
 
instance Monomo UniversalTime
 
instance Monomo UTCTime
 
instance Monomo XdgDirectory
 
instance Monomo XdgDirectoryList
 
instance Monomo ZonedTime
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* [http://h2.jaguarpaw.co.uk/posts/impure-lazy-language An impure lazy programming language], also by Tom Ellis:
Now you can start breathing again :-)
 
   
  +
:{|
 
<haskell>
 
<haskell>
  +
data IO a = IO (() -> a)
module Partible where
 
import Data.Array
 
import Data.List
 
 
class Partible a where
 
part :: a -> (a, a)
 
parts :: a -> [a]
 
 
-- Minimal complete definition: part or parts
 
part u = case parts u of u1:u2:_ -> (u1, u2)
 
parts u = case part u of (u1, u2) -> u1 : parts u2
 
 
 
instance (Ix a, Partible b) => Partible (Array a b) where
 
part arr = case unzip (map part' (assocs arr)) of
 
(al1, al2) -> (new al1, new al2)
 
where
 
new = array (bounds arr)
 
part' (i, u) = case part u of
 
(u1, u2) -> ((i, u1), (i, u2))
 
 
instance (Partible a, Partible b) => Partible (Either a b) where
 
parts (Left u) = map Left (parts u)
 
parts (Right v) = map Right (parts v)
 
 
instance (Partible a, Partible b, Partible c, Partible d, Partible e) => Partible (a, b, c, d, e) where
 
parts (u, v, w, x, y) = zipWith5 (,,,,) (parts u) (parts v) (parts w) (parts x) (parts y)
 
 
instance (Partible a, Partible b, Partible c, Partible d) => Partible (a, b, c, d) where
 
parts (u, v, w, x) = zipWith4 (,,,) (parts u) (parts v) (parts w) (parts x)
 
 
instance (Partible a, Partible b, Partible c) => Partible (a, b, c) where
 
parts (u, v, w) = zipWith3 (,,) (parts u) (parts v) (parts w)
 
 
instance (Partible a, Partible b) => Partible (a, b) where
 
parts (u, v) = zipWith (,) (parts u) (parts v)
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* page 2 of [https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.128.9269&rep=rep1&type=pdf Unique Identifiers in Pure Functional Languages] by Péter Diviánszky:
If you remember, I dispensed with an up-front explanation to try something different. Now that you've
 
  +
:{|
seen just how different this all is, here's the explanation...
 
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
 
  +
[...] The type <code>Id</code> can be hidden by the synonym data type
That abstract <code>partOI</code> and its overloaded associates <code>part</code> and <code>parts</code>? They help an optimising Haskell implementation to determine when it's safe to use those optimisations. Consider this definition:
 
  +
<pre>
 
  +
:: Create a :== Id -> a
  +
</pre>
  +
</div>
  +
<sup> </sup>
 
<haskell>
 
<haskell>
testme n = n^2 + n^2
+
type Create a = Id -> a
 
</haskell>
 
</haskell>
  +
|}
   
  +
* page 7 of [https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.701.930&rep=rep1&type=pdf Functional Reactive Animation] by Conal Elliott and Paul Hudak:
One simple optimisation would be to replace the duplicates of <code>n^2</code> with a single, shared local definition:
 
  +
:{|
 
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
  +
An early implementation of Fran represented behaviors as implied in the formal semantics:
 
<haskell>
 
<haskell>
  +
data Behavior a = Behavior (Time -> a)
testme n = let x = n^2 in x + x
 
 
</haskell>
 
</haskell>
  +
</div>
  +
|}
   
  +
* page 26 of [https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.91.3579&rep=rep1&type=pdf How to Declare an Imperative] by Philip Wadler:
This definition:
 
  +
:{|
 
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
  +
The type <code>'a io</code> is represented by a function expecting a dummy argument of type <code>unit</code> and returning a value of type <code>'a</code>.
  +
<pre>
  +
type 'a io = unit -> a
  +
</pre>
  +
</div>
  +
<sup> </sup>
 
