Can you explain why one works but not the other?
Given
//fu : unit -> unit
let fu() = ();;
This works
//exec : (unit -> unit) -> int -> unit
let exec (f:(unit -> unit)) (data:int) = f();;
//this works, and p : int -> unit
let p = exec fu;;
And it works for other types of data such as string, long, etc.
This doesn't work
//exec : (unit -> unit) -> obj -> unit
let exec (f:(unit -> unit)) (data:obj) = f();;
let p = exec fu;;
and I get the following error:
error FS0030: Value restriction. The value 'p' has been inferred to have generic type
val p : ('_a -> unit)
Either make the arguments to 'p' explicit or, if you do not intend for it to be generic, add a type annotation.
Notice the only difference between these cases is the type of the data parameter.
When it is obj or System.Object or 'a - it doesn't work.
Another thing is that if data has type obj then the following happens:
//Data type is obj
let exec (f:(unit -> unit)) (data:obj) = f();;
//specifying parameters explicitly
let p x = exec fu x;;
Now p has the signature of 'a -> unit, not obj -> unit.
So the question is: why the "shortcuted" currying doesn't work when data is obj or 'a and why the type of p is 'a -> unit when data was obj?
So I think the problem is that F# appears to be generalizing at the wrong point (from your point of view):
Here is a version of your code that at first glance shouldn't typecheck:
let exec (f:unit -> unit) (data:obj) = f();;
let p:int -> unit = exec (fun () -> ());;
This seems weird as int <> obj.
Also, here is an even simpler example which shows your behaviour (from the spec with modifications)
type Base() =
member b.X = 1
type Derived(i : int) =
inherit Base()
member d.Y = i
let exec (f:unit -> unit) (data:Base) = f();;
let p = exec (fun () -> ());;
which produces a value restriction error.
This makes it more clear that as F# inserts an implicit upcast before the function calls, the code is valid where it is a function, but once you make it an explicit value, this cast can't be used.
If you want your code to compile, you need to move the location of the upcast by adding a type annotation:
let exec (f:unit -> unit) (data:obj) = f()
let p:obj -> unit = exec (fun () -> ());;
Related
Have I correctly implemented map for Task?
let map continuation (t: Task<'A>) =
t.ContinueWith(fun (antecedent: Task<'A>) ->
if antecedent.Status <> TaskStatus.Canceled &&
antecedent.Status <> TaskStatus.Faulted then
continuation antecedent.Result
else
raise antecedent.Exception // must I?
)
I got the TaskStatus checks from the docs. I feel most uncertain about raise antecedent.Exception, but I can't think of another way to handle it.
As background, yes I'm aware of Async, but my current stack uses Entity Framework and Blazor, so I have a backend that uses things like .ToListAsync() and a front end in C#, so I'd rather just not deal with converting from Task to Async then back again.
