Syntax error in generated code inside zig-cache (translate-c?)

I'm trying to get my project working with the zig changes from the last day or two and ran into an error. Here is a branch where this error is happening: https://github.com/dbandstra/oxid/commits/not-working

I'm on Ubuntu right now.

/home/dbandstra/oxid/zig-cache/o/DH8YSD9-pmAXMXLFnwJZqo87mh-lr6V49BszWVU-To157k5V3X2S6NNthdcdKFt0/cimport.zig:1256:26: error: expected token '}', found '.'
                        }.?.* = _val;
                         ^

Here's a paste of the relevant part of that file. I marked the bad line with a comment.

pub fn SDL_memset4(dst: ?*c_void, val: Uint32, dwords: usize) void {
    var _n: usize = (dwords +% c_ulong(3)) / c_ulong(4);
    var _p: [*c]Uint32 = @ptrCast([*c]Uint32, dst);
    var _val: Uint32 = val;
    if (dwords == c_ulong(0)) return;
    __switch: {
        __case_0: {
            switch (dwords % c_ulong(4)) {
                c_ulong(0) => break :__case_0,
                c_ulong(3) => break :__case_1,
                c_ulong(2) => break :__case_2,
                c_ulong(1) => break :__case_3,
                else => break :__switch,
            }
        }
        while (true) {
            __case_3: {
                __case_2: {
                    __case_1: {
                        x: {
                            const _ref = &_p;
                            const _tmp = _ref.*;
                            _ref.* +%= 1;
                            break :x _tmp;
                        }.?.* = _val; // <------------------------------------------ here
                    }
                    x: {
                        const _ref = &_p;
                        const _tmp = _ref.*;
                        _ref.* +%= 1;
                        break :x _tmp;
                    }.?.* = _val;
                }
                x: {
                    const _ref = &_p;
                    const _tmp = _ref.*;
                    _ref.* +%= 1;
                    break :x _tmp;
                }.?.* = _val;
            }
            x: {
                const _ref = &_p;
                const _tmp = _ref.*;
                _ref.* +%= 1;
                break :x _tmp;
            }.?.* = _val;
            if (!x: {
                const _ref = &_n;
                _ref.* -%= 1;
                break :x _ref.*;
            }) break;
        }
    }
}

P.S. Zig syntax highlighting in github! Nice!

edit: here's the c code from SDL_stdinc.h:

/* Note that memset() is a byte assignment and this is a 32-bit assignment, so they're not directly equivalent. */
SDL_FORCE_INLINE void SDL_memset4(void *dst, Uint32 val, size_t dwords)
{
#if defined(__GNUC__) && defined(i386)
    int u0, u1, u2;
    __asm__ __volatile__ (
        "cld \n\t"
        "rep ; stosl \n\t"
        : "=&D" (u0), "=&a" (u1), "=&c" (u2)
        : "0" (dst), "1" (val), "2" (SDL_static_cast(Uint32, dwords))
        : "memory"
    );
#else
    size_t _n = (dwords + 3) / 4;
    Uint32 *_p = SDL_static_cast(Uint32 *, dst);
    Uint32 _val = (val);
    if (dwords == 0)
        return;
    switch (dwords % 4)
    {
        case 0: do {    *_p++ = _val;   /* fallthrough */
        case 3:         *_p++ = _val;   /* fallthrough */
        case 2:         *_p++ = _val;   /* fallthrough */
        case 1:         *_p++ = _val;   /* fallthrough */
        } while ( --_n );
    }
#endif
}

[andrewrk]

Oh wow, I didn't know that Zig was successfully translating SDL_memset4 before. That's pretty neat!

That's probably a good candidate to become a regression test of translate-c.

I'm pretty sure this was broken by the C pointers change - which although it caused this regression, it actually is going to result in much cleaner translate-c output, and make a lot of translate-c code easier to implement.

I think this regression can definitely get fixed before 0.4.0 comes out.

[Hejsil]

That switch translation also looks very incorrect, but I don't think:

        case 0: do {    *_p++ = _val;   /* fallthrough */
        case 3:         *_p++ = _val;   /* fallthrough */
        case 2:         *_p++ = _val;   /* fallthrough */
        case 1:         *_p++ = _val;   /* fallthrough */
        } while ( --_n );

can be easily be translated to Zig

[andrewrk]

Yeah it definitely looks wrong. I think it can work in theory though.

[Hejsil]

Actually, this broke with the grammar changes in #1685. Basically, we have this rule at stmt level:

Statement
    <- ...
     / LabeledStatement
     ...
     / AssignExpr SEMICOLON

LabeledStatement <- BlockLabel? (Block / LoopStatement)

AssignExpr expands into all other expressions. LabeledStatement is before AssignExpr so that rule is being parsed first, succeeds and then Statement succeeds. Then the parser tries to end the __case_1 block, which then fails the parsing.

The parser parses the grammar correctly, but the new grammar just does not allow this syntax. I can't come up with a solution on top of my head right now.

Edit: Well, a solution would be to make translate-c conform to this new grammar. Idk if we want to put in the work to make x:{}.* = 1; valid at stmt level.