<haskell>
 
<haskell>
  +
type Io a = () -> a
main' u = putchars "ha" u `seq` putchars "ha" u
 
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* The [https://www.vex.net/~trebla/haskell/IO.xhtml Haskell I/O Tutorial] by Albert Lai:
would likewise be rewritten, with the result being:
 
  +
:{|
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
  +
But I can already tell you why we cannot follow other languages and use simply <code>X</code> or <code>() -> X</code>.
  +
</div>
  +
|}
   
  +
* [http://comonad.com/reader/2011/free-monads-for-less-3 Free Monads for Less (Part 3 of 3): Yielding IO] by Edward Kmett:
  +
:{|
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
 
<haskell>
 
<haskell>
  +
newtype OI a = forall o i. OI (FFI o i) o (i -> a) deriving Functor
main' u = let x = putchars "ha" u in x `seq` x
 
 
</haskell>
 
</haskell>
  +
</div>
 
  +
<sup> </sup>
but, as noted by Philip Wadler[[#refs|[6]]]:
 
 
<blockquote>''[...] the laugh is on us: the program prints only a single <code>"ha"</code>, at the time variable <br><code>x</code> is bound. In the presence of side effects, equational reasoning in its simplest form <br>becomes invalid.''</blockquote>
 
 
''Equational reasoning'' is the basis for that simple optimisation and many others in implementations like GHC - so far they've been serving us quite well.
 
 
What - just treat I/O-centric definitions as some special case by modifying GHC? Haskell implementations are complicated enough as is!
 
 
The problem is being caused by the code being treated as though it's pure, so let's modify the code instead. In this case, one simple solution is to make all calls to I/O-centric definitions unique:
 
 
 
<haskell>
 
<haskell>
main u = case part u of
+
type Oi a = forall i . i -> a
(u1, u2) ->
 
putchars "ha" u1 `seq` putchars "ha" u2
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* page 27 of [https://blog.higher-order.com/assets/scalaio.pdf Purely Functional I/O in Scala] by Rúnar Bjarnason:
But what about:
 
  +
:{|
 
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
  +
<pre>
  +
class IO[A](run: () => A)
  +
</pre>
  +
</div>
  +
<sup> </sup>
 
<haskell>
 
<haskell>
oops g h u = g u `seq` h u
+
class Io a where run :: () -> a
 
main' = oops (putchars "ha") (putchars "ha")
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* [https://stackoverflow.com/questions/6647852/haskell-actual-io-monad-implementation-in-different-language/6706442#6706442 ysdx's answer] to [https://stackoverflow.com/questions/6647852/haskell-actual-io-monad-implementation-in-different-language this SO question]:
Will the laugh be on us, again?
 
  +
:{|
 
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
This is Haskell, not Clean[[#refs|[7]]] - there are no uniqueness types to help fend off such potentially-troublesome expressions. For now, the simplest way to make sure <code>OI</code> values are only used once is to have the implementation treat their reuse as being invalid e.g. by throwing an exception or raising an error to stop the offending program.
 
  +
Let's say you want to implement <code>IO</code> in SML :
 
  +
<pre>
In the prototype implementation, this all-important ''single-use'' property is maintained by <code>expire#</code>.
 
  +
structure Io : MONAD =
 
  +
struct
As for that curious <code>Monomo</code> class and its instances, they leverage Haskell's type system to provide an extra measure of safety for the prototype - an actual implementation would instead use an annotation[[#refs|[8]]] to achieve the same result e.g:
 
  +
type 'a t = unit -> 'a
 
  +
  +
end
  +
</pre>
  +
</div>
  +
<sup> </sup>
 
<haskell>
 
<haskell>
  +
type T a = () -> a
newEmptyMVar :: monomo a . OI -> MVar a
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* [https://stackoverflow.com/questions/45136398/is-the-monadic-io-construct-in-haskell-just-a-convention/45141523#45141523 luqui's answer] to [https://stackoverflow.com/questions/45136398/is-the-monadic-io-construct-in-haskell-just-a-convention this SO question]:
Now for the much-maligned[[#refs|[9][10]]] <code>seq</code>...you could be tempted into avoiding it by using a new data type:
 