I would suggest implementing your solution in terms of the interfaces behind the concept of awaitable in the TPL, namely INotifyCompletion and ICriticalNotifyCompletion. Also, to implement map correctly, you should really do it in terms of bind. This is something that there are already some existing solutions for in F#, such as the TaskBuilder library. Personally, I have been using the following in a library for years without any issues:
open System.Runtime.CompilerServices
open System.Threading.Tasks
type TaskStep<'result> =
| Value of 'result
| AsyncValue of 'result Task
| Continuation of ICriticalNotifyCompletion * (unit -> 'result TaskStep)
and StateMachine<'a>(firstStep) as this =
let methodBuilder = AsyncTaskMethodBuilder<'a Task>()
let mutable continuation = fun () -> firstStep
let nextAwaitable() =
try
match continuation() with
| Value r ->
methodBuilder.SetResult(Task.FromResult(r))
null
| AsyncValue t ->
methodBuilder.SetResult(t)
null
| Continuation (await, next) ->
continuation <- next
await
with
| exn ->
methodBuilder.SetException(exn)
null
let mutable self = this
member __.Run() =
methodBuilder.Start(&self)
methodBuilder.Task
interface IAsyncStateMachine with
member __.MoveNext() =
let mutable await = nextAwaitable()
if not (isNull await) then
methodBuilder.AwaitUnsafeOnCompleted(&await, &self)
member __.SetStateMachine(_) =
()
type Binder<'out> =
static member inline GenericAwait< ^abl, ^awt, ^inp
when ^abl : (member GetAwaiter : unit -> ^awt)
and ^awt :> ICriticalNotifyCompletion
and ^awt : (member get_IsCompleted : unit -> bool)
and ^awt : (member GetResult : unit -> ^inp) >
(abl : ^abl, continuation : ^inp -> 'out TaskStep) : 'out TaskStep =
let awt = (^abl : (member GetAwaiter : unit -> ^awt)(abl))
if (^awt : (member get_IsCompleted : unit -> bool)(awt))
then continuation (^awt : (member GetResult : unit -> ^inp)(awt))
else Continuation (awt, fun () -> continuation (^awt : (member GetResult : unit -> ^inp)(awt)))
module TaskStep =
let inline bind f step : TaskStep<'a> =
Binder<'a>.GenericAwait(step, f)
let inline toTask (step: TaskStep<'a>) =
try
match step with
| Value x -> Task.FromResult(x)
| AsyncValue t -> t
| Continuation _ as step -> StateMachine<'a>(step).Run().Unwrap()
with
| exn ->
let src = new TaskCompletionSource<_>()
src.SetException(exn)
src.Task
module Task =
let inline bind f task : Task<'a> =
TaskStep.bind f task |> TaskStep.toTask
let inline map f task : Task<'b> =
bind (f >> Value) task
FsToolkit.ErrorHandling implements it here. I'll paste the current version below as it's quite short. It uses the TaskBuilder library Aaron mentioned.
module Task =
let singleton value = value |> Task.FromResult
let bind (f : 'a -> Task<'b>) (x : Task<'a>) = task {
let! x = x
return! f x
}
let map f x = x |> bind (f >> singleton)
Additionally, FSharpPlus has an independent implementation of Task.map here.
Throwing the exception again in the continuation would make for an incorrect stack trace.
It's a mapping from 'A -> 'B, so it's probably best to lay it out explicitly.
let rec map (continuation: 'A -> 'B) (t: Task<'A>) =
let rec map_resolved (task: Task<'A>) =
match task.Status with
| TaskStatus.RanToCompletion -> Task.FromResult(continuation task.Result)
| TaskStatus.Faulted -> Task.FromException<'B>(task.Exception)
| TaskStatus.Canceled -> Task.FromCanceled<'B>(CancellationToken.None)
| _ -> task.ContinueWith(map_resolved).Unwrap()
map_resolved t
Previously had a very compact and comprehensive answer for my question.
I had it working for my custom type but now due to some reason I had to change it to string type which is now causing type mismatch errors.
module AsyncResult =
let bind (binder : 'a -> Async<Result<'b, 'c>>) (asyncFun : Async<Result<'a, 'c>>) : Async<Result<'b, 'c>> =
async {
let! result = asyncFun
match result with
| Error e -> return Error e
| Ok x -> return! binder x
}
let compose (f : 'a -> Async<Result<'b, 'e>>) (g : 'b -> Async<Result<'c, 'e>>) = fun x -> bind g (f x)
let (>>=) a f = bind f a
let (>=>) f g = compose f g
Railway Oriented functions
let create (json: string) : Async<Result<string, Error>> =
let url = "http://api.example.com"
let request = WebRequest.CreateHttp(Uri url)
request.Method <- "GET"
async {
try
// http call
return Ok "result"
with :? WebException as e ->
return Error {Code = 500; Message = "Internal Server Error"}
}
test
type mismatch error for the AsyncResult.bind line
let chain = create
>> AsyncResult.bind (fun (result: string) -> (async {return Ok "more results"}))
match chain "initial data" |> Async.RunSynchronously with
| Ok data -> Assert.IsTrue(true)
| Error error -> Assert.IsTrue(false)
Error details:
EntityTests.fs(101, 25): [FS0001] Type mismatch. Expecting a '(string -> string -> Async<Result<string,Error>>) -> 'a' but given a 'Async<Result<'b,'c>> -> Async<Result<'d,'c>>' The type 'string -> string -> Async<Result<string,Error>>' does not match the type 'Async<Result<'a,'b>>'.
