82. Generics and type errors#

One of the significant advantages of using generics is that they give us compile-time errors instead of run-time errors. Generics allow developers to catch type errors during the compilation process rather than at run-time. Writing a class like Pair without generics would mean that type mismatches or invalid type operations could go unnoticed until the code is executed, leading to potential runtime exceptions.

Remember the generic class Pair<T1, T2> from the chapter on generic types? Here’s a simplified version of it:

class Pair<T1, T2>
    public T1 Item1 { get; set; }
    public T2 Item2 { get; set; }

Is it possible to achieve the same level of reusability before we had generics or do we have to write different versions of the same code for each type? If we want to keep static type safety we have to write duplicated code. If we’re willing to sacrifice static type safety then we can either write code that works on the top-type object and then downcast back to whatever type we had before or use the dynamic keyword and then cast back to whatever type we had before. Since we’ve given up static type safety however we run the risk of writing code that crashes with run-time errors.

Key point

With generics, we can write reusable code without giving up static type safety. This means that type errors are identified at compile-time rather than at run-time. Which prevents type-related errors from reaching end users and streamlines debugging.

Let’s look at what both of these non-generic solutions would look like. First out is the solution that uses the top-type object that all types implicitly either inherit from or can be boxed into.

class ObjectPair
    public object Item1 { get; set; }
    public object Item2 { get; set; }

We can now instantiate the class and store elements of any type.

ObjectPair objectPair = new ObjectPair { Item1 = 10, Item2 = "Apples" };

However, since the compile-time type of both items in every instance of ObjectPair is object we have to use downcasting if we want to get back to the run-time type of the item. This is fine as long as we manage to downcast to the correct type.

int item1 = (int)objectPair.Item1; // Compiles and runs.

However, if we downcast to the wrong type the code still compiles but we will now get a run-time exception. Boom!

int item2 = (int)objectPair.Item2; // Compiles but throws exception.
System.InvalidCastException: Unable to cast object of type 'System.String' to type 'System.Int32'.

What if we instead make use of the dynamic keyword? It’s the same story. By using the dynamic keyword we tell the compiler that the type of the underlying object could be absolutely anything. Using dynamic we leave the realm of static typing. Consequently we can compile code that tries to convert our values to types that they actually are not. Only to blown up with an exception at run-time.

class DynamicPair
    public dynamic Item1 { get; set; }
    public dynamic Item2 { get; set; }
DynamicPair dynamicPair = new DynamicPair { Item1 = 10, Item2 = "Apples" };
int item1 = dynamicPair.Item1; // Compiles and runs.
int item2 = dynamicPair.Item2; // Compiles but throws exception.
Microsoft.CSharp.RuntimeBinder.RuntimeBinderException: Cannot implicitly convert type 'string' to 'int'

With generics, such errors are caught during compilation, ensuring that if you have a method expecting a certain type, it cannot inadvertently receive an unexpected type. This compile-time type checking enhances code safety, predictability, and robustness, minimizing surprises during execution and simplifying debugging.

What happens if we try to write the same incorrect code when using our generic Pair<T1, T2> type? Are we allowed to run the code? No, with generics, type-related issues are caught at compile-time, ensuring that we maintain type integrity from the outset If it doesn’t compile, we can’t run it.

Pair<int, string> genericPair = new Pair<int, string>()
    Item1 = 10,
    Item2 = "Apples"

int item1 = genericPair.Item1;
int item2 = genericPair.Item2; // Does not compile due to type error.
(8,13): error CS0029: Cannot implicitly convert type 'string' to 'int'

By getting a compile-time error instead of a run-time error we ensure that this error never lands in the hands of the end user.