Related
I’m trying to understand the applicative typeclass, and in particular the <*> function. Now I see its type signature is f (a -> b) -> f a -> f b and I take it that f is a functor, which I think of as some kind of structure wrapping some data. It appears to me that f must handle generic types, specifically it must be possible for f to have the parameterized types a, b, and in fact it must also support a -> b.
If my understanding is correct, then what we are doing is working with a typeclass f that is initially intended to wrap some data, like say a list of strings or a tree containing file buffers or whatever random thing we might want. But f must not be committed to any one type, because not only must it handle such a type but it is also required to handle functions from the data to the other data. Thus if we had an example implementation of <*> which contained
Just2 f <*> j = fmap f j
the way to read this is that j is some kind of data inside of a Just2. f is a function which maps data to data. And f is wrapped in a Just2.
Is all of that right? Fundamentally my question is: Must anything applicative be so generic that it can always simultaneously handle arbitrary data and also functions from data to data? Or is there some way that you could have an applicative such that the only data it allows inside is, say, lists?
Yes, your understanding is largely correct. In particular, any specific Applicative, say one named Foo, has an associated specialization of the function pure with type signature:
pure :: a -> Foo a
that must work for any type a selected by the caller, such as:
> pure 10 :: Foo Int
> pure length :: Foo (String -> Int)
So, whatever Foo is, it has to be able to "handle" any provided type without limitations because pure can technically be applied to any type without limitations.
One cautionary note, though. The idea that a functor f "wraps" data, so that f Int is somehow a "container" of Int values, can be a helpful intuition and is often literally correct (e.g., lists, trees, etc.), but it's not always strictly true. (Some counterexamples include the functors IO, (->) r, and Const b, which "contain" values in a very different sense than real containers.)
For "regular" Functors and Applicatives, you're right; they need to be able to handle values of any type. This is known as parametric polymorphism. If you have a type that you think is almost a Functor except that it can't do that, then consider the MonoFunctor typeclass from the mono-traversable package. It's the same idea as Functor, except with a single valid element type baked in. I'm not aware of any packages that have a monomorphic equivalent to Applicative. I think this is because <*> uses values of 3 different types inside the same container, so it doesn't have a good monomorphic analogue.
From your question, it sounds like there could be many different Applicative typeclasses. But there is only one; therefore it has to be generic.
The Applicative typeclass is defined by its functions (<*> and pure) and by the laws these functions need to adhere to. In any Haskell codebase, there can be exactly one definition only. But there can be many types that have an instance of the Applicative typeclass. You can define your own instances with the instance declaration and by defining the required functions. The compiler will not check that your definitions abide by the functor laws, though - this is up to you to ensure.
Typeclasses like Functor, Applicative and Monad do not specify a data type; they don't say that you need a "list-like" type, or a "box-like" type, event though lists and other containers like Either do have instances of these type classes. Any type that you can equip with the required functions in such a way that the Applicative laws hold becomes an Applicative.
It is often helpful to think of containers or boxes. But you need to stretch that intuition once you use Applicative instances of types like functions; e.g., a -> r.
Compared with interfaces in OO-languages, typeclasses are more powerful because you can define some data type to be an instance of a typeclass even if you do not have access to the source code of the type itself.
It seems like there are a lot of functions that do the same thing, particularly relating to Monads, Functors, and Applicatives.
Examples (from most to least generic):
fmap == liftA == liftM
(<*>) == ap
liftA[2345] == liftM[2345]
pure == return
(*>) == (>>)
An example not directly based on the FAM class tree:
fmap == map
(I thought there were quite a few more with List, Foldable, Traversable, but it looks like most were made more generic some time ago, as I only see the old, less generic type signatures in old stack overflow / message board questions)
I personally find this annoying, as it means that if I need to do x, and some function such as liftM allows me to do x, then I will have made my function less generic than it could have been, and I am only going to notice that kind of thing by thoroughly reasoning about the differences between types (such as FAM, or perhaps List, Foldable, Traversable combinations as well), which is not beginner friendly at all, as while simply using those types isn't all that hard, reasoning about their properties and laws requires a lot more mental effort.
