Make (someIntPtr is null) work

[GSPP]

Testing an IntPtr for null requires comparing to IntPtr.Zero. While this is OK, it could be simpler:

IntPtr someIntPtr;

if (someIntPtr is null) { ... } //1
if (someIntPtr == null) { ... } //2

Can this be made to work?

(2) actually works if you don't use C# strict mode (see dotnet/roslyn#18975). So it's not supposed to work and you're not supposed to do that. It would be nice if this was supported.

[RikkiGibson]

Moving to csharplang as this appears to be a request to change the language.

[GSPP]

Thanks.

[tannergooding]

IntPtr is not a pointer. It is a pointer-bit integer (in the same way Int32 is a 32-bit integer). It's name actually comes from C/C++ where there is intptr_t

Because it is a pointer sized integer, it is sometimes used to hold or otherwise manipulate pointers, addresses, or related data.

However, it is ultimately still an integer just as int and long are and the comparison against null is typically a bug in these cases (and was/is certainly a common source of bugs in C/C++).

It would not be, IMO, good to allow a comparison to null here. Instead, you're likely better off using nint and directly comparing against the literal 0. This will also result in smaller IL and ultimately better JIT throughput

[333fred]

2 "works", but does not actually do what you think it does. It does not compare against IntPtr.Zero, it compares against null the reference: this is always false. In general, I don't think we are interested in treating IntPtr or UIntPtr different from any other struct type: they cannot be null.