EntityTests.fs(101, 25): [FS0001] Type mismatch. Expecting a '(string -> string -> Async<Result<string,Error>>) -> 'a' but given a 'Async<Result<string,'b>> -> Async<Result<string,'b>>' The type 'string -> string -> Async<Result<string,Error>>' does not match the type 'Async<Result<string,'a>>'.
Edit
Curried or partial application
In context of above example, is it the problem with curried functions? for instance if create function has this signature.
let create (token: string) (json: string) : Async<Result<string, Error>> =
and then later build chain with curried function
let chain = create "token" >> AsyncResult.bind (fun (result: string) -> (async {return Ok "more results"}))
Edit 2
Is there a problem with following case?
signature
let create (token: Token) (entityName: string) (entityType: string) (publicationId: string) : Async<Result<string, Error>> =
test
let chain = create token >> AsyncResult.bind ( fun (result: string) -> async {return Ok "more results"} )
match chain "test" "article" "pubid" |> Async.RunSynchronously with
Update: At the front of the answer, even, since your edit 2 changes everything.
In your edit 2, you have finally revealed your actual code, and your problem is very simple: you're misunderstanding how the types work in a curried F# function.
When your create function looked like let create (json: string) = ..., it was a function of one parameter. It took a string, and returned a result type (in this case, Async<Result<string, Error>>). So the function signature was string -> Async<Result<string, Error>>.
But the create function you've just shown us is a different type entirely. It takes four parameters (one Token and three strings), not one. That means its signature is:
Token -> string -> string -> string -> Async<Result<string, Error>>
Remember how currying works: any function of multiple parameters can be thought of as a series of functions of one parameter, which return the "next" function in that chain. E.g., let add3 a b c = a + b + c is of type int -> int -> int -> int; this means that add3 1 returns a function that's equivalent to let add2 b c = 1 + b + c. And so on.
Now, keeping currying in mind, look at your function type. When you pass a single Token value to it as you do in your example (where it's called as create token, you get a function of type:
string -> string -> string -> Async<Result<string, Error>>
This is a function that takes a string, which returns another function that takes a string, which returns a third function which takes a string and returns an Async<Result<whatever>>. Now compare that to the type of the binder parameter in your bind function:
(binder : 'a -> Async<Result<'b, 'c>>)
Here, 'a is string, so is 'b, and 'c is Error. So when the generic bind function is applied to your specific case, it's looking for a function of type string -> Async<Result<'b, 'c>>. But you're giving it a function of type string -> string -> string -> Async<Result<string, Error>>. Those two function types are not the same!
That's the fundamental cause of your type error. You're trying to apply a function that returns a function that returns function that returns a result of type X to a design pattern (the bind design pattern) that expects a function that returns a result of type X. What you need is the design pattern called apply. I have to leave quite soon so I don't have time to write you an explanation of how to use apply, but fortunately Scott Wlaschin has already written a good one. It covers a lot, not just "apply", but you'll find the details about apply in there as well. And that's the cause of your problem: you used bind when you needed to use apply.
Original answer follows:
I don't yet know for a fact what's causing your problem, but I have a suspicion. But first, I want to comment that the parameter names for your AsyncResult.bind are wrong. Here's what you wrote:
let bind (binder : 'a -> Async<Result<'b, 'c>>)
(asyncFun : Async<Result<'a, 'c>>) : Async<Result<'b, 'c>> =
(I moved the second parameter in line with the first parameter so it wouldn't scroll on Stack Overflow's smallish column size, but that would compile correctly if the types were right: since the two parameters are lined up vertically, F# would know that they are both belonging to the same "parent", in this case a function.)