I am guessing a lot of these equivalencies come from the Applicative Monad Proposal. If that is the reason for them (and not some other reason I am missing for having less generic functions available for confusion), are they going to be deprecated / deleted ever? I can understand waiting a long time to delete them, due to breaking existing code, but surely deprecation is a good idea?
The short answers are "history" and "regularity".
Originally "map" was defined for lists. Then type-classes were introduced, with the Functor type class, so the generalised version of "map" for any functor had to be called something different, otherwise existing code would be broken. Hence "fmap".
Then monads came along. Instances of monads did not need to be functors, so "liftM" was created, along with "liftM2", "liftM3" etc. Of course if a type is an instance of both Monad and Functor then fmap = liftM.
Monads also have "ap", used in expressions like f `ap` arg1 `ap` arg2. This was very handy, but then Applicative Functors were added. (<*>) did the same job for applicative functors as 'ap', but because many applicative functors are not monads it had to be called something different. Likewise liftAx versus liftMx and "pure" versus "return".
They aren't equivalent though. equivalent things in haskell can be interchanged with no difference at all in functionality. Consider for example pure and return
EDIT: I wrote some examples down, but they were really bad since they involved Maybe a, a type that is both an applicative and a monad, so the functions could be used pretty interchangeably.
There are types that are applicatives but not monads though (see this question for examples), and by studying the type of the following expression, we can see that this could lead to some roadbumps:
pure 1 >>= pure :: (Monad m, Num b) => m b
I personally find this annoying, as it means that if I need to do x, and some function such as liftM allows me to do x, then I will have made my function less generic than it could have been
This logic is backwards.
Normally you know in advance the type of the thing you want to write, be it IO String or (Foldable f, Monoid t, Monad m) => f (m t) -> m t or whatever. Let's take the first case, getLineCapitalized :: IO String. You could write it as
getLineCapitalized = liftM (map toUpper) getLine
or
getLineCapitalized = fmap (fmap toUpper) getLine
Is the former "less generic" because it uses the specialized functions liftM and map? Of course not. This is intrinsically an IO action that produces a list. It cannot become "more generic" by changing it to the second version since those fmaps will have their types fixed to IO and [] anyways. So, there is no advantage to the second version.
By writing the first version, you provide contextual information to the reader for free. In liftM (map foo) bar, the reader knows that bar is going to be an action in some monad that returns a list. In fmap (fmap foo) bar, it could be any sort of doubly-nested structure whatsoever. If bar is something complicated rather than just getLine, then this kind of information is helpful for understanding more easily what is going on in bar.
In general, you should write a function in two steps.
Decide what the type of the function should be. Make it as general or as specific as you want. The more general the type of the function, the stronger guarantees you get on its behavior from parametricity.
Once you have decided on the type of your function, implement it using the most specific available functions. By doing so, you are providing the most information to the reader of your function. You never lose any generality or parametricity guarantees by doing so, since those only depend on the type, which you already determined in step 1.
Edit in response to comments: I was reminded of the biggest reason to use the most specific function available, which is catching bugs. The type length :: [a] -> Int is essentially the entire reason that I still use GHC 7.8. It's never happened that I wanted to take the length of an unknown Foldable structure. On the other hand, I definitely do not want to ever accidentally take the length of a pair, or take the length of foo bar baz which I think has type [a], but actually has type Maybe [a].
In the use cases for Foldable that are not already covered by the rest of the Haskell standard, lens is a vastly more powerful alternative. If I want the "length" of a Maybe t, lengthOf _Just :: Maybe t -> Int expresses my intent clearly, and the compiler can check that the program actually matches my intent; and I can go on to write lengthOf _Nothing, lengthOf _Left, etc. Explicit is better than implicit.
