First Draft: cast(ref T)... as shorthand for *cast(T*)&...

[Paul Backus]

On Tuesday, 25 February 2025 at 12:46:53 UTC, Dukc wrote:
> If anything, we want a library helper function like Paul 
> suggested, but I'm slightly reserved even about that. When 
> doing reinterpret casts, it's extremely important to understand 
> exactly how the cast works. Any abstraction over that always 
> tends to more or less hide it.

IMO `reinterpretCast!T` is actually more explicit about what's 
going on than `*cast(T*) &`.

The pointer used in these casts is a language-level abstraction 
that doesn't actually exist at runtime. If you look at the 
generated assembly, it's just a `mov` instruction.

For example, this code:

     ulong doubleAsInt(double num)
     {
         return *cast(ulong*) #
     }

     size_t ptrAsSize(void* p)
     {
         return *cast(size_t*) &p;
     }


...produces the following assembly:

     ulong example.doubleAsInt(double):
             push    rbp
             mov     rbp, rsp
             movsd   qword ptr [rbp - 8], xmm0
             mov     rax, qword ptr [rbp - 8] ; *cast(ulong*) &num
             pop     rbp
             ret

     ulong example.ptrAsSize(void*):
             push    rbp
             mov     rbp, rsp
             mov     qword ptr [rbp - 8], rdi
             mov     rax, qword ptr [rbp - 8] ; *cast(size_t*) &p
             pop     rbp
             ret

Godbolt link: https://d.godbolt.org/z/xGGKnPrWs