add built-in complex number types for C ABI compatibility

[thestinger]

These are sadly not just treated as a simple struct on various platforms and have ABI requirements. I think support for the existing C math library is important because this isn't something we want to recreate on every platform.

[nikomatsakis]

On Thu, Nov 14, 2013 at 10:15:43PM -0800, Daniel Micay wrote:

These are sadly not just treated as a simple struct on various
platforms and have ABI requirements. I think support for the
existing C math library is important because this isn't something we
want to recreate on every platform.

By "built-in", do you mean lang items? I'd imagine that is sufficient.

[thestinger]

@nikomatsakis: Yeah, language items would work.

(In fact, it would be nice to move to language items for all built-in types in order to allow defining methods on them. The constant expression evaluation and syntax wouldn't have to change.)

[huonw]

Example of this in C:

#include<complex.h>

struct test {
    float x, y;
};
struct test s() {
    struct test ret = { 1.0, 2.0 };
    return ret;
}

float complex c() {
    return 1.0 + 2.0 * I;
}

Compiled with clang -S complex.c -O --target=i686-unknown-linux gives, respectively:

s:                                      # @s
# BB#0:
    movl    4(%esp), %eax
    movl    $1073741824, 4(%eax)    # imm = 0x40000000
    movl    $1065353216, (%eax)     # imm = 0x3F800000
    retl    $4

c:                                      # @c
# BB#0:
    movl    $1065353216, %eax       # imm = 0x3F800000
    movl    $1073741824, %edx       # imm = 0x40000000
    retl

i.e. the struct (equivalent to num::Complex<f32>) is returned on the stack but _Complex is returned in registers.

[steveklabnik]

I'm pulling a massive triage effort to get us ready for 1.0. As part of this, I'm moving stuff that's wishlist-like to the RFCs repo, as that's where major new things should get discussed/prioritized.

This issue has been moved to the RFCs repo: rust-lang/rfcs#793