There are some "redundant" functions like liftM, ap, and liftA that have a very real use and taking them out would cause loss of functionality --- you can use liftM, ap, and liftA to implement your Functor or Applicative instances if all you've written is a Monad instance. It lets you be lazy and do, say:
instance Monad Foo where
return = ...
(>>=) = ...
Now you've done all of the rewarding work of defining a Monad instance, but this won't compile. Why? Because you also need a Functor and Applicative instance.
So, because you're quickly prototyping, or lazy, or can't think of a better way, you can just get a free Functor and Applicative instance:
instance Functor Foo where
fmap = liftM
instance Applicative Foo where
pure = return
(<*>) = ap
In fact, you can just copy-and-paste that chunk of code everywhere you need to quickly define a Functor or Applicative instance when you already have a Monad instance defined.
The same goes for fmapDefault from Data.Traversable. If you've implemented Traversable, you can also implement Foldable and Functor:
instance Functor Bar where
fmap = fmapDefault
no extra work required!
There are some redundant functions, however, that really have no actual usage other than being historical accidents from a time when Functor was not a superclass of Monad. These have literally zero use/point in existing...and include things like the liftM2, liftM3 etc., and (>>) and friends.
I am working through the section on Functors in the Typeclassopedia.
A simple intuition is that a Functor represents a “container” of some sort, along with the ability to apply a function uniformly to every element in the container.
OK. So, functors appear pretty natural for inductive types like lists or trees.
Functors also appear pretty simple if the number of elements is fixed to a low number. For example, with Maybe you just have to be concerned about "Nothing" or "Just a" -- two things.
So, how would you make something like a graph, that could potentially have loops, an instance of Functor? I think a more generalized way to put it is, how do non-inductive types "fit into" Functors?
The more I think about it, the more I realize that inductive / non-inductive doesn't really matter. Inductive types are just easier to define fmap for...
If I wanted to make a graph an instance of Functor, I would have to implement a graph traversal algorithm inside fmap; for example it would probably have to use a helper function that would keep track of the visited nodes. At this point, I am now wondering why bother defining it as a Functor instead of just writing this as a function itself? E.g. map vs fmap for lists...?
I hope someone with experience, war stories, and scars can shed some light. Thanks!
Well let's assume you define a graph like this
data Graph a = Node a [Graph a]
Then fmap is just defined precisely as you would expect
instance Functor Graph where
fmap f (Node a ns) = Node (f a) (map (fmap f) ns)
Now, if there's a loop then we'd have had to do something like
foo = Node 1 [bar]
bar = Node 2 [foo]
Now fmap is sufficiently lazy that you can evaluate part of it's result without forcing the rest of the computation, so it works just as well as any knot-tied graph representation would!
In general this is the trick: fmap is lazy so you can treat it's results just as you would treat any non-inductive values in Haskell (: carefully).
Also, you should define fmap vs the random other functions since
fmap is a good, well known API with rules
Your container now places well with things expecting Functors
You can abstract away other bits of your program so they depend on Functor, not your Graph
In general when I see something is a functor I think "Ah wonderful, I know just how to use that" and when I see
superAwesomeTraversal :: (a -> b) -> Foo a -> Foo b
I get a little worried that this will do unexpected things..
Comonoids are mentioned, for example, in Haskell's distributive library docs:
Due to the lack of non-trivial comonoids in Haskell, we can restrict ourselves to requiring a Functor rather than some Coapplicative class.
After a little searching I found a StackOverflow answer that explains this a bit more with the laws that comonoids would have to satisfy. So I think I understand why there's only one possible instance for a hypothetical Comonoid typeclass in Haskell.
Thus, to find a nontrivial comonoid, I suppose we'd have to look in some other category. Surely, if category theorists have a name for comonoids, then there are some interesting ones. The other answers on that page seem to hint at an example involving Supply, but I couldn't figure one out that still satisfies the laws.
