Monday, October 26, 2009

New features in C# 4.0 - Dynamic Lookup

Dynamic lookup allows you a unified approach to invoking things dynamically. With dynamic lookup, when you have an object in your hand you do not need to worry about whether it comes from COM, IronPython, the HTML DOM or reflection; you just apply operations to it and leave it to the runtime to figure out what exactly those operations mean for that particular object.

This affords you enormous flexibility, and can greatly simplify your code, but it does come with a significant drawback: Static typing is not maintained for these operations. A dynamic object is assumed at compile time to support any operation, and only at runtime will you get an error if it wasn’t so. Oftentimes this will be no loss, because the object wouldn’t have a static type anyway, in other cases it is a tradeoff between brevity and safety. In order to facilitate this tradeoff, it is a design goal of C# to allow you to opt in or opt out of dynamic behavior on every single call.

The dynamic type

C# 4.0 introduces a new static type called dynamic. When you have an object of type dynamic you can “do things to it” that are resolved only at runtime:
   
dynamic d = GetDynamicObject(…);
d.M(7);
  
The C# compiler allows you to call a method with any name and any arguments on d because it is of type dynamic. At runtime the actual object that d refers to will be examined to determine what it means to “call M with an int” on it.

The type dynamic can be thought of as a special version of the type object, which signals that the object can be used dynamically. It is easy to opt in or out of dynamic behavior: any object can be implicitly converted to dynamic, “suspending belief” until runtime. Conversely, there is an “assignment conversion” from dynamic to any other type, which allows implicit conversion in assignment-like constructs:

dynamic d = 7; // implicit conversion
int i = d; // assignment conversion

Dynamic operations

Not only method calls, but also field and property accesses, indexer and operator calls and even delegate invocations can be dispatched dynamically:
  
dynamic d = GetDynamicObject(…);
d.M(7); // calling methods
d.f = d.P; // getting and settings fields and properties
d[“one”] = d[“two”]; // getting and setting thorugh indexers
int i = d + 3; // calling operators
string s = d(5,7); // invoking as a delegate

The role of the C# compiler here is simply to package up the necessary information about “what is being done to d”, so that the runtime can pick it up and determine what the exact meaning of it is given an actual object d. Think of it as deferring part of the compiler’s job to runtime.

The result of any dynamic operation is itself of type dynamic.

Runtime lookup

At runtime a dynamic operation is dispatched according to the nature of its target object d:
COM objects
If d is a COM object, the operation is dispatched dynamically through COM IDispatch. This allows calling to COM types that don’t have a Primary Interop Assembly (PIA), and relying on COM features that don’t have a counterpart in C#, such as indexed properties and default properties.

Dynamic objects

If d implements the interface IDynamicObject d itself is asked to perform the operation. Thus by implementing IDynamicObject a type can completely redefine the meaning of dynamic operations. This is used intensively by dynamic languages such as IronPython and IronRuby to implement their own dynamic object models. It will also be used by APIs, e.g. by the HTML DOM to allow direct access to the object’s properties using property syntax.

Plain objects

Otherwise d is a standard .NET object, and the operation will be dispatched using reflection on its type and a C# “runtime binder” which implements C#’s lookup and overload resolution semantics at runtime. This is essentially a part of the C# compiler running as a runtime component to “finish the work” on dynamic operations that was deferred by the static compiler.

Example

Assume the following code:
dynamic d1 = new Foo();
dynamic d2 = new Bar();
string s;
d1.M(s, d2, 3, null);

Because the receiver of the call to M is dynamic, the C# compiler does not try to resolve the meaning of the call. Instead it stashes away information for the runtime about the call. This information (often referred to as the “payload”) is essentially equivalent to:

“Perform an instance method call of M with the following arguments:
1.    a string
2.    a dynamic
3.    a literal int 3
4.    a literal object null”

At runtime, assume that the actual type Foo of d1 is not a COM type and does not implement
IDynamicObject. In this case the C# runtime binder picks up to finish the overload resolution job based on runtime type information, proceeding as follows:
1.    Reflection is used to obtain the actual runtime types of the two objects, d1 and d2, that did not have a static type (or rather had the static type dynamic). The result is Foo for d1 and Bar for d2.
2.    Method lookup and overload resolution is performed on the type Foo with the call M(string,Bar,3,null) using ordinary C# semantics.
3.    If the method is found it is invoked; otherwise a runtime exception is thrown.