Look at your second parameter. You've named it asyncFun, but there's no arrow in its type description. That's not a function, it's a value. A function would look like something -> somethingElse. You should name it something like asyncValue, not asyncFun. By naming it asyncFun, you're setting yourself up for confusion later.
Now for the answer to the question you asked. I think your problem is this line, where you've fallen afoul of the F# "offside rule":
let chain = create
>> AsyncResult.bind (fun (result: string) -> (async {return Ok "more results"}))
Note the position of the >> operator, which is to the left of its first operand. Yes, the F# syntax appears to allow that in most situations, but I suspect that if you simply change that function definition to the following, your code will work:
let chain =
create
>> AsyncResult.bind (fun (result: string) -> (async {return Ok "more results"}))
Or, better yet because it's good style to make the |> (and >>) operators line up with their first operand:
let chain =
create
>> AsyncResult.bind (fun (result: string) -> (async {return Ok "more results"}))
If you look carefully at the rules that Scott Wlaschin lays out in https://fsharpforfunandprofit.com/posts/fsharp-syntax/, you'll note that his examples where he shows exceptions to the "offside rule", he writes them like this:
let f g h = g // defines a new line at col 15
>> h // ">>" allowed to be outside the line
Note how the >> character is still to the right of the = in the function definition. I don't know exactly what the F# spec says about the combination of function definitions and the offside rule (Scott Wlaschin is great, but he's not the spec so he could be wrong, and I don't have time to look up the spec right now), but I've seen it do funny things that I didn't quite expect when I wrote functions with part of the function definition on the same line as the function, and the rest on the next line.
E.g., I once wrote something like this, which didn't work:
let f a = if a = 0 then
printfn "Zero"
else
printfn "Non-zero"
But then I changed it to this, which did work:
let f a =
if a = 0 then
printfn "Zero"
else
printfn "Non-zero"
I notice that in Snapshot's answer, he made your chain function be defined on a single line, and that worked for him. So I suspect that that's your problem.
Rule of thumb: If your function has anything after the = on the same line, make the function all on one line. If your function is going to be two lines, put nothing after the =. E.g.:
let f a b = a + b // This is fine
let g c d =
c * d // This is also fine
let h x y = x
+ y // This is asking for trouble
I would suspect that the error stems from a minor change in indentation since adding a single space to an FSharp program changes its meaning, the FSharp compiler than quickly reports phantom errors because it interprets the input differently. I just pasted it in and added bogus classes and removed some spaces and now it is working just fine.
module AsyncResult =
[<StructuralEquality; StructuralComparison>]
type Result<'T,'TError> =
| Ok of ResultValue:'T
| Error of ErrorValue:'TError
let bind (binder : 'a -> Async<Result<'b, 'c>>) (asyncFun : Async<Result<'a, 'c>>) : Async<Result<'b, 'c>> =
async {
let! result = asyncFun
match result with
| Error e -> return Error e
| Ok x -> return! binder x
}
let compose (f : 'a -> Async<Result<'b, 'e>>) (g : 'b -> Async<Result<'c, 'e>>) = fun x -> bind g (f x)
let (>>=) a f = bind f a
let (>=>) f g = compose f g
open AsyncResult
open System.Net
type Assert =
static member IsTrue (conditional:bool) = System.Diagnostics.Debug.Assert(conditional)
type Error = {Code:int; Message:string}
[<EntryPoint>]
let main args =
let create (json: string) : Async<Result<string, Error>> =
let url = "http://api.example.com"
let request = WebRequest.CreateHttp(Uri url)
request.Method <- "GET"
async {
try
// http call
return Ok "result"
with :? WebException as e ->
return Error {Code = 500; Message = "Internal Server Error"}
}
let chain = create >> AsyncResult.bind (fun (result: string) -> (async {return Ok "more results"}))
match chain "initial data" |> Async.RunSynchronously with
| Ok data -> Assert.IsTrue(true)
| Error error -> Assert.IsTrue(false)
0
I have a made a simple function that wraps common F# functions of signature 'a -> 'b -> 'c option to more "C# compliant" function as : 'a -> b -> byref<'c> -> bool. But somehow when I try to wrap such a method in a class I am getting error FS0001 and I can't locate the error.