  +
:{|
 
<haskell>
+
|<haskell>
newtype Result a = Is a
+
newtype IO a = IO { runIO :: () -> a }
 
getchar' :: OI -> Result Char
 
putchar' :: Char -> OI -> Result ()
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* [https://stackoverflow.com/questions/15418075/the-reader-monad/15419592#15419592 luqui's answer] to [https://stackoverflow.com/questions/15418075/the-reader-monad this SO question]:
and case-expressions:
 
  +
:{|
 
<haskell>
+
|<haskell>
respond' :: Request -> OI -> Response
+
newtype Supply r a = Supply { runSupply :: r -> a }
respond' Getq = \u -> case getchar' u of Is c -> Getp c
 
respond' (Putq c) = \u -> case putchar' c u of Is _ -> Putp
 
 
</haskell>
 
</haskell>
  +
|}
   
  +
* [https://stackoverflow.com/questions/51770808/how-exactly-does-ios-work-under-the-hood/51772273#51772273 chi's answer] to [https://stackoverflow.com/questions/51770808/how-exactly-does-ios-work-under-the-hood this SO question]:
But before you succumb:
 
  +
:{|
  +
|<div style="border-left:1px solid lightgray; padding: 1em" alt="blockquote">
  +
As long as we have its special case <code>IO c = () ~> c</code>, we can represent (up to isomorphism) […] <code>a ~> c</code> […]
  +
</div>
  +
|}
   
  +
Of these, it is the [https://hackage.haskell.org/package/oi/docs/src/Data-OI-Internal.html#OI implementation of <code>OI a</code>] in Yamashita's [https://hackage.haskell.org/package/oi oi] package which is most interesting as its values are ''monousal'' - once used, their contents remain constant. This single-use property also appears in the implementation of the abstract <code>decision</code> type described by F. Warren Burton in [https://academic.oup.com/comjnl/article-pdf/31/3/243/1157325/310243.pdf Nondeterminism with Referential Transparency in Functional Programming Languages].
<haskell>
 
unit_Result :: a -> Result a
 
unit_Result = Is
 
   
  +
----
bind_Result :: Result a -> (a -> Result b) -> Result b
 
  +
=== <code>IO</code><u>, redefined</u> ===
bind_Result (Is x) k = k x
 
</haskell>
 
   
  +
Based on these and other observations, a reasonable distillment of these examples would be <code>OI -> a</code>, which then implies:
Oh look - <code>Result</code> is one of '''those''' types[[#refs|[11]]]...
 
 
The bang-pattern[[#refs|[12]]] extension? So you can instead write:
 
   
 
<haskell>
 
<haskell>
  +
type IO a = OI -> a
respond'' :: Request -> OI -> Response
 
respond'' Getq = \u -> let !c = getchar u in Getp c
 
respond'' (Putq c) = \u -> let !z = putchar c u in Putp
 
 
</haskell>
 
</haskell>
   
  +
Using Burton's ''pseudodata'' approach:
As you can see, <code>z</code> isn't used anywhere - there is no need for it. This being Haskell, if it isn't needed, it normally isn't evaluated. For now, the bang-pattern extension modifies the evaluation of
 
<code>z</code> in order to prevent <code>respond&apos;&apos;</code> being rewritten as:
 
   
 
<haskell>
 
<haskell>
  +
-- abstract; single-use I/O-access mediator
respond'' :: Request -> OI -> Response
 
  +
data Exterior
respond'' Getq = \u -> let !c = getchar u in Getp c
 
  +
getchar :: Exterior -> Char
respond'' (Putq c) = \u -> Putp
 
  +
putchar :: Char -> Exterior -> ()
</haskell>
 
   
  +
-- from section 2 of Burton's paper
Will bang-patterns ever be included in a future Haskell standard? If so, will you still be able to use them like this? If not, will you be left with the clean-up job?
 
