Operator precedence is wrong when casted value is left-shifted in if/match expr

[lemoncmd]

Describe the bug

I'm so afraid of this bug that I can't even use any expression

Reproduction Steps

println(u8(1) << 1 | 1)
println(match 0 {
        0 { u8(1) << 1 | 1 }
        else { 0 }
})
println(if 0 == 0 {
        u8(1) << 1 | 1
} else {
        0
})

Expected Behavior

3
3
3

Current Behavior

3
2
2

u8(1) << (1 | 1) is calculated instead (why?)

Possible Solution

No response

Additional Information/Context

No response

V version

V 0.4.3 043ebb8

Environment details (OS name and version, etc.)

OS: Arch Linux

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.

[felipensp]

Code generated:

VV_LOCAL_SYMBOL void main__main(void) {
        println(u8_str(((((u8)(1)) << 1) | 1)));
        println(u8_str(((0 == (0))? ((((u8)(1)) << 1)) : (0))));
        println(u8_str((true ? ((((u8)(1)) << 1)) : (0))));
}

So it is not about precedence, but not taking care of the whole expression when parsing.

[lemoncmd]

a := 1
println(match 0 {
        0 { u8(1) << 1 | a }
        else { 0 }
})

generates

VV_LOCAL_SYMBOL void main__main(void) {
        int a = 1;
        println(u8_str(((0 == (0))? ((((u8)(1)) << (1 | a))) : (0))));
}

so it is about precedence

[felipensp]

Ah right. I've checked it with:

fn main() {
	println(if false {
			u8(1) << 1 | 2
	} else {
			0
	})
}

Generates:
println(u8_str((false ? ((((u8)(1)) << 3)) : (0))));

[Krchi]

I have the same problem, I handled it locally, but it still has problem in array

fn test_test() {
	mut arr := [u8(1), 9]
	arr << 3
	t := if true {
		arr[0] << 1 | arr[1]
	} else {
		1
	}
	println(t) // print 0, should be 11
}

u8(id) is CastExpr, but arr[0] is IndexExpr, I cant get what type store in array

[lemoncmd]

I found it radically difficult to solve this kind of problem:

mut a := [0]
b := 1
c := match 0 {
    0 {
        a << 1 | 1 // compiler have to parse it as a << (1 | 1)
        b << 1 | 1 // compiler have to parse it as (b << 1) | 1
    }
    else {
        0
    }
}

This is completely context-sensitive.

[felipensp]

I found it radically difficult to solve this kind of problem:

mut a := [0]
b := 1
c := match 0 {
    0 {
        a << 1 | 1 // compiler have to parse it as a << (1 | 1)
        b << 1 | 1 // compiler have to parse it as (b << 1) | 1
    }
    else {
        0
    }
}

This is completely context-sensitive.

Yes, for Ident it is. But for the CastExpr treated here it is okay for now.