We Keep Coding

C# generic type inference on delegate types [duplicate]

Given this code:
class C
{
C()
{
Test<string>(A); // fine
Test((string a) => {}); // fine
Test((Action<string>)A); // fine
Test(A); // type arguments cannot be inferred from usage!
}
static void Test<T>(Action<T> a) { }
void A(string _) { }
}
The compiler complains that Test(A) can't figure out T to be string.
This seems like a pretty easy case to me, and I swear I've relied far more complicated inference in other generic utility and extension functions I've written. What am I missing here?
Update 1: This is in the C# 4.0 compiler. I discovered the issue in VS2010 and the above sample is from a simplest-case repro I made in LINQPad 4.
Update 2: Added some more examples to the list of what works.

Test(A);
This fails because the only applicable method (Test<T>(Action<T>)) requires type inference, and the type inference algorithm requires that each each argument be of some type or be an anonymous function. (This fact is inferred from the specification of the type inference algorithm (§7.5.2)) The method group A is not of any type (even though it is convertable to an appropriate delegate type), and it is not an anonymous function.
Test<string>(A);
This succeeds, the difference being that type inference is not necessary to bind Test, and method group A is convertable to the required delegate parameter type void Action<string>(string).
Test((string a) => {});
This succeeds, the difference being that the type inference algorithm makes provision for anonymous functions in the first phase (§7.5.2.1). The parameter and return types of the anonymous function are known, so an explicit parameter type inference can be made, and a correspondense is thereby made between the types in the anonymous function (void ?(string)) and the type parameter in the delegate type of the Test method’s parameter (void Action<T>(T)). No algorithm is specified for method groups that would correspond to this algorithm for anonymous functions.
Test((Action<string>)A);
This succeeds, the difference being that the untyped method group parameter A is cast to a type, thereby allowing the type inference of Test to proceed normally with an expression of a particular type as the only argument to the method.
I can think of no reason in theory why overload resolution could not be attempted on the method group A. Then—if a single best binding is found—the method group could be given the same treatment as an anonymous function. This is especially true in cases like this where the method group contains exactly one candidate and it has no type parameters. But the reason it does not work in C#4 appears to be the fact that this feature was not designed and implemented. Given the complexity of this feature, the narowness of its application, and the existance of three easy work-arounds, I am not going to be holding my breath for it!

I think it's because it's a two-step inference:
It has to infer that you want to convert A to a generic delegate
It has to infer what the type of the delegate parameter should be
I'm not sure if this is the reason, but my hunch is that a two-step inference isn't necessarily easy for the compiler.
Edit:
Just a hunch, but something is telling me the first step is the problem. The compiler has to figure out to convert to a delegate with a different number of generic parameters, and so it can't infer the types of the parameters.

This looks like a vicious circle to me.
Test method expects a parameter of delegate type constructed from generic type Action<T>. You pass in a method group instead: Test(A). This means compiler has to convert your parameter to a delegate type (method group conversion).
But which delegate type? To know the delegate type we need to know T. We didn't specify it explicitly, so compiler has to infer it to figure out the delegate type.
To infer the type parameters of the method we need to know the types of the method arguments, in this case the delegate type. Compiler doesn't know the argument type and thus fails.
In all other cases either type of argument is apparent:
// delegate is created out of anonymous method,
// no method group conversion needed - compiler knows it's Action<string>
Test((string a) => {});
// type of argument is set explicitly
Test((Action<string>)A);
or type parameter is specified explicitly:
Test<string>(A); // compiler knows what type of delegate to convert A to
P.S. more on type inference

You're passing the name of the Method A. The .Net framework CAN convert it to an Action, but it's implicit and it will not take responsibility for it.
But still, a method name is NOT an explicit Action<> Object. And therefor it won't infer the type as an Action type.

I could be wrong, but I imagine the real reason C# cannot infer the type is due to method overloading and the ambiguity that arises. For example, suppose I have the following methods: void foo (int) and void foo (float). Now if I write var f = foo. Which foo should the compiler pick? Likewise, the same problem happens with your example using Test(foo).

