A way to generalize Haskell's Either type for arbitrarily many types? - haskell

I am creating a turn based game. I want to define a datatype that encodes one type out of many possible types. Here is the motivating example:
I have defined a Turn type using GADTs, so the type of each value of Turn a says something about it's value.
data Travel
data Attack
data Flee
data Quit
data Turn a where
Travel :: Location -> Turn Travel
Attack :: Int -> Turn Attack
Flee :: Turn Flee
Quit :: Turn Quit
Now I can write types like this decideTravel :: GameState -> Turn Travel, very expressive and nice.
The problem arises when I want to return one of multiple possible turn types. I want to write functions similar to the following:
-- OneOf taking two types
decideFightingTurn :: GameState -> OneOf (Turn Attack) (Turn Flee)
-- OneOf takes three types
decideTurn :: GameState -> OneOf (Turn Attack) (Turn Travel) (Turn Quit)
Where this OneOf datatype carries the notion of "one type out of many possible types". The problem lies in defining this datatype, I need to somehow handle a list of types at the type level.
There are two subpar solutions I have as of now:
Solution 1: Create a wrapper sum type
Simply create a new type that has a constructor for each Turn a constructor:
data AnyTurn
= TravelTurn (Turn Travel)
| TravelAttack (Turn Attack)
| TravelFlee (Turn Flee)
| TravelQuit (Turn Quit)
This doesn't help me in the way I want it to. Now I have to pattern match all cases of AnyTurn and account for invalid input types. Additionally the type level information is obscured by the AnyTurn type, since it fails to indicate which specific turns are actually possible at the type level.
Solution 2: Create "Either" types for different numbers of types
This solution gives me what I want at the type level, but is cumbersome to use. Basically create an Either-like type for any number of combinations, like so:
data OneOf2 a b
= OneOf2A a
| OneOf2B b
data OneOf3 a b c
= OneOf3A a
| OneOf3B b
| OneOf3C c
-- etc, for as many as are needed.
This conveys what I want at the type level, so I can now write the previous examples as:
decideFightingTurn :: GameState -> OneOf2 (Turn Travel) (Turn Flee)
decideTurn :: GameState -> OneOf3 (Turn Attack) (Turn Travel) (Turn Quit)
This works, however it would be nice to express this with just one type OneOf. Is there any way to write the generalized OneOf type in Haskell?

Something like this, I guess:
data OneOf as where
ThisOne :: a -> OneOf (a : as)
Later :: OneOf as -> OneOf (a : as)
Then you can write, for example:
decideFightingTurn :: GameState -> OneOf [Turn Travel, Turn Flee]
You might also like to:
type family Map f xs where
Map f '[] = '[]
Map f (x:xs) = f x : Map f xs
This way you can write something like:
decideFightingTurn :: GameState -> OneOf (Map Turn [Travel, Flee])
Or you could build the Map into the GADT if you think you'll always be doing it:
data OneOfF f as where
ThisOneF :: f a -> OneOfF f (a : as)
LaterF :: OneOfF f as -> OneOfF f (a : as)
And then:
decideFightingTurn :: GameState -> OneOfF Turn [Travel, Flee]
Various things can be done to make this more efficient if that becomes a concern; the first thing I'd try would be using binary indexing rather than unary as shown here.

There's a Haskell library called row-types that might give you what you want. Specifically, it provides an extensible variant that is kind of like Either but tagged and with arbitrarily many options.
In your case, you could make your types:
import Data.Row
decideFightingTurn :: GameState -> Var ("travel" .== Turn Travel .+ "flee" .== Turn Flee)
decideTurn :: GameState -> Var ("attack" .== Turn Attack .+ "travel" .== Turn Travel .+ "quit" .== Turn Quit)
The easiest way to work with variants is to turn OverloadedLabels on. Then, to create one, you write:
decideFightingTurn gs = case gs of
Foo -> IsJust #travel $ Travel loc
Bar -> IsJust #flee Flee
Deconstructing can be done in a few different ways. If you just want to check for a particular type, you can use trial or view (exported by Data.Row.Variants), but if you want to deal with all the options at once, you may want to use switch. You'd write something like this:
handleTravelOrFlee :: Var ("travel" .== Turn Travel .+ "flee" .== Turn Flee) -> GameState -> GameState
handleTravelOrFlee v gs = switch v $
#travel .== (\(Travel loc) -> updateGameStateWithLoc gs loc)
.+ #flee .== (\Flee -> updateGameStateFlee gs)
Check out the documentation and the examples file for more info.

