DMD 2.0.6
This behaviour seems inconsistent and unintuitive:
void main() {
int[3] a = [1,2,3]; // The array is fine
a = [4, a[0], 6];
struct S { // The stuct is the problem
int a, b, c;
}
S s = S(1,2,3);
s = S(4, s.a, 6);
assert(a == [4,1,6]); // What I'd expect
assert(s == S(4,4,6)); // Unhelpful
}
Setting the struct writes s.a before evaluating it while the
reverse is true of the array assignment. GDC
does what I'd expect and gives both as 4,1,6. This seems to be a bug to me, it creates an easy to miss bug and behaves differently to another common data structure and to the same data structure with a different compiler.
Creating a custom constructor for the struct fixes the issue:
struct S {
int a, b, c;
this(int a, int b, int c)
{
this.a = a;
this.b = b;
this.c = c;
}
}
assert(s == S(4,1,6));
Comment #1 by clugdbug — 2013-09-23T22:33:00Z
This works in CTFE. I think it's a bug in DMD's glue layer.