Generic Function Type Inference in C# [duplicate]

Given this code:
class C
{
C()
{
Test<string>(A); // fine
Test((string a) => {}); // fine
Test((Action<string>)A); // fine
Test(A); // type arguments cannot be inferred from usage!
}
static void Test<T>(Action<T> a) { }
void A(string _) { }
}
The compiler complains that Test(A) can't figure out T to be string.
This seems like a pretty easy case to me, and I swear I've relied far more complicated inference in other generic utility and extension functions I've written. What am I missing here?
Update 1: This is in the C# 4.0 compiler. I discovered the issue in VS2010 and the above sample is from a simplest-case repro I made in LINQPad 4.
Update 2: Added some more examples to the list of what works.

Test(A);
This fails because the only applicable method (Test<T>(Action<T>)) requires type inference, and the type inference algorithm requires that each each argument be of some type or be an anonymous function. (This fact is inferred from the specification of the type inference algorithm (§7.5.2)) The method group A is not of any type (even though it is convertable to an appropriate delegate type), and it is not an anonymous function.
Test<string>(A);
This succeeds, the difference being that type inference is not necessary to bind Test, and method group A is convertable to the required delegate parameter type void Action<string>(string).
Test((string a) => {});
This succeeds, the difference being that the type inference algorithm makes provision for anonymous functions in the first phase (§7.5.2.1). The parameter and return types of the anonymous function are known, so an explicit parameter type inference can be made, and a correspondense is thereby made between the types in the anonymous function (void ?(string)) and the type parameter in the delegate type of the Test method’s parameter (void Action<T>(T)). No algorithm is specified for method groups that would correspond to this algorithm for anonymous functions.
Test((Action<string>)A);
This succeeds, the difference being that the untyped method group parameter A is cast to a type, thereby allowing the type inference of Test to proceed normally with an expression of a particular type as the only argument to the method.
I can think of no reason in theory why overload resolution could not be attempted on the method group A. Then—if a single best binding is found—the method group could be given the same treatment as an anonymous function. This is especially true in cases like this where the method group contains exactly one candidate and it has no type parameters. But the reason it does not work in C#4 appears to be the fact that this feature was not designed and implemented. Given the complexity of this feature, the narowness of its application, and the existance of three easy work-arounds, I am not going to be holding my breath for it!

I think it's because it's a two-step inference:
It has to infer that you want to convert A to a generic delegate
It has to infer what the type of the delegate parameter should be
I'm not sure if this is the reason, but my hunch is that a two-step inference isn't necessarily easy for the compiler.
Edit:
Just a hunch, but something is telling me the first step is the problem. The compiler has to figure out to convert to a delegate with a different number of generic parameters, and so it can't infer the types of the parameters.

This looks like a vicious circle to me.
Test method expects a parameter of delegate type constructed from generic type Action<T>. You pass in a method group instead: Test(A). This means compiler has to convert your parameter to a delegate type (method group conversion).
But which delegate type? To know the delegate type we need to know T. We didn't specify it explicitly, so compiler has to infer it to figure out the delegate type.
To infer the type parameters of the method we need to know the types of the method arguments, in this case the delegate type. Compiler doesn't know the argument type and thus fails.
In all other cases either type of argument is apparent:
// delegate is created out of anonymous method,
// no method group conversion needed - compiler knows it's Action<string>
Test((string a) => {});
// type of argument is set explicitly
Test((Action<string>)A);
or type parameter is specified explicitly:
Test<string>(A); // compiler knows what type of delegate to convert A to
P.S. more on type inference

You're passing the name of the Method A. The .Net framework CAN convert it to an Action, but it's implicit and it will not take responsibility for it.
But still, a method name is NOT an explicit Action<> Object. And therefor it won't infer the type as an Action type.

I could be wrong, but I imagine the real reason C# cannot infer the type is due to method overloading and the ambiguity that arises. For example, suppose I have the following methods: void foo (int) and void foo (float). Now if I write var f = foo. Which foo should the compiler pick? Likewise, the same problem happens with your example using Test(foo).

Why can't C# infer type from this seemingly simple, obvious case

Given this code:
class C
{
C()
{
Test<string>(A); // fine
Test((string a) => {}); // fine
Test((Action<string>)A); // fine
Test(A); // type arguments cannot be inferred from usage!
}
static void Test<T>(Action<T> a) { }
void A(string _) { }
}
The compiler complains that Test(A) can't figure out T to be string.
This seems like a pretty easy case to me, and I swear I've relied far more complicated inference in other generic utility and extension functions I've written. What am I missing here?
Update 1: This is in the C# 4.0 compiler. I discovered the issue in VS2010 and the above sample is from a simplest-case repro I made in LINQPad 4.
Update 2: Added some more examples to the list of what works.