I also turned to Wikipedia: there's a page for monoids that doesn't reference category theory, which seems to me as an adequate description of Haskell's Monoid typeclass, but "comonoid" redirects to a category-theoretic description of monoids and comonoids together that I can't understand, and there still don't seem to be any interesting examples.
So my questions are:
Can comonoids be explained in non-category-theoretic terms like monoids?
What is a simple example of an interesting comonoid, even if it's not a Haskell type? (Could one be found in a Kleisli category over a familiar Haskell monad?)
edit: I am not sure if this is actually category-theoretically correct, but what I was imagining in the parenthetical of question 2 was nontrivial definitions of delete :: a -> m () and split :: a -> m (a, a) for some specific Haskell type a and Haskell monad m that satisfy Kleisli-arrow versions of the comonoid laws in the linked answer. Other examples of comonoids are still welcome.
As Phillip JF mentioned, comonoids are interesting to talk about in substructural logics. Let's talk about linear lambda calculus. This is much like your normal typed lambda calculus except that every variable must be used exactly once.
To get a feel, let's count linear functions of given types, i.e.
a -> a
has exactly one inhabitant, id. While
(a,a) -> (a,a)
has two, id and flip. Note that in regular lambda calculus (a,a) -> (a,a) has four inhabitants
(a, b) ↦ (a, a)
(a, b) ↦ (b, b)
(a, b) ↦ (a, b)
(a, b) ↦ (b, a)
but the first two require that we use one of the arguments twice while discarding the other. This is exactly the essence of linear lambda calculus—disallowing those kinds of functions.
As a quick aside, what's the point of linear LC? Well, we can use it to model linear effects or resource usage. If, for instance, we have a file type and a few transformers it might look like
data File
open :: String -> File
close :: File -> () -- consumes a file, but we're ignoring purity right now
t1 :: File -> File
t2 :: File -> File
and then the following are valid pipelines:
close . t1 . t2 . open
close . t2 . t1 . open
close . t1 . open
close . t2 . open
but this "branching" computation isn't
let f1 = open "foo"
f2 = t1 f1
f3 = t2 f1
in close f3
since we used f1 twice.
Now, you might be wondering something at this point about what things must follow the linear rules. For instance, I decided that some pipelines don't have to include both t1 and t2 (compare the enumeration exercise from before). Further, I introduced the open and close functions which happily create and destroy the File type despite that being a violation of linearity.
Indeed, we might posit the existence of functions which violate linearity—but not all clients may. It's much like the IO monad—all of the secrets live inside the implementation of IO so that users work in a "pure" world.
And this is where Comonoid comes in.
class Comonoid m where
destroy :: m -> ()
split :: m -> (m, m)
A type that instantiates Comonoid in a linear lambda calculus is a type which has carry-along destruction and duplication rules. In other words, it's a type which isn't very much bound by linear lambda calculus at all.
Since Haskell doesn't implement the linear lambda calculus rules at all, we can always instantiate Comonoid
instance Comonoid a where
destroy a = ()
split a = (a, a)
Or, perhaps the other way to think of it is that Haskell is a linear LC system that just happens to instantiate Comonoid for every type and applies destroy and split for you automatically.
A monoid in the usual sense is the same as a categorical monoid in the category of sets. One would expect that a comonoid in the usual sense is the same as a categorical comonoid in the category of sets. But every set in the category of sets is a comonoid in a trivial way, so apparently there is no non-categorical description of comonoids which would be parallel to that of monoids.
Just like a monad is a monoid in the category of endofunctors (what's the problem?), a comonad is a comonoid in the category of endofunctors (what's the coproblem?) So yes, any comonad in Haskell would be an example of a comonoid.