Another option could be activating and deactivating constructors in the sum type:
{-# LANGUAGE GADTs, TypeOperators, DataKinds, PolyKinds,
TypeFamilies, StandaloneKindSignatures #-}
{-# OPTIONS_GHC -Werror=incomplete-patterns -Werror=overlapping-patterns #-}
import Data.Void
import Data.Kind
data TurnType =
Travel
| Attack
| Flee
| Quit
type Turn :: TurnType -> Type
data Turn tt where
TurnTravel :: Location -> Turn Travel
TurnAttack :: Int -> Turn Attack
TurnFlee :: Turn Flee
TurnQuit :: Turn Quit
type X :: TurnType -> [TurnType] -> Type
type family X x tts :: Type where
X _ '[] = Void
X x (x : xs) = ()
X x (_ : xs) = X x xs
-- Exhaustiveness checker knows that branches where
-- X returns Void are impossible.
type AnyTurn :: [TurnType] -> Type
data AnyTurn allowed
= TravelTurn !(X Travel allowed) (Turn Travel)
| TravelAttack !(X Attack allowed) (Turn Attack)
| TravelFlee !(X Flee allowed) (Turn Flee)
| TravelQuit !(X Quit allowed) (Turn Quit)
Putting it to work:
someTurn :: AnyTurn '[Travel, Attack]
someTurn = TravelAttack () (TurnAttack 0)
-- This doesn't give an exhaustiveness error because other branches are impossible.
-- The problem is that we now have those annoying () cluttering every match.
turn2string :: AnyTurn '[Travel, Attack] -> String
turn2string t = case t of
TravelTurn () _ -> "foo"
TravelAttack () _ -> "bar"
-- TravelQuit _ _ -> "doesn't compile"

Related

Generics : run-time ADT for types with instances

Is it possible with Haskell / GHC, to extract an algebraic data type representing all types with Eq and Ord instances ? This would probably need Generics, Typeable, etc.
What I would like is something like :
data Data_Eq_Ord = Data_String String
| Data_Int Int
| Data_Bool Bool
| ...
deriving (Eq, Ord)
For all types known to have instances for Eq and Ord. If it makes the solution easier, we can limit our scope to Ord instances, since Eq is implied by Ord. But is would be interesting to know if constraints intersection is possible.
This data type would be useful because it gives the possibility to use it where Eq and Ord constraints are required, and pattern-match at runtime to refine on types.
I would need this to implement a generic Map Key Value, where Key would be this type, in a Document Indexing library, where the keys and their type is known at run-time. This library is here. For the moment I worked around the issue by defining a data DocIndexKey, and a FieldKey class, but this is not fully satisfactory since it requires boilerplate and can't cover all legit candidates.
Any good alternative approach to this situation is welcome. For some reasons, I prefer to avoid Template Haskell.
Well, it's not an ADT, but this definitely works:
data Satisfying c = forall a. c a => Satisfy a
class (l a, r a) => And l r a where
instance (l a, r a) => And l r a where
ex :: [Satisfying (Typeable `And` Show `And` Ord)]
ex = [ Satisfy (7 :: Int)
, Satisfy "Hello"
, Satisfy (5 :: Int)
, Satisfy [10..20 :: Int]
, Satisfy ['a'..'z']
, Satisfy ((), 'a')]
-- An example of use, with "complicated" logic
data With f c = forall a. c a => With (f a)
-- vvvvvvvvvvvvvvvvvvvvvvvvvv QuantifiedConstraints chokes on this, which is probably a bug...
partitionTypes :: (forall a. c a => TypeRep a) -> [Satisfying c] -> [[] `With` c]
partitionTypes rep = foldr go []
where go (Satisfy x) [] = [With [x]]
go x'#(Satisfy (x :: a)) (xs'#(With (xs :: [b])) : xss) =
case testEquality rep rep :: Maybe (a :~: b) of
Just Refl -> With (x : xs) : xss
Nothing -> xs' : go x' xss
main :: IO ()
main = traverse_ (\(With xs) -> print (sort xs)) $ partitionTypes typeRep ex
Exhaustivity is much harder. Perhaps with a plugin, you could get GHC to do it, but why bother? I don't believe GHC actually tries to keep track of what types it has seen. In particular, you'd have to scan all modules in the project and its dependencies, even those that haven't been loaded by the module containing the type definition. You'd have to implement it from the ground-up. And, as this answer shows, I very much doubt you would actually be able to use such exhaustivity for anything that you can't already do by just taking the open universe as it is.