Test(A);
This fails because the only applicable method (Test<T>(Action<T>)) requires type inference, and the type inference algorithm requires that each each argument be of some type or be an anonymous function. (This fact is inferred from the specification of the type inference algorithm (§7.5.2)) The method group A is not of any type (even though it is convertable to an appropriate delegate type), and it is not an anonymous function.
Test<string>(A);
This succeeds, the difference being that type inference is not necessary to bind Test, and method group A is convertable to the required delegate parameter type void Action<string>(string).
Test((string a) => {});
This succeeds, the difference being that the type inference algorithm makes provision for anonymous functions in the first phase (§7.5.2.1). The parameter and return types of the anonymous function are known, so an explicit parameter type inference can be made, and a correspondense is thereby made between the types in the anonymous function (void ?(string)) and the type parameter in the delegate type of the Test method’s parameter (void Action<T>(T)). No algorithm is specified for method groups that would correspond to this algorithm for anonymous functions.
Test((Action<string>)A);
This succeeds, the difference being that the untyped method group parameter A is cast to a type, thereby allowing the type inference of Test to proceed normally with an expression of a particular type as the only argument to the method.
I can think of no reason in theory why overload resolution could not be attempted on the method group A. Then—if a single best binding is found—the method group could be given the same treatment as an anonymous function. This is especially true in cases like this where the method group contains exactly one candidate and it has no type parameters. But the reason it does not work in C#4 appears to be the fact that this feature was not designed and implemented. Given the complexity of this feature, the narowness of its application, and the existance of three easy work-arounds, I am not going to be holding my breath for it!

I think it's because it's a two-step inference:
It has to infer that you want to convert A to a generic delegate
It has to infer what the type of the delegate parameter should be
I'm not sure if this is the reason, but my hunch is that a two-step inference isn't necessarily easy for the compiler.
Edit:
Just a hunch, but something is telling me the first step is the problem. The compiler has to figure out to convert to a delegate with a different number of generic parameters, and so it can't infer the types of the parameters.

This looks like a vicious circle to me.
Test method expects a parameter of delegate type constructed from generic type Action<T>. You pass in a method group instead: Test(A). This means compiler has to convert your parameter to a delegate type (method group conversion).
But which delegate type? To know the delegate type we need to know T. We didn't specify it explicitly, so compiler has to infer it to figure out the delegate type.
To infer the type parameters of the method we need to know the types of the method arguments, in this case the delegate type. Compiler doesn't know the argument type and thus fails.
In all other cases either type of argument is apparent:
// delegate is created out of anonymous method,
// no method group conversion needed - compiler knows it's Action<string>
Test((string a) => {});
// type of argument is set explicitly
Test((Action<string>)A);
or type parameter is specified explicitly:
Test<string>(A); // compiler knows what type of delegate to convert A to
P.S. more on type inference

You're passing the name of the Method A. The .Net framework CAN convert it to an Action, but it's implicit and it will not take responsibility for it.
But still, a method name is NOT an explicit Action<> Object. And therefor it won't infer the type as an Action type.

I could be wrong, but I imagine the real reason C# cannot infer the type is due to method overloading and the ambiguity that arises. For example, suppose I have the following methods: void foo (int) and void foo (float). Now if I write var f = foo. Which foo should the compiler pick? Likewise, the same problem happens with your example using Test(foo).

Write a method which accepts a lambda expression

I have a method with the following signature:
void MyMethod(Delegate d){};
void MyMethod(Expression exp){};
void MyMethod(object obj){};
However, this fails to compile:
MyMethod((int a) => a)
with the following error:
"Cannot convert lambda expression to type 'object' because it is not a delegate type"
Why doesn't this work?
Edit: I know that this works. The compiler compiles the lambda expression to a delgate in this case I think.
void MyMethod(Func<int, int> d){};
Kind regards,

Because the type System.Delegate isn't a "Delegate". It's just the base class. You have to use a delegate type with the correct signature. Define your Method as follows:
void MyMethod(Func<int, int> objFunc)
EDIT:
MyMethod(object) doesn't work because a lambda expression has no type at it's own, but the type is inferred from the type of the location it is assigned to. So object doesn't work either. You HAVE to use a delegate type with the correct signature.

void MyMethod(Action<int> lambdaHereLol)
{
lambdaHereLol(2);
}
in use:
var hurrDurr = 5;
MyMethod(x => Console.Write(x * hurrDurr));
C# is a statically typed language. The compiler needs to know the Type of everything it deals with. Lambdas are a bit hard to nail down, and sometimes the compiler can't figure it out. In my example above, if MyMethod took an object, the compiler couldn't figure out that x is an int (my example is simple, but there's nothing that says it can't be much more complex and harder to determine). So I have to be more explicit in defining the method that takes my lambda.