Overload resolution with dynamic arguments

Even if the receiver of a method call is of a static type, overload resolution can still happen at runtime. This can happen if one or more of the arguments have the type dynamic:

Foo foo = new Foo();
dynamic d = new Bar();
var result = foo.M(d);

The C# runtime binder will choose between the statically known overloads of M on Foo, based on the runtime type of d, namely Bar. The result is again of type dynamic.

The Dynamic Language Runtime

An important component in the underlying implementation of dynamic lookup is the Dynamic Language Runtime (DLR), which is a new API in .NET 4.0.

The DLR provides most of the infrastructure behind not only C# dynamic lookup but also the implementation of several dynamic programming languages on .NET, such as IronPython and IronRuby. Through this common infrastructure a high degree of interoperability is ensured, but just as importantly the DLR provides excellent caching mechanisms which serve to greatly enhance the efficiency of runtime dispatch.

To the user of dynamic lookup in C#, the DLR is invisible except for the improved efficiency. However, if you want to implement your own dynamically dispatched objects, the IDynamicObject interface allows you to interoperate with the DLR and plug in your own behavior. This is a rather advanced task, which requires you to understand a good deal more about the inner workings of the DLR. For API writers, however, it can definitely be worth the trouble in order to vastly improve the usability of e.g. a library representing an inherently dynamic domain.

Open issues

There are a few limitations and things that might work differently than you would expect.
•    The DLR allows objects to be created from objects that represent classes. However, the current implementation of C# doesn’t have syntax to support this.
•    Dynamic lookup will not be able to find extension methods. Whether extension methods apply or not depends on the static context of the call (i.e. which using clauses occur), and this context information is not currently kept as part of the payload.
•    Anonymous functions (i.e. lambda expressions) cannot appear as arguments to a dynamic method call.The compiler cannot bind (i.e. “understand”) an anonymous function without knowing what type it is converted to.

One consequence of these limitations is that you cannot easily use LINQ queries over dynamic objects:

dynamic collection = …;
var result = collection.Select(e => e + 5);

If the Select method is an extension method, dynamic lookup will not find it. Even if it is an instance method, the above does not compile, because a lambda expression cannot be passed as an argument to a dynamic operation.
There are no plans to address these limitations in C# 4.0.

Named and Optional Arguments

Named and optional parameters are really two distinct features, but are often useful together. Optional parameters allow you to omit arguments to member invocations, whereas named arguments is a way to provide an argument using the name of the corresponding parameter instead of relying on its position in the parameter list.

Some APIs, most notably COM interfaces such as the Office automation APIs, are written specifically with named and optional parameters in mind. Up until now it has been very painful to call into these APIs from C#, with sometimes as many as thirty arguments having to be explicitly passed, most of which have reasonable default values and could be omitted.

Even in APIs for .NET however you sometimes find yourself compelled to write many overloads of a method with different combinations of parameters, in order to provide maximum usability to the callers. Optional parameters are a useful alternative for these situations.

Optional parameters
A parameter is declared optional simply by providing a default value for it:
  
public void M(int x, int y = 5, int z = 7);
Here y and z are optional parameters and can be omitted in calls:
M(1, 2, 3); // ordinary call of M
M(1, 2); // omitting z – equivalent to M(1, 2, 7)
M(1); // omitting both y and z – equivalent to M(1, 5, 7)

No comments:

Post a Comment