IsInstanceOf<T>() is not consistent accross types

[dupdob]

While working on #66 I realized that IsInstanceOf<Person>() fails for null values (NullRefExceprion), but it works there: Check.That( (int?) null).IsInstanceOf<int?>().

It raises multiple questions:
1. is this ok?
I do not think so.

2. What should be the behaviour?
Based on the check name, I propose that null always fail

3. What should we do for Nullables?
Reminder: Nullable types cannot be identified as such at compilation time. At run time, they are either null or a value of the underlying type. As such, GetType() behaves as expected, either throwing a NullRefExc or the appropriate struct type. Furthermore Nullable cannot be inherited from.

It implies that the nullableness can only be checked at compilation time, Therefore IsInstanceOf<int ?> does not bring any value. By the same token, Check.That((int?) 4).IsInstanceOf<T>() can only succeed for T= int?

[dupdob]

Now I understand where it comes from. IsInstanceOf has no interest for actual value type.
But it does for boxed type

[dupdob]

Ultimately, I re implemented IsInstanceOf with the same contract, but it is now made available to all types

[tpierrain]

Since:

• You finally find a way to check whether a Type is nullable or not (i.e. the IsNullable(this Type...) of the ExtensionsCommonHelpers)
• It was very odd to have messages such as: The checked value is not of the given type... The checked value [null] of type [int?] The given type: [int?]

I finally changes the implementation so that a nullable type without a value may succeed if it is the same nullable type.

[dupdob]

Ok, as I usually say for such edge cases: let's see the feedback we get, if
any :-)

[tpierrain]

I think that there is indeed no real use case for such edge case checks (on
the type of a nullable instance without a value), but the previous error
messages was so screaming at us: "stricto sensu, you can´t say that!" ;-)
that I finally could´t let it be...

T

[dupdob]

I fully agree with your statement, but I lack the energy for longer
explaination today ;-)
I was thinking: one extra reason to get rid of those (checks)
Thanks for checking message quality