cgen incorrect order in for loop

[Xyndra]

Describe the bug

VV_LOC int main__find_word_end(Array_rune text, int start) {
	int i = start;
	for (;;) {
		bool _t1 = (i < text.len);
		bool _t2 = (Array_rune_contains(_const_main__consonants, (*(rune*)builtin__array_get(text, i))));
		if (!( _t1 && ( _t2 || Array_rune_contains(_const_main__vowels, (*(rune*)builtin__array_get(text, i)))))) break;
		i++;
	}
	return i;
}

Reproduction Steps

fn find_word_end(text []rune, start int) int {
	mut i := start
	for i < text.len && (consonants.contains(text[i]) || vowels.contains(text[i])) {
		i++
	}
	return i
}

Expected Behavior

works in the correct order (left to right)

Current Behavior

part get taken out, leading to V panic: array.get: index out of range (i,a.len):61, 61

Possible Solution

No response

Additional Information/Context

No response

V version

V 0.4.12 6762c9a

Environment details (OS name and version, etc.)

V full version	V 0.4.12 c221b32.6762c9a
OS	windows, Microsoft Windows 11 Pro 22635 64-bit
Processor	8 cpus, 64bit, little endian, 11th Gen Intel(R) Core(TM) i7-1165G7 @ 2.80GHz
Memory	34.62GB/63.67GB
V executable	D:\software\v\v.exe
V last modified time	2025-09-26 19:51:30
V home dir	OK, value: D:\software\v
VMODULES	OK, value: C:\Users\Sammy.vmodules
VTMP	OK, value: C:\Users\Sammy\AppData\Local\Temp\v_0
Current working dir	OK, value: D:\Code\v\bwinf_44_1_2
Git version	git version 2.51.0.windows.1
V git status	weekly.2025.16-830-g6762c9a6
.git/config present	true
cc version	cc (Rev2, Built by MSYS2 project) 13.2.0
gcc version	gcc (Rev2, Built by MSYS2 project) 13.2.0
clang version	clang version 21.1.2
msvc version	N/A
tcc version	tcc version 0.9.27 (x86_64 Windows)
tcc git status	thirdparty-windows-amd64 b425ac82
emcc version	emcc (Emscripten gcc/clang-like replacement + linker emulating GNU ld) 4.0.3 (a9651ff57165f5710bb09a5fe52590fd6ddb72df)
glibc version	N/A

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.

[dy-tea]

This works:

fn find_word_end(text []rune, start int) int {
	mut i := start
	for i < text.len {
		if consonants.contains(text[i]) || vowels.contains(text[i]) {
			break
		}
		i++
	}
	return i
}

I don't think this is a bug, since you are using && both sides need to be evaluated before the decision can be made to stop or not. It won't break immediately when i < text.len is encountered because the right side is required to see if the expression it true or not.

So when it reaches the index after the end of a string with no word end such as hello, i < text.len will be false and the right side needs to be checked so it will index out of bounds.

[Xyndra]

since you are using && both sides need to be evaluated before the decision can be made to stop or not

@dy-tea
No, this is wrong, and C also handles this differently. If it is false, the other side does not need to be evaluated.

#include <stdio.h>

int main() {
    int x = 0;

    if (x && (++x)) {
        printf("Inside if\n");
    } else {
        printf("Inside else, x = %d\n", x);
    }

    return 0;
}

Inside else, x = 0

[Xyndra]

The asm for the code above

[dy-tea]

#include <stdio.h>

int main() {
    int x = 1;
    if (x && (--x))
        printf("true, x: %d\n", x);
    else
        printf("false, x: %d\n", x);
    return 0;
}

Prints false, x: 0

[Xyndra]

yes, since the first statement is true. If the first statement is false, it will not execute. That is the reason it is a problem. i < text.len is false if it will lead to an error

[dy-tea]

Yeah I see your point, since the first comparison is false the right doesn't need to be checked.

[Xyndra]

fn print_true(i int) bool {
	println(i)
	return true
}

const other_list = [3, 4]

fn main() {
	list := [1, 2, 3]
	index := 3
	if index < list.len && list[index] == 3 {
		println('Index is within bounds and value is 3')
	} else {
		println('Index is out of bounds or value is not 3')
	}
	if index < list.len && print_true(list[index]) {
		println('Index is within bounds')
	} else {
		println('Index is out of bounds')
	}
	if index < list.len && other_list.contains(list[index]) {
		println('Index is within bounds and value is in other_list')
	} else {
		println('Index is out of bounds or value is not in other_list')
	}
	if index < list.len && (other_list.contains(list[index]) || other_list.contains(list[index])) {
		println('Index is within bounds and value is in other_list')
	} else {
		println('Index is out of bounds or value is not in other_list')
	}
}

Index is out of bounds or value is not 3
Index is out of bounds
Index is out of bounds or value is not in other_list
V panic: array.get: index out of range (i,a.len):3, 3
v hash: 35bbb54
C:/Users/Sammy/AppData/Local/Temp/v_0/cgen_if_order.01K66FC32HPCHDQ4SE3XAJ85R9.tmp.c:3383: at builtin___v_panic: Backtrace
C:/Users/Sammy/AppData/Local/Temp/v_0/cgen_if_order.01K66FC32HPCHDQ4SE3XAJ85R9.tmp.c:3414: by builtin__panic_n2
C:/Users/Sammy/AppData/Local/Temp/v_0/cgen_if_order.01K66FC32HPCHDQ4SE3XAJ85R9.tmp.c:3012: by builtin__array_get
C:/Users/Sammy/AppData/Local/Temp/v_0/cgen_if_order.01K66FC32HPCHDQ4SE3XAJ85R9.tmp.c:5109: by main__main
C:/Users/Sammy/AppData/Local/Temp/v_0/cgen_if_order.01K66FC32HPCHDQ4SE3XAJ85R9.tmp.c:5146: by wmain
00434110 : by ???
00434273 : by ???
7ffcb877259d : by ???

[Xyndra]

therefore, please add back the "bug" type, since it is

[dy-tea]

Ah sorry I'm not sure how it was removed, I'm having internet problems right now. I will also look into this issue, thanks for explaining it to me.

[Xyndra]

this is potentially related (the part about resetting if branches): https://github.com/vlang/v/pull/25399/files#diff-4b4444571bba392cad6c9324bb16bf645af6b61192dbae1b795e46af61730a9e