How to interact with pure algorithm in IO code

To illustrate the point with a trivial example, say I have implemented filter:
filter :: (a -> Bool) -> [a] -> [a]
And I have a predicate p that interacts with the real world:
p :: a -> IO Bool
How do it make it work with filter without writing a separate implementation:
filterIO :: (a -> IO Bool) -> [a] -> IO [a]
Presumably if I can turn p into p':
p': IO (a -> Bool)
Then I can do
main :: IO ()
main = do
p'' <- p'
print $ filter p'' [1..100]
But I haven't been able to find the conversion.
Edited:
As people have pointed out in the comment, such a conversion doesn't make sense as it would break the encapsulation of the IO Monad.
Now the question is, can I structure my code so that the pure and IO versions don't completely duplicate the core logic?
How do it make it work with filter without writing a separate implementation
That isn't possible and the fact this sort of thing isn't possible is by design - Haskell places firm limits on its types and you have to obey them. You cannot sprinkle IO all over the place willy-nilly.
Now the question is, can I structure my code so that the pure and IO versions don't completely duplicate the core logic?
You'll be interested in filterM. Then, you can get both the functionality of filterIO by using the IO monad and the pure functionality using the Identity monad. Of course, for the pure case, you now have to pay the extra price of wrapping/unwrapping (or coerceing) the Identity wrapper. (Side remark: since Identity is a newtype this is only a code readability cost, not a runtime one.)
ghci> data Color = Red | Green | Blue deriving (Read, Show, Eq)
Here is a monadic example (note that the lines containing only Red, Blue, and Blue are user-entered at the prompt):
ghci> filterM (\x -> do y<-readLn; pure (x==y)) [Red,Green,Blue]
Red
Blue
Blue
[Red,Blue] :: IO [Color]
Here is a pure example:
ghci> filterM (\x -> Identity (x /= Green)) [Red,Green,Blue]
Identity [Red,Blue] :: Identity [Color]
As already said, you can use filterM for this specific task. However, it is usually better to keep with Haskell's characteristic strict seperation of IO and calculations. In your case, you can just tick off all necessary IO in one go and then do the interesting filtering in nice, reliable, easily testable pure code (i.e. here, simply with the normal filter):
type A = Int
type Annotated = (A, Bool)
p' :: Annotated -> Bool
p' = snd
main :: IO ()
main = do
candidates <- forM [1..100] $ \n -> do
permitted <- p n
return (n, permitted)
print $ fst <$> filter p' candidates
Here, we first annotate each number with a flag indicating what the environment says. This flag can then simply be read out in the actual filtering step, without requiring any further IO.
In short, this would be written:
main :: IO ()
main = do
candidates <- forM [1..100] $ \n -> (n,) <$> p n
print $ fst <$> filter snd candidates
While it is not feasible for this specific task, I'd also add that you can in principle achieve the IO seperation with something like your p'. This requires that the type A is “small enough” that you can evaluate the predicate with all values that are possible at all. For instance,
import qualified Data.Map as Map
type A = Char
p' :: IO (A -> Bool)
p' = (Map.!) . Map.fromList <$> mapM (\c -> (c,) <$> p c) ['\0'..]
This evaluates the predicate once for all of the 1114112 chars there are and stores the results in a lookup table.

