Hi there, I'm bumping into a type error I can't make sense of when making a record of natural transformations. I've made a gist which I think is a minimal reproduction : https://try.purescript.org/?gist=404eefaba98d2e42c16fcd333487ce8b
It works with a simple data constructor
Nat m n = Nat (m ~> n) but not with a record NatR m n = Nat { nat :: (m ~> n) }
use :: forall m n. Nat m n -> m Unit -> n Unit
use (Nat nat) = nat
useR :: forall m n. NatR m n -> m Unit -> n Unit
useR (NatR { nat }) = nat -- Error!useR: Could not match type
a2
with type
Unit
while trying to match type m0 a2
with type m0 Unit
while checking that expression nat
has type m0 Unit -> n1 Unit
in value declaration useR
where n1 is a rigid type variable
bound at (line 0, column 0 - line 0, column 0)
m0 is a rigid type variable
bound at (line 0, column 0 - line 0, column 0)
a2 is a rigid type variable
bound at (line 17, column 34 - line 17, column 54)@Ebmtranceboy Thanks, neat trick! I had to extend my minimal example because unfortunately, in my case it still didn't work. https://try.purescript.org/?gist=c544c34971cecfd61528fa06c996460b
data Nat m n = Nat (m ~> n)
data NatR m n = NatR { nat :: (m ~> n) }
morphS :: forall m n. Monad m => Monad n => (m ~> n) -> m Unit -> n Unit
morphS t fm = t fm
-- useR :: forall m n. Monad m => Monad n => NatR m n -> m Unit -> n Unit
-- useR (NatR { nat }) x = morphS nat x -- Error!
-- Error: Could not match type a4 with a1
-- a4 rigid type variable bound at line 17, column 34-54...
-- Stays at line 17 even if I move the code around.
-- Actually it's also the same location in my local code base... Wat?
useR' :: forall m n. Monad m => Monad n => NatR m n -> m Unit -> n Unit
useR' (NatR h) x = morphS h.nat x -- Works
useR'' :: forall m n. Monad m => Monad n => NatR m n -> m Unit -> n Unit
useR'' (NatR h) = morphS h.nat -- Pointfree also worksThere's also some added strangeness going on with the code location reported by the type checker... Should some of this be documented as gotchas somewhere so that they're googleable?
@jmatsushita I'm not sure where is heading your investigations. The compiler error you mention goes away if you replace
morphS with identity in the useR definition but you probably have something else in mind :) I advise you to ask on the functional programming Slack, channel #purescript or #purescript-beginners. They're quite crowded
Hi everyone hope you are well. How do I import
Nothing from Data.Maybe I am getting an error as shown above
The
(..) asks to not only import the type, but also the type constructors (Just and Nothing)
I have a type system question.
I can express head as a natural transformation like so:
head' ∷ List ~> Maybe
head' = headMy intuition is that join is also a natural transformation of the type forall m. m m ~> m
But if the kind of m :: Type -> Type then m m won't compile because the kind of the type argument to the outer m is type -> type and it expects the kind type
join' ∷ forall (m :: Type → Type) . Bind m ⇒ m m ~> m
join' = join
join' ∷ forall (m :: Type → Type) . Bind m ⇒ m m ~> m
^
Could not match kind
Type
with kind
Type -> TypeI also tried using a type lambda there to get the the type of the inner m to be correct
43 join'' ∷ forall (m :: Type → Type) . Bind m ⇒ (∀ a. a → m (m a)) ~> m
^^^^^^^^^^^^^^^^
Could not match kind
Type -> Type
with kind
TypeWhich results in a different type error. Which I am very confused by, the compiler seems to be expect the underlined section to have the kind type but I would expect that it should be type -> type because it's on the left of ~>
Does anyone know how I can express join as NaturalTransformation or am I just completely lost.
(never seen that syntax before)
module Main where
import Data.Array
import Data.Array.ST
import Effect.Ref
import Prelude
import Control.Monad.ST.Internal
import Control.Monad.ST (run, for)
import Control.Monad.ST.Ref as Ref
import Effect (Effect, forE)
import Effect.Class.Console (log, logShow)
import Data.Maybe (Maybe(..), fromMaybe)
import Partial.Unsafe (unsafePartial)
main :: Effect Unit
main = void $ unsafePartial do
let nodes = [
[-100.0, -100.0, -100.0],
[-100.0, -100.0, 100.0],
[-100.0, 100.0, -100.0],
[-100.0, 100.0, 100.0],
[ 100.0, -100.0, -100.0],
[ 100.0, -100.0, 100.0],
[ 100.0, 100.0, -100.0],
[ 100.0, 100.0, 100.0]
]
ref <- empty
void $ forE 0 8 $ \i -> do
let xx = (nodes !! i)
let yy = fromMaybe [] xx
void $ push yy ref
log "x"
here ref is mutable array on doing
push yy ref it's throwing
Could not match type
ST t1
with type
Effect
while trying to match type t0 Unit
with type Effect Unit
while checking that expression (apply void) ((push yy) ref)
has type Effect Unit
in value declaration main
where t0 is an unknown type
t1 is an unknown type
someone help me please
This returns me false „wrapped“ within Conj monoid, how can I get the Boolean out?
let Conj result = Conj false <> Conj true
It appears not (see here https://github.com/purescript/purescript-prelude/blob/v4.1.1/src/Data/Monoid/Conj.purs#L15-L15). You could make a PR and see what the author thinks.
Hey! just started with PS and with JS for that matter. Playing with a JS framework and studying FFI I stumbled onto a weird problem:
first one works just fine, but second one produces a this ~ undefined problem
first one works just fine, but second one produces a this ~ undefined problem
Hooks.on uses this._hooks under the hood so I get an error: "Object._hooks is undefined". However I verified that the right function is found and executed. Does anyone know what can cause such behaviour?
@vlj: in Haskell it’s TypeApplicatons extension which helps to annotate just type variables. regular type annotation is done like this:
(func :: SomeTypeAnnotation) both in Haskell and PureScript
there is no extension in purescript ?
@vlj: i’m not even sure if there are any extensions in PureScript. every feature of the language in turned on already
I'm having some trouble with my record types.
I have a type `
data MeasuredTask = MeasuredTask {
uid :: Int,
threadId :: Int,
linkedParentId :: Int,
label :: String,
nestedTasks :: Array MeasuredTask,
linkedTasks :: Array MeasuredTask,
timeStart :: String,
timeEnd :: String,
cpuStart :: Int,
cpuEnd :: Int,
cpuThread :: Int,
cpuLinked :: Int,
timeLinked :: Int,
lineStart :: String,
lineEnd :: String}It's not "just" a record because it 's inductively defined so I put a MeasuredTask constructor