Question

How good is C# type inference? I read somewhere that it’s only for local variables? Does it work for class level attributes? For method signatures? Method return types? etc.

Answer 1

There are a few main kinds of type inference in C#:

• Implicitly typed local variables:

  • Only for local variables
  • Only when the value is assigned as part of the declaration
  • Value cannot be null
  • Value cannot be a lambda expression, anonymous method or method group (without a cast)
  • The compile-time type of the value is used for the type of the variable
  • Any further uses of the variable are only checked against the type determined by the initial declaration+assignment; they don’t contribute to the inference itself.

• Generic method type argument inference, i.e. you don’t specify the type arguments in a call to a generic method, the compiler figures them out based on the arguments.

  • Would be really handy to have this for generic types as well as generic methods
  • Really handy anyway – LINQ would be hard or impossible to use without it
  • Anonymous types would be fairly useless without it
  • Really complicated rules, even the spec is wrong in a few places

• Lambda expression parameter type inference

  • Compiler tries to work out the types of the parameters for lambda expressions based on the context in which it’s used
  • Usually works pretty well, in my experience

• Array type inference, e.g. new[] { 'Hi', 'there' } instead of new string[] { 'Hi', 'there' }

  • Various small restrictions, nothing major

I’ve probably forgotten some other features which might be called ‘type inference’. I suspect you’re mostly interested in the first, but the others might be relevant to you too 🙂