data type with a default field and that needs a function that works with it

Say, I have a data type
data FooBar a = Foo String Char [a]
| Bar String Int [a]
I need to create values of this type and give empty list as the second field:
Foo "hello" 'a' []
or
Bar "world" 1 []
1) I do this everywhere in my code and I think it would be nice if I could omit the empty list part somehow and have the empty list assigned implicitly. Is this possible? Something similar to default function arguments in other languages.
2) Because of this [] "default" value, I often need to have a partial constructor application that results in a function that takes the first two values:
mkFoo x y = Foo x y []
mkBar x y = Bar x y []
Is there a "better" (more idiomatic, etc) way to do it? to avoid defining new functions?
3) I need a way to add things to the list:
add (Foo u v xs) x = Foo u v (x:xs)
add (Bar u v xs) x = Bar u v (x:xs)
Is this how it is done idiomatically? Just a general purpose function?
As you see I am a beginner, so maybe these questions make little sense. Hope not.
I'll address your questions one by one.
Default arguments do not exist in Haskell. They are simply not worth the added complexity and loss of compositionally. Being a functional language, you do a lot more function manipulation in Haskell, so funkiness like default arguments would be tough to handle.
One thing I didn't realize when I started Haskell is that data constructors are functions just like everything else. In your example,
Foo :: String -> Char -> [a] -> FooBar a
Thus you can write functions for filling in various arguments of other functions, and then those functions will work with Foo or Bar or whatever.
fill1 :: a -> (a -> b) -> b
fill1 a f = f a
--Note that fill1 = flip ($)
fill2 :: b -> (a -> b -> c) -> (a -> c)
--Equivalently, fill2 :: b -> (a -> b -> c) -> a -> c
fill2 b f = \a -> f a b
fill3 :: c -> (a -> b -> c -> d) -> (a -> b -> d)
fill3 c f = \a b -> f a b c
fill3Empty :: (a -> b -> [c] -> d) -> (a -> b -> d)
fill3Empty f = fill3 [] f
--Now, we can write
> fill3Empty Foo x y
Foo x y []
The lens package provides elegant solutions to questions like this. However, you can tell at a glance that this package is enormously complicated. Here is the net result of how you would call the lens package:
_list :: Lens (FooBar a) (FooBar b) [a] [b]
_list = lens getter setter
where getter (Foo _ _ as) = as
getter (Bar _ _ as) = as
setter (Foo s c _) bs = Foo s c bs
setter (Bar s i _) bs = Bar s i bs
Now we can do
> over _list (3:) (Foo "ab" 'c' [2,1])
Foo "ab" 'c' [3,2,1]
Some explanation: the lens function produces a Lens type when given a getter and a setter for some type. Lens s t a b is a type that says "s holds an a and t holds a b. Thus, if you give me a function a -> b, I can give you a function s -> t". That is exactly what over does: you provide it a lens and a function (in our case, (3:) was a function that adds 3 to the front of a List) and it applies the function "where the lens indicates". This is very similar to a functor, however, we have significantly more freedom (in this example, the functor instance would be obligated to change every element of the lists, not operate on the lists themselves).
Note that our new _list lens is very generic: it works equally well over Foo and Bar and the lens package provides many functions other than over for doing magical things.
The idiomatic thing is to take those parameters of a function or constructor that you commonly want to partially apply, and move them toward the beginning:
data FooBar a = Foo [a] String Char
| Bar [a] String Int
foo :: String -> Char -> FooBar a
foo = Foo []
bar :: String -> Int -> FooBar a
bar = Bar []
Similarly, reordering the parameters to add lets you partially apply add to get functions of type FooBar a -> FooBar a, which can be easily composed:
add :: a -> FooBar a -> FooBar a
add x (Foo xs u v) = Foo (x:xs) u v
add123 :: FooBar Int -> FooBar Int
add123 = add 1 . add 2 . add 3
add123 (foo "bar" 42) == Foo [1, 2, 3] "bar" 42
(2) and (3) are perfectly normal and idiomatic ways of doing such things. About (2) in particular, one expression you will occasionally hear is "smart constructor". That just means a function like your mkFoo/mkBar that produces a FooBar a (or a Maybe (FooBar a) etc.) with some extra logic to ensure only reasonable values can be constructed.
Here are some additional tricks that might (or might not!) make sense, depending on what you are trying to do with FooBar.
If you use Foo values and Barvalues in similar ways most of the time (i.e. the difference between having the Char field and the Int one is a minor detail), it makes sense to factor out the similarities and use a single constructor:
data FooBar a = FooBar String FooBarTag [a]
data FooBarTag = Foo Char | Bar Int
Beyond avoiding case analysis when you don't care about the FooBarTag, that allows you to safely use record syntax (records and types with multiple constructors do not mix well).
data FooBar a = FooBar
{ fooBarName :: String
, fooBarTag :: FooBarTag
, fooBarList :: [a]
}
Records allow you to use the fields without having to pattern match the whole thing.
If there are sensible defaults for all fields in a FooBar, you can go one step beyond mkFoo-like constructors and define a default value.
defaultFooBar :: FooBar a
defaultFooBar = FooBar
{ fooBarName = ""
, fooBarTag = Bar 0
, fooBarList = []
}
You don't need records to use a default, but they allow overriding default fields conveniently.
myFooBar = defaultFooBar
{ fooBarTag = Foo 'x'
}
If you ever get tired of typing long names for the defaults over and over, consider the data-default package:
instance Default (FooBar a) where
def = defaultFooBar
myFooBar = def { fooBarTag = Foo 'x' }
Do note that a significant number of people do not like the Default class, and not without reason. Still, for types which are very specific to your application (e.g. configuration settings) Default is perfectly fine IMO.
Finally, updating record fields can be messy. If you end up annoyed by that, you will find lens very useful. Note that it is a big library, and it might be a little overwhelming to a beginner, so take a deep breath beforehand. Here is a small sample:
{-# LANGUAGE TemplateHaskell #-} -- At the top of the file. Needed for makeLenses.
import Control.Lens
-- Note the underscores.
-- If you are going to use lenses, it is sensible not to export the field names.
data FooBar a = FooBar
{ _fooBarName :: String
, _fooBarTag :: FooBarTag
, _fooBarList :: [a]
}
makeLenses ''FooBar -- Defines lenses for the fields automatically.
defaultFooBar :: FooBar a
defaultFooBar = FooBar
{ _fooBarName = ""
, _fooBarTag = Bar 0
, _fooBarList = []
}
-- Using a lens (fooBarTag) to set a field without record syntax.
-- Note the lack of underscores in the name of the lens.
myFooBar = set fooBarTag (Foo 'x') defaultFooBar
-- Using a lens to access a field.
myTag = view fooBarTag myFooBar -- Results in Foo 'x'
-- Using a lens (fooBarList) to modify a field.
add :: a -> FooBar a -> FooBar a
add x fb = over fooBarList (x :) fb
-- set, view and over have operator equivalents, (.~). (^.) and (%~) respectively.
-- Note that (^.) is flipped with respect to view.
Here is a gentle introduction to lens which focuses on aspects I have not demonstrated here, specially in how nicely lenses can be composed.

