[solved] Which []u8 type is the most compatible with C?

[golden creek]

☝️ basically

[round summit]

[*c]u8

[golden creek]

right, but that IS the C type

I mean native/normal zig slice that can be sent to C

[round summit]

Other than that, it depends on the context. For a C string ?[*:0]u8

[idle spear]

there is no one answer to this question, since the semantics of []u8 dont directly exist in C, can you provide a code sample of what the C side looks like?

[golden creek]

passing const thing = ""; to C where a const char* is expected

[idle spear]

just pass "", ittl coerce

[golden creek]

but I need to pass the string around my zig code

[idle spear]

then [:0]u8

[round summit]

[:0]u8 should be fine. Then pass the .ptr property to C

[empty hinge]

wouldn't it just be

struct foo {
    char *ptr;
    size_t len;
}

(not sure)

[idle spear]

no

[golden creek]

the .ptr idea might be the way to go, yeah

[idle spear]

[]u8 has no set memory layout, the compiler is free to lay it out in any way

[idle spear]

.ptr on [:0]u8 returns [*:0]u8, which should freely coerce to a [*c]u8

[golden creek]

✍️

[hard notch]

in the future, this will only be usable in auto-translated files: https://github.com/ziglang/zig/issues/2984

if you're writing code by hand, any of the other pointer types are preferred, e.g. [*:0]u8 if you're using the C ABI and it needs to be null-terminated

[golden creek]

pub const A = [:0]const u8;
pub const B = [*:0]const u8;

btw, what is the difference from Zig's code from using these two?

is it reasonable to use B in general, or is there any reason to avoid it and prefer A?

[hoary zenith]

Top has length information

Bottom have to iterate through the values it points to to find lenght

[golden creek]

ah, i see. definitely A for the zig code then, tyty