zig segfaults with array and pointer code

[Akuli]

I'm new to zig, so my code might be doing something that's a horribly bad idea. Anyway, it makes zig crash with no output.

// run with:  zig test nameofthisfile.zig

const A = struct {
    b_list_pointer: *const []B,
};
const B = struct {
    a_pointer: *const A,
};

const b_list: []B = []B{ };
const a = A{ .b_list_pointer = &b_list };

test "segfault bug" {
    const assert = @import("std").debug.assert;
    const obj = B{ .a_pointer = &a };
    assert(obj.a_pointer == &a);    // this makes zig crash
}

Edit: information about versions of things:

akuli@Akuli-Desktop:~$ zig version
0.3.0+bbe857be
akuli@Akuli-Desktop:~$ uname -a
Linux Akuli-Desktop 4.9.0-8-amd64 #1 SMP Debian 4.9.130-2 (2018-10-27) x86_64 GNU/Linux
akuli@Akuli-Desktop:~$ lsb_release -a
No LSB modules are available.
Distributor ID:	Devuan
Description:	Devuan GNU/Linux 2.0 (ascii)
Release:	2.0
Codename:	ascii
akuli@Akuli-Desktop:~$ 

I can't try this with the latest master because I'm having some issues with cmake.

[Akuli]

I managed to get cmake and llvm 7 and all the things to work, and turns out the zig on master no longer has this bug.

[Akuli]

The bug doesn't seem to be quite gone yet, even on latest master. This code makes the compiler crash with zig from latest master (0.3.0+c90c2568):

// run with:  zig test nameofthisfile.zig

const std = @import("std");

const A = struct {
    pointer: *B,
};

pub const B = struct {
    // the one with ? works, the one without ? segfaults zig

    //pointer_array: ?[]*A,
    pointer_array: []*A,
};

var b_value = B { .pointer_array = []*A { } };

test "basic stuff" {
    std.debug.assert(&b_value == &b_value);
}

This seems to have something to do with types that refer to each other. Interestingly, the problem goes away if the reference is made nullable. Is this the idiomatic way to create cyclic-referring things, or is this a bug?

Edit: pub and var don't seem to have anything to do with this problem. The same behaviour happens without pub, and const instead of var.

[emekoi]

possibly related: #1850.

[Akuli]

I don't think that's related. This one is a segfault, not an assertion error.

[Akuli]

Maybe #1735 is related?

[andrewrk]

I don't think that's related. This one is a segfault, not an assertion error.

An assertion in a debug build is likely a segfault in a release build.

[andrewrk]

This is fixed in master branch. I don't know when it was fixed, but I added your tests as additional regression tests, just to be safe.

[Akuli]

I think it might be good to figure out when exactly this got fixed, and what caused it to get fixed. Note that the first code example got "fixed" at some point, but changing it slightly showed that it wasn't quite fixed after all.