  +
data Tree a = Node { contents :: a,
  +
left :: Tree a,
  +
right :: Tree a }
   
  +
-- utility definitions
Perhaps you'll find some other way for correctly sequencing the evaluation that you like; all well and good...but others might not. For me, the simplest way I've found to make this approach to I/O work is with <code>seq</code> - one that's actually sequential.
 
  +
type OI = Tree Exterior
   
  +
getChar' :: OI -> Char
But maybe - after all that - you still want <code>seq</code> banished from Haskell. Perhaps you still don't understand I/O in Haskell. It could be that you're dismayed by what you've read here. Alternately, you may have seen or tried this all before, and know it doesn't work - darn...
 
  +
getChar' = getchar . contents
   
  +
putChar' :: Char -> OI -> ()
If that's you, the corresponding language proposal[[#refs|[13]]] has a list of other articles and research papers I've found which describe or refer to alternative approaches - perhaps one (or more) of them will be more acceptable.
 
  +
putChar' c = putchar c . contents
   
  +
part :: OI -> (OI, OI)
As noted by Owen Stephens[[#refs|[14]]]:
 
  +
parts :: OI -> [OI]
   
  +
part t = (left t, right t)
<blockquote>''I/O is not a particularly active area of research, but new approaches are still being discovered, <br>iteratees being a case in point.''</blockquote>
 
  +
parts t = let !(t1, t2) = part t in
  +
t1 : parts t2
  +
</haskell>
   
  +
Of course, in an actual implementation <code>OI</code> would be abstract like <code>World</code>, and for similar reasons. This permits a simpler implementation for <code>OI</code> and its values, instead of being based on (theoretically) infinite structured values like binary trees. That simplicity has benefits for the <code>OI</code> interface, in this case:
Who knows - the Haskell language could return to having a pure, fully-defined approach to I/O...and it could be you that finds it :-D
 
   
  +
<haskell>
  +
data OI
  +
part :: OI -> (OI, OI)
  +
getChar' :: OI -> Char
  +
putChar' :: Char -> OI -> ()
  +
</haskell>
  +
<sup> </sup>
   
  +
----
P.S: Why the name <code>OI</code>? Many years ago I was tinkering with arrows for performing I/O, labelling them <code>OI a b</code> out of expediency. More recently, I discovered a set of slides[[#refs|[15]]] describing another approach to I/O which used values of type <code>OI a</code> in a similar fashion to what I've been describing here. I've reused the name because of that similarity.
 
 
 
<span id="refs">References</span>:
 
 
[1] [[Sequential ordering of evaluation]]; Haskell Wiki.<br>
 
 
[2] [https://gitlab.haskell.org/ghc/ghc/-/issues/5129 Ticket# 5129: "evaluate" optimized away]; GHC bug tracker.<br>
 
   
  +
=== <u>Various questions</u> ===
[3] [https://mail.haskell.org/pipermail/glasgow-haskell-users/2006-November/011480.html Thread: seq vs. pseq]; Haskell mail archive.<br>
 
   
  +
* Is the C language "purely functional"?
[4] [https://www.cs.nott.ac.uk/~pszgmh/appsem-slides/peytonjones.ppt Wearing the hair shirt: a retrospective on Haskell]; Simon Peyton Jones.<br>
 
   
  +
::No:
[5] [https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.168.4008&rep=rep1&type=pdf A History of Haskell: being lazy with class]; Paul Hudak, John Hughes, Simon Peyton Jones and Philip Wadler.<br>
 
  +
::* C isn't "pure" - it allows unrestricted access to observable effects, including those of I/O.
  +
::* C isn't "functional" - it was never intended to be [[Referential transparency|referentially transparent]], which severely restricts the ability to use [[Equational reasoning examples|equational reasoning]].
   