A lambda like (int a) => a will fit any delegate that takes an int and returns an int. Func<int,int> is just a single example, and you could easily declare one yourself with delegate int Foo(int x);. In fact this lambda expression will even fit a delegate that takes an int and returns a double, because the result of the lambda (a) is implicitly convertible to double.
In order for a lambda to be assignable to all the delegate types that it would fit, the lambda itself doesn't inherently have a type. Instead it takes on the type of the delegate you're using it as, as long as that's possible. ((int a) => a can't be assigned to Func<byte, byte> of course.)
While both Func<int, int> and the Foo delegate I defined can of course be converted to Delegate, a lambda can not be directly converted to Delegate because it is unclear what its actual signature would then be. After Delegate d = (int a) => a, would d be Foo, or Func<int, int>, or even Func<int, double>? All are valid possibilities, and the compiler has no idea what you intended. It could make a best guess, but C# is not the kind of language that does that kind of guesswork. This is also why you can't do something like var = (int a) => a.
I do think the error message that the compiler gives for Delegate d = (int a) => a; is very unclear:
Cannot convert lambda expression to type 'System.Delegate' because it is not a delegate type
Intuitively you would think Delegate is a delegate type, but that's not how things work. :)

Try this:
void MyMethod(Action<int> func) { }
You need to a strongly-typed delegate as a parameter to the method. The reason the other calls fail is because the C# compiler will not allow you to pass a lambda expression to a method expecting an Object because a lambda expression isn't necessarily always a delegate in all cases. This same rule applies for passing the lambda expression as a Delegate.
When you pass the lambda to a function like I have showed above, the compile can safely assume that you want the lambda expression to be converted to a specific delegate type and does so.

It's simply the nature of the compiler that you need to explicitly cast a delegate object to Delegate when passing it as a parameter of type Delegate. In fact, lambda expressions complicate things even further in that they are not implicitly convertable to delegates in this case.
What you require is a double cast, as such:
MyMethod((Delegate)(Func<int, int>)((int a) => a));
which of course corresponds to the method signature:
void MyMethod(Delegate d);
Depending on your situation, you may want to define a parameter of type Func<int> rather than Delegate (though I would hesitate adding an overload, just because it adds unnecessary complexity in honesty).

The reason this fails is the same reason an expression like "object del = (int a) => a" or even "var del = (int a) => a" fails. You might think that the compiler could figure out the type of your lambda expression since you explicitly give the type of the argument, but even knowing that the expression takes an int and returns an int, there are a number of delegate types it could be converted to. The Func delegate type is the one most used for generic functions such as this, but that is just a convention and nothing the compiler is aware of.
What you need to do is to cast the lambda expression to a concrete delegate type in order to have the compiler select the Delegate overload, either using the normal cast syntax (Func)((int a) => a), or using the delegate constructor syntax new Func((int a) => a).
Also, you normally do not want to use the untyped Delegate class unless you need to invoke something differently depending on the number of argument it accepts. It's almost always better to accept a Func or Action for things like callbacks.

Convert this delegate to an anonymous method or lambda

I am new to all the anonymous features and need some help. I have gotten the following to work:
public void FakeSaveWithMessage(Transaction t)
{
t.Message = "I drink goats blood";
}
public delegate void FakeSave(Transaction t);
public void SampleTestFunction()
{
Expect.Call(delegate { _dao.Save(t); }).Do(new FakeSave(FakeSaveWithMessage));
}
But this is totally ugly and I would like to have the inside of the Do to be an anonymous method or even a lambda if it is possible. I tried:
Expect.Call(delegate { _dao.Save(t); }).Do(delegate(Transaction t2) { t2.Message = "I drink goats blood"; });
and
Expect.Call(delegate { _dao.Save(t); }).Do(delegate { t.Message = "I drink goats blood"; });
but these give me
Cannot convert anonymous method to type 'System.Delegate' because it is not a delegate type** compile errors.
What am I doing wrong?
Because of what Mark Ingram posted, seems like the best answer, though nobody's explicitly said it, is to do this:
public delegate void FakeSave(Transaction t);
Expect.Call(delegate { _dao.Save(t); }).Do( new FakeSave(delegate(Transaction t2) { t.Message = expected_msg; }));