Why constraints on data are a bad thing?

I know this question has been asked and answered lots of times but I still don't really understand why putting constraints on a data type is a bad thing.
For example, let's take Data.Map k a. All of the useful functions involving a Map need an Ord k constraint. So there is an implicit constraint on the definition of Data.Map. Why is it better to keep it implicit instead of letting the compiler and programmers know that Data.Map needs an orderable key.
Also, specifying a final type in a type declaration is something common, and one can see it as a way of "super" constraining a data type.
For example, I can write
data User = User { name :: String }
and that's acceptable. However is that not a constrained version of
data User' s = User' { name :: s }
After all 99% of the functions I'll write for the User type don't need a String and the few which will would probably only need s to be IsString and Show.
So, why is the lax version of User considered bad:
data (IsString s, Show s, ...) => User'' { name :: s }
while both User and User' are considered good?
I'm asking this, because lots of the time, I feel I'm unnecessarily narrowing my data (or even function) definitions, just to not have to propagate constraints.
Update
As far as I understand, data type constraints only apply to the constructor and don't propagate. So my question is then, why do data type constraints not work as expected (and propagate)? It's an extension anyway, so why not have a new extension doing data properly, if it was considered useful by the community?
TL;DR:
Use GADTs to provide implicit data contexts.
Don't use any kind of data constraint if you could do with Functor instances etc.
Map's too old to change to a GADT anyway.
Scroll to the bottom if you want to see the User implementation with GADTs
Let's use a case study of a Bag where all we care about is how many times something is in it. (Like an unordered sequence. We nearly always need an Eq constraint to do anything useful with it.
I'll use the inefficient list implementation so as not to muddy the waters over the Data.Map issue.
GADTs - the solution to the data constraint "problem"
The easy way to do what you're after is to use a GADT:
Notice below how the Eq constraint not only forces you to use types with an Eq instance when making GADTBags, it provides that instance implicitly wherever the GADTBag constructor appears. That's why count doesn't need an Eq context, whereas countV2 does - it doesn't use the constructor:
{-# LANGUAGE GADTs #-}
data GADTBag a where
GADTBag :: Eq a => [a] -> GADTBag a
unGADTBag (GADTBag xs) = xs
instance Show a => Show (GADTBag a) where
showsPrec i (GADTBag xs) = showParen (i>9) (("GADTBag " ++ show xs) ++)
count :: a -> GADTBag a -> Int -- no Eq here
count a (GADTBag xs) = length.filter (==a) $ xs -- but == here
countV2 a = length.filter (==a).unGADTBag
size :: GADTBag a -> Int
size (GADTBag xs) = length xs
ghci> count 'l' (GADTBag "Hello")
2
ghci> :t countV2
countV2 :: Eq a => a -> GADTBag a -> Int
Now we didn't need the Eq constraint when we found the total size of the bag, but it didn't clutter up our definition anyway. (We could have used size = length . unGADTBag just as well.)
Now lets make a functor:
instance Functor GADTBag where
fmap f (GADTBag xs) = GADTBag (map f xs)
oops!
DataConstraints_so.lhs:49:30:
Could not deduce (Eq b) arising from a use of `GADTBag'
from the context (Eq a)
That's unfixable (with the standard Functor class) because I can't restrict the type of fmap, but need to for the new list.
Data Constraint version
Can we do as you asked? Well, yes, except that you have to keep repeating the Eq constraint wherever you use the constructor:
{-# LANGUAGE DatatypeContexts #-}
data Eq a => EqBag a = EqBag {unEqBag :: [a]}
deriving Show
count' a (EqBag xs) = length.filter (==a) $ xs
size' (EqBag xs) = length xs -- Note: doesn't use (==) at all
Let's go to ghci to find out some less pretty things:
ghci> :so DataConstraints
DataConstraints_so.lhs:1:19: Warning:
-XDatatypeContexts is deprecated: It was widely considered a misfeature,
and has been removed from the Haskell language.
[1 of 1] Compiling Main ( DataConstraints_so.lhs, interpreted )
Ok, modules loaded: Main.
ghci> :t count
count :: a -> GADTBag a -> Int
ghci> :t count'
count' :: Eq a => a -> EqBag a -> Int
ghci> :t size
size :: GADTBag a -> Int
ghci> :t size'
size' :: Eq a => EqBag a -> Int
ghci>
So our EqBag count' function requires an Eq constraint, which I think is perfectly reasonable, but our size' function also requires one, which is less pretty. This is because the type of the EqBag constructor is EqBag :: Eq a => [a] -> EqBag a, and this constraint must be added every time.
We can't make a functor here either:
instance Functor EqBag where
fmap f (EqBag xs) = EqBag (map f xs)
