Cannot test for void return_type

[rosshadden]

Describe the bug

You seem to be unable to test whether a FunctionData#return_type is void.

Reproduction Steps

struct Struct {}

fn (s Struct) func() {}

fn main() {
	$for method in Struct.methods {
		println('${method.name}: ${method.return_type}')

		$if method.return_type {
			println('this runs but it is true regardless of the return type')
		}
		$if method.return_type is ? {
			println('this never runs but I think it should')
		}
		$if method.return_type !is ? {
			println('this never runs either and that makes sense')
		}
		$if method.return_type == 1 {
			println('this runs which is kind of expected')
		}
		$if method.return_type == 2 {
			println('this runs too which is REALLY WEIRD')
		}
	}
}

$ v run .
func: 1
this runs but it is true regardless of the return type
this runs which is kind of expected
this runs too which is REALLY WEIRD

Expected Behavior

Based on this I think method.return_type is void should be a test that works, or at least is ? which is what the compiler yells at you into using.

I also don't understand why return_type is true when compared against any integer. It should either work as expected with return_type == 1 working in this case (and == 42 being made to not work), or if that was never intended than no integer comparison should pass.

This should probably be the output of the above example:

$ v run .
func: 1
this runs but it is true regardless of the return type
this never runs but I think it should
this runs which is kind of expected

Current Behavior

• return_type is void says to "use ? instead of ?void"
• return_type is ?void says the same
• return_type is ? compiles but does not pass
• return_type == 1 is a false positive, because it turns out comparing return_type to any int passes

Possible Solution

No response

Additional Information/Context

No response

V version

V 0.4.10 875c165

Environment details (OS name and version, etc.)

V full version	V 0.4.10 b76b1ee.875c165
OS	linux, "Artix Linux"
Processor	32 cpus, 64bit, little endian, AMD Ryzen 9 7950X3D 16-Core Processor
Memory	4.14GB/62GB
V executable	/home/ross/x/v/v
V last modified time	2025-05-02 14:46:13
V home dir	OK, value: /home/ross/x/v
VMODULES	OK, value: /home/ross/.vmodules
VTMP	OK, value: /tmp/v_1000
Current working dir	OK, value: /tmp/ciea
Git version	git version 2.49.0
V git status	weekly.2025.17-39-g875c1654
.git/config present	true
cc version	cc (GCC) 14.2.1 20250207
gcc version	gcc (GCC) 14.2.1 20250207
clang version	clang version 19.1.7
tcc version	tcc version 0.9.28rc 2025-02-13 HEAD@f8bd136d (x86_64 Linux)
tcc git status	thirdparty-linux-amd64 696c1d84
emcc version	N/A
glibc version	ldd (GNU libc) 2.41

Note

You can use the 👍 reaction to increase the issue's priority for developers.

Please note that only the 👍 reaction to the issue itself counts as a vote.
Other reactions and those to comments will not be taken into account.

[huly-for-github]

Connected to Huly®: V_0.6-22765

[rosshadden]

Note that @Xyndra found some possibly disturbing things in this area when investigating the compiled C code.

[Xyndra]

The problematic compiler code can be found here, but I couldn't manage to get it working:
https://github.com/vlang/v/blob/master/vlib/v/gen/c/comptime.v#L603

I fixed the impossible condition by locking the second half of the write behind if got_type.idx() != 1 {, but there was another problem that made the code inside the if get ignored

[JalonSolov]

It actually makes logical sense... functions don't return void. They return something, or they don't return anything at all. Seems like there should be a different check for this...

Yes, yes, I know, C represents no return with void, but V doesn't do void (as part of the language).

[rosshadden]

Yeah I can see it both ways, as I'm sure you can. If return_type is taken as "the type of the returned value" then what you said makes sense. If it is instead taken as "the return signature of the function" then I think it still makes sense to be void. And I would posit that the latter is the more universal way to think about things like reflection.

[Xyndra]

you could make the returntype optional

[JalonSolov]

The problem is that the comptime stuff needs to work with non-optional return types of nothing at all.

The problem with void is that V doesn't have void as a keyword, so how can you compare to something that doesn't exist?

[felipensp]

I'll fix the $if var.return_type == 1 { part.

[Xyndra]

you could make void a const and make is work with numbers

[JalonSolov]

It needs to be something in V, and we don't want void in V.