Code below
open System
open System.Runtime.InteropServices
// Given a function, f: 'a -> 'b -> 'c option returns
// A new function g: 'a -> 'b -> byref<'c> -> bool
let wrapOptionF f a b (g:byref<'c>) =
match f a b with
| Some v ->
do g <- v
true
| None ->
false
let tryDivide (a:int) (b:int) =
match Math.DivRem(a,b) with
| v, 0 -> Some v
| _ -> None
type TryDivideWrapper() =
static member TryDivide(a, b, [<Out>]cRef:byref<int>) : bool =
let f = wrapOptionF tryDivide a b
f cRef
The offending line is f cRef.
This post contains a more in-depth explanation, but in short you can replace your final type definition with the following:
type TryDivideWrapper() =
static member TryDivide(a, b, [<Out>]cRef:byref<int>) : bool =
wrapOptionF tryDivide a b &cRef
The reason for this is that your wrapOptionF takes a byref parameter. However, byref<int> isn't really a type like int or int ref - it's just an indication to the compiler that your parameter should be passed by reference (like out in C#). Once inside your function, however, what you have is a regular int.
Edit: Note that Intellisense will show cRef as having type byRef<int>. However, if you bind another variable to cRef, you'll see that the type you get is a regular int. You can put this line inside TryDivide and then hover your mouse over a to see it:
let a = cRef
Using the & operator tells the compiler that you're passing cRef into f by reference - which is exactly what f needs - making the type system happy. I've tested this using a C# project, and TryDivideWrapper.TryDivide(a, b, out c) works as expected. Add in #MarkSeemann's tryDivide and you should be good to go.
I'm not exactly sure I understand the reason, but this seems to work:
type TryDivideWrapper() =
static member TryDivide(a, b, [<Out>]cRef:byref<int>) : bool =
wrapOptionF tryDivide a b &cRef
BTW, the OP tryDivide implementation throws an exception on tryDivide 1 0. Here's an alternative implementation that works:
let tryDivide (a:int) (b:int) =
if b = 0
then None
else Some (a / b)
FSI:
> tryDivide 1 0;;
val it : int option = None
> tryDivide 10 5;;
val it : int option = Some 2
I'm trying to learn F# and have put together the following code:
open System.Collections.Generic
type Version = Version of int
type AggregateId = AggregateId of int
type IEventdata = interface end
type EventMeta = EventMeta of AggregateId * Version
type Event = Event of EventMeta * IEventdata
type ICommandData = interface end
type CommandMeta = CommandMeta of AggregateId * Version
type Command = CommandMeta * ICommandData
type EventStore = {
GetEvents : AggregateId -> Event seq;
Insert : Event seq -> unit
}
let commandDispatcherFactory (mapping: (Command -> Event seq -> Event seq), es: EventStore) =
(
fun (command: Command) ->
match command with
| (CommandMeta(aggregateId, version), commandData) ->
let events = es.GetEvents (aggregateId)
let resultingEvents = mapping command events
(es.Insert(resultingEvents))
)
type DummyCommand =
| Command1
interface ICommandData
type DummyCommand2 =
| Command2
interface ICommandData
type UnsupportedCommand =
| Command3
let dummyMapping = (fun ((command, commandData):Command) (events: Event seq) ->
match commandData with
| :? DummyCommand as dummyCommand ->
(printfn "Handling Command: %A" dummyCommand)
Seq.empty<Event>
| :? DummyCommand2 as dummyCommand ->
(printfn "Handling Command: %A" dummyCommand)
Seq.empty<Event>
| _ ->
(printfn "Unsupported command: %A" commandData)
Seq.empty<Event>
)
let dummyEs = {
GetEvents = (fun (aggregateId) -> Seq.empty<Event>);
Insert = (fun (events:Event seq) -> ())
}
dummyMapping (CommandMeta(AggregateId(1),Version(1)), Command1) Seq.empty<Event>
dummyMapping (CommandMeta(AggregateId(1),Version(1)), Command2) Seq.empty<Event> // Why does this work?