for exactly the same reason as with the GADTBag
Constraintless bags
data ListBag a = ListBag {unListBag :: [a]}
deriving Show
count'' a = length . filter (==a) . unListBag
size'' = length . unListBag
instance Functor ListBag where
fmap f (ListBag xs) = ListBag (map f xs)
Now the types of count'' and show'' are exactly as we expect, and we can use standard constructor classes like Functor:
ghci> :t count''
count'' :: Eq a => a -> ListBag a -> Int
ghci> :t size''
size'' :: ListBag a -> Int
ghci> fmap (Data.Char.ord) (ListBag "hello")
ListBag {unListBag = [104,101,108,108,111]}
ghci>
Comparison and conclusions
The GADTs version automagically propogates the Eq constraint everywhere the constructor is used. The type checker can rely on there being an Eq instance, because you can't use the constructor for a non-Eq type.
The DatatypeContexts version forces the programmer to manually propogate the Eq constraint, which is fine by me if you want it, but is deprecated because it doesn't give you anything more than the GADT one does and was seen by many as pointless and annoying.
The unconstrained version is good because it doesn't prevent you from making Functor, Monad etc instances. The constraints are written exactly when they're needed, no more or less. Data.Map uses the unconstrained version partly because unconstrained is generally seen as most flexible, but also partly because it predates GADTs by some margin, and there needs to be a compelling reason to potentially break existing code.
What about your excellent User example?
I think that's a great example of a one-purpose data type that benefits from a constraint on the type, and I'd advise you to use a GADT to implement it.
(That said, sometimes I have a one-purpose data type and end up making it unconstrainedly polymorphic just because I love to use Functor (and Applicative), and would rather use fmap than mapBag because I feel it's clearer.)
{-# LANGUAGE GADTs #-}
import Data.String
data User s where
User :: (IsString s, Show s) => s -> User s
name :: User s -> s
name (User s) = s
instance Show (User s) where -- cool, no Show context
showsPrec i (User s) = showParen (i>9) (("User " ++ show s) ++)
instance (IsString s, Show s) => IsString (User s) where
fromString = User . fromString
Notice since fromString does construct a value of type User a, we need the context explicitly. After all, we composed with the constructor User :: (IsString s, Show s) => s -> User s. The User constructor removes the need for an explicit context when we pattern match (destruct), becuase it already enforced the constraint when we used it as a constructor.
We didn't need the Show context in the Show instance because we used (User s) on the left hand side in a pattern match.
Constraints
The problem is that constraints are not a property of the data type, but of the algorithm/function that operates on them. Different functions might need different and unique constraints.
A Box example
As an example, let's assume we want to create a container called Box which contains only 2 values.
data Box a = Box a a
We want it to:
be showable
allow the sorting of the two elements via sort
Does it make sense to apply the constraint of both Ord and Show on the data type? No, because the data type in itself could be only shown or only sorted and therefore the constraints are related to its use, not it's definition.
instance (Show a) => Show (Box a) where
show (Box a b) = concat ["'", show a, ", ", show b, "'"]
instance (Ord a) => Ord (Box a) where
compare (Box a b) (Box c d) =
let ca = compare a c
cb = compare b d
in if ca /= EQ then ca else cb
The Data.Map case
Data.Map's Ord constraints on the type is really needed only when we have > 1 elements in the container. Otherwise the container is usable even without an Ord key. For example, this algorithm:
transf :: Map NonOrd Int -> Map NonOrd Int
transf x =
if Map.null x
then Map.singleton NonOrdA 1
else x
Live demo
works just fine without the Ord constraint and always produce a non empty map.
Using DataTypeContexts reduces the number of programs you can write. If most of those illegal programs are nonsense, you might say it's worth the runtime cost associated with ghc passing in a type class dictionary that isn't used. For example, if we had
data Ord k => MapDTC k a
then #jefffrey's transf is rejected. But we should probably have transf _ = return (NonOrdA, 1) instead.
In some sense the context is documentation that says "every Map must have ordered keys". If you look at all of the functions in Data.Map you'll get a similar conclusion "every useful Map has ordered keys". While you can create maps with unordered keys using
mapKeysMonotonic :: (k1 -> k2) -> Map k1 a -> Map k2 a
singleton :: k2 a -> Map k2 a
But the moment you try to do anything useful with them, you'll wind up with No instance for Ord k2 somewhat later.