Well one way we can think of a monoid is as hooked to any particular product construction that we're using, so in Set we'd take this signature:
mul : A * A -> A
one : A
to this one:
dup : A -> A * A
one : A
but the idea of duality is that the logical statements that you can make all have duals which can be applied to the dual objects, and there is another way of stating what a monoid is, and that's being agnostic to the choice of product construction and then when we take the costructure we can take the coproduct in the output, like:
div : A -> A + A
one : A
where + is a tagged sum. Here we essentially have that every single term which is in this type is always ready to produce a new bit, which is implicitly derived from the tag used to denote the left or the right instance of A. I personally think this is really damn cool. I think the cool version of the things that people were talking about above is when you don't particularly construct that for monoids, but for monoid actions.
A monoid M is said to act on a set A if there's a function
act : M * A -> A
where we have the following rules
act identity a = a
act f (act g a) = act (f * g) a
If we want a co-action, what exactly do we want?
act : A -> M * A
this generates us a stream of the type of our comonoid! I'm having a lot of trouble coming up with the laws for these systems, but I think they must be around somewhere so I'm gonna keep looking tonight. If somebody can tell me them or that I'm wrong about these things in some way or another, also interested in that.
As a physicist, the most common example I deal with is coalgebras, which are comonoid objects in the category of vector spaces, with the monoidal structure usually given by the tensor product.
In that case, there is a bijection between monoid and comonoid objects, since you can just take the adjoint or transpose of the product and unit maps to get a coproduct and a counit that satisfy the comonoid axioms.
In some branches of physics, it is very common to see objects that have both an algebra and a coalgebra structure with some compatibility axioms. The two most common cases are Hopf algebras and Frobenius algebras. They are very convenient for constructing states or solution that are entangled or correlated.
In programming, the simplest nontrivial example I can think of would be reference counted pointers such as shared_ptr in C++ and Rc in Rust, along with their weak equivalents. You can copy them, which is a nontrivial operation that bumps up the refcount (so the two copies are distinct from the initial state). You can drop (call the destructor) on one, which is nontrivial because it bumps down the refcount of any other refcounted pointer that points to the same piece of data.
Furthermore, weak pointers are a great example of a comonoid action. You can use the co-action to generate a weak pointer from a shared pointer. This can be easily checked by noting that creating one from a shared pointer and immediately dropping it is a unit operation, and creating one & cloning it is equivalent to creating two from the shared pointer.
This is a general thing you see with nontrivial coproducts and their co-actions: when they don't reduce to a copying operation, they intuitively imply some form of action at a distance between the two halves, while also adding an operation that erases one half to leave the other independent.
To clarify the question: it is about the merits of the monad type class (as opposed to just its instances without the unifying class).
After having read many references (see below),
I came to the conclusion that, actually, the monad class is there to solve only one, but big and crucial, problem: the 'chaining' of functions on types with context. Hence, the famous sentence "monads are programmable semicolons".
In fact, a monad can be viewed as an array of functions with helper operations.
I insist on the difference between the monad class, understood as a general interface for other types; and these other types instantiating the class (thus, "monadic types").
I understand that the monad class by itself, only solves the chaining of operators because mainly, it only mandates its type instances
to have bind >>= and return, and tell us how they must behave. And as a bonus, the compiler greatyly helps the coding providing do notation for monadic types.
On the other hand,
it is each individual type instantiating the monad class which solves each concrete problem, but not merely for being a instance of Monad. For instance Maybe solves "how a function returns a value or an error", State solves "how to have functions with global state", IO solves "how to interact with the outside world", and so on. All theses classes encapsulate a value within a context.
But soon or later, we will need to chain operations on such context-types. I.e., we will need to organize calls to functions on these types in a particular sequence (for an example of such a problem, please read the example about multivalued functions in You could have invented monads).
And you get solved the problem of chaining, if you have each type be an instance of the monad class.
For the chaining to work you need >>= just with the exact signature it has, no other. (See this question).