  +
* Is the Haskell language "purely functional"?
[6] [https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.91.3579&rep=rep1&type=pdf How to Declare an Imperative]; Philip Wadler.<br>
 
   
  +
::[https://chadaustin.me/2015/09/haskell-is-not-a-purely-functional-language Haskell is not a purely functional language], but is often described as being referentially transparent.
[7] [https://clean.cs.ru.nl/Clean The Clean homepage]; Radboud University, Nijmegen, The Netherlands.<br>
 
   
  +
* Can functional programming be liberated from the von Neumann paradigm?
[8] [[Monomorphism by annotation of type variables]]; Haskell Wiki.<br>
 
   
  +
::That remains an [[Open research problems|open research problem]].
[9] [https://mail.haskell.org/pipermail/haskell/2002-May/009622.html Thread: State monads don't respect the monad laws in Haskell]; Haskell mail archive.<br>
 
   
  +
* Can a language be "purely functional" or "denotative"?
[10] [http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.71.1777&rep=rep1&type=pdf The Impact of ''seq'' on Free Theorems-Based Program Transformations]; Patricia Johann and Janis Voigtlander.
 
   
  +
::Conditionally, yes - the condition being the language is restricted in what domains it can be used in:
[11] [[Monad]]; Haskell Wiki.<br>
 
   
  +
::* If a language is free of observable effects, including those of I/O, then the only other place where those effects can reside is within its implementation.
[12] [https://downloads.haskell.org/~ghc/7.8.4/docs/html/users_guide/bang-patterns.html 7.18. Bang patterns]; GHC user's guide.<br>
 
  +
::* There is no bound on the ways in which observable effects can be usefully combined, leading to a similarly-unlimited variety of imperative computations.
  +
::* A finite implementation cannot possibly accommodate all of those computations, so a subset of them must be chosen. This restricts the implementation and language to those domains supported by the chosen computations.
   
  +
* Why do our programs need to read input and write output?
[13] [[Partibles for composing monads]]; Haskell Wiki.<br>
 
   
  +
::Because programs are usually written for [https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.628.7053&rep=rep1&type=pdf practical] purposes, such as implementing domain-specific [https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.7.2089&rep=rep1&type=pdf little languages] like [https://dhall-lang.org Dhall].
[14] [https://www.owenstephens.co.uk/assets/static/research/masters_report.pdf Approaches to Functional I/O]; Owen Stephens.<br>
 
   
  +
----
[15] <span style="color:#ba0000">Non-Imperative Functional Programming</span>; Nobuo Yamashita.<br>
 
   
  +
=== <u>See also</u> ===
   
  +
* [https://pqnelson.github.io/2021/07/29/monadic-io-in-ml.html Monadic IO in Standard ML]
[[User:Atravers|Atravers]] 03:05, 20 August 2020 (UTC)
 
  +
* [[Disposing of dismissives]]
  +
* [[IO then abstraction]]
  +
* [https://okmij.org/ftp/Computation/IO-monad-history.html The IO monad in 1965]

Revision as of 12:36, 14 January 2022


Clearing away the smoke and mirrors

The implementation in GHC uses the following one:

type IO a  =  World -> (a, World)

An IO computation is a function that (logically) takes the state of the world, and returns a modified world as well as the return value. Of course, GHC does not actually pass the world around; instead, it passes a dummy “token,” to ensure proper sequencing of actions in the presence of lazy evaluation, and performs input and output as actual side effects!

A History of Haskell: Being Lazy With Class, Paul Hudak, John Hughes, Simon Peyton Jones and Philip Wadler.

...so what starts out as an I/O action of type:

World -> (a, World)

is changed by GHC to approximately:

() -> (a, ())

As the returned unit-value () contains no useful information, that type can be simplified further:

() -> a

Why "approximately"? Because "logically" a function in Haskell has no observable effects.


Previously seen

The type () -> a (or variations of it) have appeared elsewhere - examples include:

The use of λ, and in particular (to avoid an irrelevant bound variable) of λ() , to delay and possibly avoid evaluation is exploited repeatedly in our model of ALGOL 60. A function that requires an argument-list of length zero is called a none-adic function.