Haskell data serialization of some data implementing a common type class

Let's start with the following
data A = A String deriving Show
data B = B String deriving Show
class X a where
spooge :: a -> Q
[ Some implementations of X for A and B ]
Now let's say we have custom implementations of show and read, named show' and read' respectively which utilize Show as a serialization mechanism. I want show' and read' to have types
show' :: X a => a -> String
read' :: X a => String -> a
So I can do things like
f :: String -> [Q]
f d = map (\x -> spooge $ read' x) d
Where data could have been
[show' (A "foo"), show' (B "bar")]
In summary, I wanna serialize stuff of various types which share a common typeclass so I can call their separate implementations on the deserialized stuff automatically.
Now, I realize you could write some template haskell which would generate a wrapper type, like
data XWrap = AWrap A | BWrap B deriving (Show)
and serialize the wrapped type which would guarantee that the type info would be stored with it, and that we'd be able to get ourselves back at least an XWrap... but is there a better way using haskell ninja-ery?
EDIT
Okay I need to be more application specific. This is an API. Users will define their As, and Bs and fs as they see fit. I don't ever want them hacking through the rest of the code updating their XWraps, or switches or anything. The most i'm willing to compromise is one list somewhere of all the A, B, etc. in some format. Why?
Here's the application. A is "Download a file from an FTP server." B is "convert from flac to mp3". A contains username, password, port, etc. information. B contains file path information. There could be MANY As and Bs. Hundreds. As many as people are willing to compile into the program. Two was just an example. A and B are Xs, and Xs shall be called "Tickets." Q is IO (). Spooge is runTicket. I want to read the tickets off into their relevant data types and then write generic code that will runTicket on the stuff read' from the stuff on disk. At some point I have to jam type information into the serialized data.
I'd first like to stress for all our happy listeners out there that XWrap is a very good way, and a lot of the time you can write one yourself faster than writing it using Template Haskell.
You say you can get back "at least an XWrap", as if that meant you couldn't recover the types A and B from XWrap or you couldn't use your typeclass on them. Not true! You can even define
separateAB :: [XWrap] -> ([A],[B])
If you didn't want them mixed together, you should serialise them seperately!
This is nicer than haskell ninja-ery; maybe you don't need to handle arbitrary instances, maybe just the ones you made.
Do you really need your original types back? If you feel like using existential types because you just want to spooge your deserialised data, why not either serialise the Q itself, or have some intermediate data type PoisedToSpooge that you serialise, which can deserialise to give you all the data you need for a really good spooging. Why not make it an instance of X too?
You could add a method to your X class that converts to PoisedToSpooge.
You could call it something fun like toPoisedToSpooge, which trips nicely off the tongue, don't you think? :)
Anyway this would remove your typesystem complexity at the same time as resolving the annoying ambiguous type in
f d = map (\x -> spooge $ read' x) d -- oops, the type of read' x depends on the String
You can replace read' with
stringToPoisedToSpoogeToDeserialise :: String -> PoisedToSpooge -- use to deserialise
and define
f d = map (\x -> spooge $ stringToPoisedToSpoogeToDeserialise x) -- no ambiguous type
which we could of course write more succincly as
f = map (spooge.stringToPoisedToSpoogeToDeserialise)
although I recognise the irony here in suggesting making your code more succinct. :)
If what you really want is a heterogeneous list then use existential types. If you want serialization then use Cereal + ByteString. If you want dynamic typing, which is what I think your actual goal is, then use Data.Dynamic. If none of this is what you want, or you want me to expand please press the pound key.