let commandDispatcher = commandDispatcherFactory(dummyMapping, dummyEs)
commandDispatcher (CommandMeta(AggregateId(1),Version(1)), Command1)
commandDispatcher (CommandMeta(AggregateId(1),Version(1)), Command2) // this doesn't work
I know it might be a wall of text, but I didn't manage to narrow it down. It's the last call to my commandDispatcher that doesn't work, and I can't figure out why. I know there probably are other things to do better, but that is something I'm working on :)
Here's a cut down repro:
let f x (y:System.IComparable) = ()
let g = f 1
g 3
g "test" // not okay
let g' : System.IComparable -> unit = f 1
g' 3
g' "test" // okay
let g'' x = f 1 x
g'' 3
g'' "test" // okay
What happens is that the compiler infers a generic type for g (g:('a -> unit) when 'a :> System.IComparable in the case of my example). However, g is a syntactic value and therefor can't be generic (see questions about the value restriction), so the compiler needs to pick a specific type. If you don't use g at all, you'll get a compiler error. If you use it at exactly one concrete type, the compiler will pick that type. If you use it at more than one type, the compiler will specialize based on the first type used and throw an error after that.
Two workarounds are to either annotate with a non-generic type or to eta-expand so that g isn't a syntactic value.
The compiler is specializing the type of commandDispatcher based on the initial usage. You can fix it with a cast:
commandDispatcher (CommandMeta(AggregateId(1),Version(1)), Command1 :> ICommandData)
commandDispatcher (CommandMeta(AggregateId(1),Version(1)), Command2) //works
I'm quite sure that I run into some kind of limitation, but I do not understand it:
type IRunner =
abstract member Run : (string -> 'a) -> 'a
type T() =
let run4 doFun = doFun "4"
let run5 doFun = doFun "5"
let parseInt s = System.Int32.Parse(s)
let parseFloat s = System.Double.Parse(s)
let doSomething () =
let i = parseInt |> run4
let f = parseFloat |> run4
f |> ignore
// Make it more generic ->
//let doSomething2 (runner:(string->'a)->'b) =
let doSomething2 runner =
// Error on the following lines with both declarations
let i = parseInt |> runner
let f = parseFloat |> runner
f |> ignore
// Want to do something like
let test () =
doSomething2 run4
doSomething2 run5
// Workaround
let workaround (runner:IRunner) =
let run f = runner.Run f
let i = parseInt |> run
let f = parseFloat |> run
f |> ignore
Can somebody bring some light over this? I did not find any related question, sorry if i duplicated something.
The problem is, if doSomething2 has type ((string->'a) -> 'b) -> unit, then 'a and 'b are fixed during each invocation of doSomething2, which isn't what you want - in your case 'a needs to treated as both int and float during a single invocation of doSomething2.
It seems like what you really want is more like: doSomething2 : (forall 'a. (string -> 'a) -> 'a) -> unit, but that kind of direct universal quantification doesn't exist in F#. As you've discovered, the way to work around this is to use a type with a generic method.
And even if F# did support forall types, as I mentioned in a comment inference still wouldn't be possible. Consider your doSomething2 function - we know that runner needs to be able to take an input of type string -> int to some output type and an input of type string -> float to some (possibly different) output type. Here are several different signatures for doSomething2 that all meet this requirement:
forall 'a. 'a -> 'a
forall 'a. (string -> 'a) -> 'a
forall 'a. 'a -> unit
Note that none of these types is more general than the others, they are all incompatible. In the first case, we could pass id to the function, in the second case, we could pass run4 to it, and in the third case, we could pass ignore to it (but none of those functions is compatible with the other possible signatures!).