(\ () -> ) :: () -> a
abstype 'a Job = JOB of unit -> 'a

data Job a = JOB (() -> a)

A value of type Obs 𝜏 is called an observer. Such a value observes (i.e. views or inspects) a state and returns a value of type 𝜏. [...] An observer type Obs 𝜏 may be viewed as an implicit function space from the set of states to the type 𝜏.

type Obs tau = State -> tau
  • page 15 of Non-Imperative Functional Programming by Nobuo Yamashita:
type a :-> b = OI a -> b
data Time_ a = GetCurrentTime (UTCTime -> a)
data IO a = IO (() -> a)

[...] The type Id can be hidden by the synonym data type

:: Create a  :==  Id -> a

type Create a = Id -> a

An early implementation of Fran represented behaviors as implied in the formal semantics:

data Behavior a = Behavior (Time -> a)

The type 'a io is represented by a function expecting a dummy argument of type unit and returning a value of type 'a.

type 'a io = unit -> a

type Io a = () -> a

But I can already tell you why we cannot follow other languages and use simply X or () -> X.

newtype OI a = forall o i. OI (FFI o i) o (i -> a) deriving Functor

type Oi a = forall i . i -> a
class IO[A](run: () => A)

class Io a where run :: () -> a

Let's say you want to implement IO in SML :

structure Io : MONAD =
struct
  type 'a t = unit -> 'a
         ⋮
end

type T a = () -> a
newtype IO a = IO { runIO :: () -> a }
newtype Supply r a = Supply { runSupply :: r -> a }

As long as we have its special case IO c = () ~> c, we can represent (up to isomorphism) […] a ~> c […]

Of these, it is the implementation of OI a in Yamashita's oi package which is most interesting as its values are monousal - once used, their contents remain constant. This single-use property also appears in the implementation of the abstract decision type described by F. Warren Burton in Nondeterminism with Referential Transparency in Functional Programming Languages.


IO, redefined

Based on these and other observations, a reasonable distillment of these examples would be OI -> a, which then implies:

type IO a = OI -> a

Using Burton's pseudodata approach:

 -- abstract; single-use I/O-access mediator
data Exterior
getchar :: Exterior -> Char
putchar :: Char -> Exterior -> ()

 -- from section 2 of Burton's paper
data Tree a = Node { contents :: a,
                     left     :: Tree a,
                     right    :: Tree a }

 -- utility definitions
type OI  =  Tree Exterior

getChar' :: OI -> Char
getChar' =  getchar . contents

putChar' :: Char -> OI -> ()
putChar' c = putchar c . contents

part     :: OI -> (OI, OI)
parts    :: OI -> [OI]

part t   =  (left t, right t)
parts t  =  let !(t1, t2) = part t in
            t1 : parts t2

Of course, in an actual implementation OI would be abstract like World, and for similar reasons. This permits a simpler implementation for OI and its values, instead of being based on (theoretically) infinite structured values like binary trees. That simplicity has benefits for the OI interface, in this case:

data OI
part :: OI -> (OI, OI)
getChar' :: OI -> Char
putChar' :: Char -> OI -> ()


Various questions

  • Is the C language "purely functional"?
No:
  • C isn't "pure" - it allows unrestricted access to observable effects, including those of I/O.
  • C isn't "functional" - it was never intended to be referentially transparent, which severely restricts the ability to use equational reasoning.
  • Is the Haskell language "purely functional"?
Haskell is not a purely functional language, but is often described as being referentially transparent.
  • Can functional programming be liberated from the von Neumann paradigm?
That remains an open research problem.
  • Can a language be "purely functional" or "denotative"?
Conditionally, yes - the condition being the language is restricted in what domains it can be used in:
  • If a language is free of observable effects, including those of I/O, then the only other place where those effects can reside is within its implementation.
  • There is no bound on the ways in which observable effects can be usefully combined, leading to a similarly-unlimited variety of imperative computations.
  • A finite implementation cannot possibly accommodate all of those computations, so a subset of them must be chosen. This restricts the implementation and language to those domains supported by the chosen computations.
  • Why do our programs need to read input and write output?
Because programs are usually written for practical purposes, such as implementing domain-specific little languages like Dhall.

See also