Therefore, I guess that the next time you define a context data type T for solving something, if you need to sequence calls of functions (on values of T) consider making T an instance of Monad (if you need "chaining with choice" and if you can benefit from the do notation). And to make sure you are doing it right, check that T satisfies the monad laws
Then, I ask two questions to the Haskell experts:
A concrete question: is there any other problem that the monad class solves by ifself (leaving apart monadic classes)? If any, then, how it compares in relevance to the problem of chaining operations?
An optional general question: are my conclusions right, am I misunderstanding something?
References
Tutorials
Monads in pictures Definitely worth it; read this one first.
Fistful of monads
You could have invented monads
Monads are trees (pdf)
StackOverflow Questions & Answers
How to detect a monad
On the signature of >>= monad operator
You're definitely on to something in the way that you're stating this—there are many things that Monad means and you've separated them out well.
That said, I would definitely say that chaining operations is not the primary thing solved by Monads. That can be solved using plain Functors (with some trouble) or easily with Applicatives. You need to use the full monad spec when "chaining with choice". In particular, the tension between Applicative and Monad comes from Applicative needing to know the entire structure of the side-effecting computation statically. Monad can change that structure at runtime and thus sacrifices some analyzability for power.
To make the point more clear, the only place you deal with a Monad but not any specific monad is if you're defining something with polymorphism constrained to be a Monad. This shows up repeatedly in the Control.Monad module, so we can examine some examples from there.
sequence :: [m a] -> m [a]
forever :: m a -> m b
foldM :: (a -> b -> m a) -> a -> [b] -> m a
Immediately, we can throw out sequence as being particular to Monad since there's a corresponding function in Data.Traversable, sequenceA which has a type slightly more general than Applicative f => [f a] -> f [a]. This ought to be a clear indicator that Monad isn't the only way to sequence things.
Similarly, we can define foreverA as follows
foreverA :: Applicative f => f a -> f b
foreverA f = flip const <$> f <*> foreverA f
So more ways to sequence non-Monad types. But we run into trouble with foldM
foldM :: (Monad m) => (a -> b -> m a) -> a -> [b] -> m a
foldM _ a [] = return a
foldM f a (x:xs) = f a x >>= \fax -> foldM f fax xs
If we try to translate this definition to Applicative style we might write
foldA :: (Applicative f) => (a -> b -> f a) -> a -> [b] -> f a
foldA _ a [] = pure a
foldA f a (x:xs) = foldA f <$> f a x <*> xs
But Haskell will rightfully complain that this doesn't typecheck--each recursive call to foldA tries to put another "layer" of f on the result. With Monad we could join those layers down, but Applicative is too weak.
So how does this translate to Applicatives restricting us from runtime choices? Well, that's exactly what we express with foldM, a monadic computation (a -> b -> m a) which depends upon its a argument, a result from a prior monadic computation. That kind of thing simply doesn't have any meaning in the more purely sequential world of Applicative.
To solve the problem of chaining operations on an individual monadic type, it's not at all necessary to make it an instance of Monad and be sure the monad laws are satisfied. You could just implement a chaining operation directly on your type.
It would probably be very similar to the monadic bind, but not necessarily exactly the same (recall that bind for lists is concatMap, a function that exists anyway, but with the arguments in a different order). And you wouldn't have to worry about the monad laws, because you would have a slightly different interface for each type, so they wouldn't have any common requirements.
To ask what problem the Monad type class itself solves, look at all the functions (in Control.Monad and else where) that work on values in any monadic type. The problem solved is code reuse! Monad is exactly the part of all the monadic types that is common to each and every one of them. That part is sufficient on its own to write useful computations. All of these functions could be implemented for any individual monadic type (often more directly), but they've already been implemented for all monadic types, even the ones that don't exist yet.
You don't write a Monad instance so that you can chain operations on your type (often you already have a way of chaining, in fact). You write a Monad instance for all the code that automatically comes along with the Monad instance. Monad isn't about solving any problem for any single type, it's about a way of viewing many disparate types as instances of a single unifying concept.