Based on your edit, I don't see any reason a list of thunks won't work. In what way does IO () fail to represent both the operations of "Download a file from an FTP server" and "convert from flac to MP3"?
I'll assume you want to do more things with deserialised Tickets
than run them, because if not you may as well ask the user to supply a bunch of String -> IO()
or similar, nothing clever needed at all.
If so, hooray! It's not often I feel it's appropriate to recommend advanced language features like this.
class Ticketable a where
show' :: a -> String
read' :: String -> Maybe a
runTicket :: a -> IO ()
-- other useful things to do with tickets
This all hinges on the type of read'. read' :: Ticket a => String -> a isn't very useful,
because the only thing it can do with invalid data is crash.
If we change the type to read' :: Ticket a => String -> Maybe a this can allow us to read from disk and
try all the possibilities or fail altogether.
(Alternatively you could use a parser: parse :: Ticket a => String -> Maybe (a,String).)
Let's use a GADT to give us ExistentialQuantification without the syntax and with nicer error messages:
{-# LANGUAGE GADTs #-}
data Ticket where
MkTicket :: Ticketable a => a -> Ticket
showT :: Ticket -> String
showT (MkTicket a) = show' a
runT :: Ticket -> IO()
runT (MkTicket a) = runTicket a
Notice how the MkTicket contstuctor supplies the context Ticketable a for free! GADTs are great.
It would be nice to make Ticket and instance of Ticketable, but that won't work, because there would be
an ambiguous type a hidden in it. Let's take functions that read Ticketable types and make them read
Tickets.
ticketize :: Ticketable a => (String -> Maybe a) -> (String -> Maybe Ticket)
ticketize = ((.).fmap) MkTicket -- a little pointfree fun
You could use some unusual sentinel string such as
"\n-+-+-+-+-+-Ticket-+-+-+-Border-+-+-+-+-+-+-+-\n" to separate your serialised data or better, use separate files
altogether. For this example, I'll just use "\n" as the separator.
readTickets :: [String -> Maybe Ticket] -> String -> [Maybe Ticket]
readTickets readers xs = map (foldr orelse (const Nothing) readers) (lines xs)
orelse :: (a -> Maybe b) -> (a -> Maybe b) -> (a -> Maybe b)
(f `orelse` g) x = case f x of
Nothing -> g x
just_y -> just_y
Now let's get rid of the Justs and ignore the Nothings:
runAll :: [String -> Maybe Ticket] -> String -> IO ()
runAll ps xs = mapM_ runT . catMaybes $ readTickets ps xs
Let's make a trivial ticket that just prints the contents of some directory
newtype Dir = Dir {unDir :: FilePath} deriving Show
readDir xs = let (front,back) = splitAt 4 xs in
if front == "dir:" then Just $ Dir back else Nothing
instance Ticketable Dir where
show' (Dir p) = "dir:"++show p
read' = readDir
runTicket (Dir p) = doesDirectoryExist p >>= flip when
(getDirectoryContents >=> mapM_ putStrLn $ p)
and an even more trivial ticket
data HelloWorld = HelloWorld deriving Show
readHW "HelloWorld" = Just HelloWorld
readHW _ = Nothing
instance Ticketable HelloWorld where
show' HelloWorld = "HelloWorld"
read' = readHW
runTicket HelloWorld = putStrLn "Hello World!"
and then put it all together:
myreaders = [ticketize readDir,ticketize readHW]
main = runAll myreaders $ unlines ["HelloWorld",".","HelloWorld","..",",HelloWorld"]
Just use Either. Your users don't even have to wrap it themselves. You have your deserializer wrap it in the Either for you. I don't know exactly what your serialization protocol is, but I assume that you have some way to detect which kind of request, and the following example assumes the first byte distinguishes the two requests:
deserializeRequest :: IO (Either A B)
deserializeRequest = do
byte <- get1stByte
case byte of
0 -> do
...
return $ Left $ A <A's fields>
1 -> do
...
return $ Right $ B <B's fields>
Then you don't even need to type-class spooge. Just make it a function of Either A B:
spooge :: Either A B -> Q

Resources