clang::Cursor::typedef_type should return Option<Type>

[fitzgen]

The FFI call that typedef_type makes will return an invalid type if the cursor's referent is not a typedef. We should check whether the referent is a typedef and return None if it is not, and make the FFI call and return Some if it is a typedef. This would be one less foot gun.

I can mentor anyone who'd like to pick up this bug.

[highfive]

Please make a comment here if you intend to work on this issue. Thank you!

[codingHahn]

How difficult is this? I am just getting my foots wet with rust. I am almost halfway through the Rust book form the website.

[fitzgen]

@codingHahn, I think you should be able to handle this, it is pretty straight forward.

You don't actually need to do the check before the FFI call, you can instead check the validity of the resulting cursor. Something like this:

let inner = Type {
    x: unsafe { clang_getTypedefDeclUnderlyingType(self.x) }
};
if inner.is_valid() { Some(inner) } else { None }

Then, you'll need to fix up the places where we call this method. For each call site: if the call site is explicitly checking the resulting cursor for validity, change that to an is_some()/is_none()/if let Some(...) = ... { ... } check; if the call site is not checking the validity of the result, change it to a .expect("expected inner typedef type") call to unwrap the Option and panic if we get an invalid cursor.

Does that make sense? Would you like to take this issue?

[codingHahn]

Ok. I should implement the check inside the function and return Some() or None? And every calling that checks the validity of this function should be wrapped into is_some() etc.? And every other function calling should have .expect(sth)? Then a am working on it in the next couple days.

[fitzgen]

Yep that sounds right! Let me know if you run into any roadblocks.

[codingHahn]

I'll let you know :)

[impowski]

@codingHahn Have you done any progress on that or I can take it?