Let's have a following simplified example:
void Foo<T>(IEnumerable<T> collection, params T[] items)
{
// ...
}
void Foo<C, T>(C collection, T item)
where C : ICollection<T>
{
// ...
}
void Main()
{
Foo((IEnumerable<int>)new[] { 1 }, 2);
}
Compiler says:
The type 'System.Collections.Generic.IEnumerable' cannot be used as type parameter 'C' in the generic type or method 'UserQuery.Foo(C, T)'. There is no implicit reference conversion from 'System.Collections.Generic.IEnumerable' to 'System.Collections.Generic.ICollection'.
If I change Main
to:
void Main()
{
Foo<int>((IEnumerable<int>)new[] { 1 }, 2);
}
It will work ok. Why compiler does not choose the right overload?
Your question is answered here.
http://blogs.msdn.com/ericlippert/archive/2009/12/10/constraints-are-not-part-of-the-signature.aspx
Please also read the approximately one million comments telling me that I am wrong for some interesting additional commentary on this issue.
My guess is that the compiler chooses the best match before it uses the generic constraint. In your example the method with the constraint is preferable because it doesn't have a
params
last parameter.Edit - Eric Lippert confirms this in his answer.