Improve documentation for ListT instances

Abierto
  • ¿Algo no está bien?
#117 6 comentarios 0 reacciones 0 asignados Ver en GitHub

Nadie ha tomado este issue todavía.

Evaluación

Dificultad
4/5
Tiempo estimado
3-5 días
Aptitud para principiantes
30/100
Tipo de issue
Documentación
Claridad
Bastante claro
Estado de actividad
Estancado
Área
documentation

Línea de trabajo

Start by reading the ListT instances in the transformers package and compare their effect sequencing with the Compose List Effect examples in the issue. Document which sequencing behavior the instances provide and how it differs in the shown examples; done when the documentation explains the distinction clearly. The issue has had no comment for years, so check whether maintainers still want this work.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

type: documentation

What kind of effect-sequencing behavior is ListT supposed to exhibit? I ask because in haskell, ListT m a is a newtype wrapper of m [a], and if purescript's ListT is meant to be modeled after haskell's, ListT Effect Int should perhaps then behave like Effect [Int]. However, the haskell community has also considered alternative models for ListT like the following:

-- https://github.com/nikita-volkov/list-t/blob/master/library/ListT.hs
newtype ListT m a =  ListT (m (Maybe (a, ListT m a)))

Because ListT in Pursuit's 'transformers' package is defined as ...

newtype ListT f a = ListT (f (Step a (ListT f a)))
data Step a s = Yield a (Lazy s)  | Skip (Lazy s)  | Done

... Volkov's ListT (shown above) seems to be the underlying model for purescript's own ListT.

What kind of effect-sequencing behavior, then, should Volkov's ListT exhibit? Unfortunately, at the moment, I can't experiment with his code directly; however, it seems that, under specific conditions, Volkov's ListT m Int should behave similarly to [m Int] -- in particular, when a list-like construct in either case (ListT m Int or [m Int]) is fully traversed/consumed with pre-order congruence and the final effect of ListT is pure or ignored. (Or at least I assume; perhaps there are details that invalidate this comparison.)

If this comparison is sound, users in purescript-land might expect values of the analogous purescript types ListT Effect Int, List (Effect Int) and Compose List Effect Int, under appropriate conditions, also to exhibit identical behavior. I'm not sure that's the case, however.

Consider, for example, product0 and product1 below:

import Control.Monad.List.Trans as T
-- other imports ...

io :: forall a b. Show b => b -> a -> Effect a
io x y = log (show x) *> pure y

-- `list0` and `list1` below are meant to have congruent sequences of data AND effects.

list0 :: ListT Effect Int
list0 = fromEffect (io 2 2) <|> fromEffect (io 1 1) <|> fromEffect (io 0 0) <|> T.nil

list1 :: Compose List Effect Int
list1 = Compose $ pure (io 2 2) <|> pure (io 1 1) <|> pure (io 0 0) <|> Nil

f0 :: ListT Effect (Int -> Int)
f0 = fromEffect (io "id" identity)

f1 :: Compose List Effect (Int -> Int)
f1 = Compose (pure (io "id" identity))

product0 = f0 <*> list0 :: ListT Effect Int
product1 = f1 <*> list1 :: Compose List Effect Int

If product0 and product1 are fully traversed, results like the following occur:

sequenceListT :: forall a. ListT Effect a -> Effect (List a)
sequenceListT = foldl' (\x y -> pure (x <> pure y)) Nil

sequenceCompose :: forall a. Compose List Effect a -> Effect (List a)
sequenceCompose = sequence <<< unwrap

-- PROMPT> sequenceListT product0
-- "id"
-- 2
-- 1
-- 0
-- (2 :  1 : 0 : Nil)

-- PROMPT> sequenceCompose product1
-- "id"
-- 2
-- "id"
-- 1
-- "id"
-- 0
-- (2 : 1 : 0 : Nil)

Other helpful comparisons to try might include the following:

product2 = T.nil <*> list0 :: ListT Effect Int
product3 = pure Nil <*> list1 :: Compose List Effect Int
product4 = Backwards f0 <*> Backwards list0 :: Backwards (ListT Effect) Int
product5 = Backwards f1 <*> Backwards list1 :: Backwards (Compose List Effect) Int

Although I've tried to model the same sequence of bundled data and effects as both an instance of ListT Effect Int and as an instance of Compose List Effect Int and although I (myself at least) expect similar behavior because of my assumptions listed above, the two values product0 and product1 have different overall effects when traversed.

The source of this difference in behavior appears to be related to the Apply typeclass. Whereas List's apply member has a static flavor, ListT's implementation of apply is monadic and therefore directionally biased.

I'm not sure which style of applicative multiplication might be generally preferable. I'm not even sure that List and ListT should have exactly identical effect-sequencing semantics (in relevant scenarios). However, because this difference in behavior surprised me at least, I figured I'd note this distinction as an issue for maintainers to consider. Perhaps it's a point confusing/interesting enough to include in documentation.

Thanks much

Lenguaje dominante
PureScript
Estrellas
72
Forks
43
Merge medio
2 d 23 h
PR fusionados (30 d)
1

Preparar el entorno

  • Sin Dockerfile ni archivo de Docker Compose
  • Tiene una plantilla de pull request
  • Sin guía de contribución

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de purescript/purescript-transformers

Todos los issues de purescript/purescript-transformers

Issues similares

Más issues de Documentation

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.