That's a well known error message. Check the link below for a more detailed discussion.
http://staceyw1.wordpress.com/2007/12/22/they-are-anonymous-methods-not-anonymous-delegates/
Basically you just need to put a cast in front of your anonymous delegate (your lambda expression).
In case the link ever goes down, here is a copy of the post:
They are Anonymous Methods, not
Anonymous Delegates.
Posted on December 22, 2007 by staceyw1
It is not just a talking point because
we want to be difficult. It helps us
reason about what exactly is going on.
To be clear, there is *no such thing
as an anonymous delegate. They don’t
exist (not yet). They are "Anonymous
Methods" – period. It matters in how
we think of them and how we talk about
them. Lets take a look at the
anonymous method statement "delegate()
{…}". This is actually two different
operations and when we think of it
this way, we will never be confused
again. The first thing the compiler
does is create the anonymous method
under the covers using the inferred
delegate signature as the method
signature. It is not correct to say
the method is "unnamed" because it
does have a name and the compiler
assigns it. It is just hidden from
normal view. The next thing it does
is create a delegate object of the
required type to wrap the method. This
is called delegate inference and can
be the source of this confusion. For
this to work, the compiler must be
able to figure out (i.e. infer) what
delegate type it will create. It has
to be a known concrete type. Let
write some code to see why.
private void MyMethod()
{
}
Does not compile:
1) Delegate d = delegate() { }; // Cannot convert anonymous method to type ‘System.Delegate’ because it is not a delegate type
2) Delegate d2 = MyMethod; // Cannot convert method group ‘MyMethod’ to non-delegate type ‘System.Delegate’
3) Delegate d3 = (WaitCallback)MyMethod; // No overload for ‘MyMethod’ matches delegate ‘System.Threading.WaitCallback’
Line 1 does not compile because the
compiler can not infer any delegate
type. It can plainly see the signature
we desire, but there is no concrete
delegate type the compiler can see.
It could create an anonymous type of
type delegate for us, but it does not
work like that. Line 2 does not
compile for a similar reason. Even
though the compiler knows the method
signature, we are not giving it a
delegate type and it is not just going
to pick one that would happen to work
(not what side effects that could
have). Line 3 does not work because
we purposely mismatched the method
signature with a delegate having a
different signature (as WaitCallback
takes and object).
Compiles:
4) Delegate d4 = (MethodInvoker)MyMethod; // Works because we cast to a delegate type of the same signature.
5) Delegate d5 = (Action)delegate { }; // Works for same reason as d4.
6) Action d6 = MyMethod; // Delegate inference at work here. New Action delegate is created and assigned.
In contrast, these work. Line 1 works
because we tell the compiler what
delegate type to use and they match,
so it works. Line 5 works for the
same reason. Note we used the special
form of "delegate" without the parens.
The compiler infers the method
signature from the cast and creates
the anonymous method with the same
signature as the inferred delegate
type. Line 6 works because the
MyMethod() and Action use same
signature.
I hope this helps.
Also see:
http://msdn.microsoft.com/msdnmag/issues/04/05/C20/

What Mark said.
The problem is that Do takes a Delegate parameter. The compiler can't convert the anonymous methods to Delegate, only a "delegate type" i.e. a concrete type derived from Delegate.
If that Do function had took Action<>, Action<,> ... etc. overloads, you wouldn't need the cast.

The problem is not with your delegate definition, it's that the parameter of the Do() method is of type System.Delegate, and the compiler generated delegate type (FakeSave) does not implicitly convert to System.Delegate.
Try adding a cast in front of your anonymous delegate:
Expect.Call(delegate { _dao.Save(t); }).Do((Delegate)delegate { t.Message = "I drink goats blood"; });

Try something like:
Expect.Call(delegate { _dao.Save(t); }).Do(new EventHandler(delegate(Transaction t2) { t2.CheckInInfo.CheckInMessage = "I drink goats blood"; }));
Note the added EventHandler around the delegate.
EDIT: might not work since the function signatures of EventHandler and the delegate are not the same... The solution you added to the bottom of your question may be the only way.
Alternately, you could create a generic delegate type:
public delegate void UnitTestingDelegate<T>(T thing);
So that the delegate is not Transaction specific.