Is this a bug with goto?

[Andrei Alexandrescu]

On 04/17/2010 11:11 AM, Jerry Quinn wrote:
> In the spec it says that it's illegal to skip an initialization using goto.  Unless I'm mistaken, the code below does that for b, s, and c.  However, it compiles without complaint.
>
> So, should the compiler be complaining, or is the text about goto really saying that behavior is undefined.  Obviously the latter makes it easier for the compiler writer :-)
>
> void foo() {
>    goto L0;
>      int b;
>      struct S { int a; this(int c) { a=c; } }
>      S s = S(5);
>      class C { int a; }
>      C c = new C;
>   L0: ;
>      s.a++;
>      c.a++;
> }
>
> thanks,
> Jerry

(Hi Jerry! Glad to see you're ramping up participation lately.) Yes, 
that's a bug. There are many other bugs related to jumps, including in 
switch statements.

Andrei