29 – undefined behaviour of: scope(...){ return X; }

void main(){
	scope(exit){
		return 0; // should fail to compile
	}
}
int main(){
	scope(exit){
		return 0; // undefined behaviour
	}
	return 1;
}
test cases:
http://dstress.kuehne.cn/nocompile/s/scope_08_A.d
http://dstress.kuehne.cn/nocompile/s/scope_08_C.d
http://dstress.kuehne.cn/nocompile/s/scope_08_B.d
http://dstress.kuehne.cn/undefined/scope_08_D.d
http://dstress.kuehne.cn/undefined/scope_08_E.d
http://dstress.kuehne.cn/undefined/scope_08_F.d 

It shouldn't fail to compile. It means the same thing as a return inside a finally block. Not a bug.

But it *does* fail to compile now. For example, this program ...
-------------------
 import std.stdio;
 int Foo(int x)
 {
 scope(exit) if (x < 5) return x+9;
 return x;
 }
 void main()
 {
 for(int i = 0; i < 10; i++)
 writefln("IN %s OUT %s", i, Foo(i));
 }
-------------------
gives this compiler message ...
"test.d(4): return statements cannot be in finally bodies"

(In reply to comment #1)
> It shouldn't fail to compile. It means the same thing as a return inside a
> finally block. Not a bug.
The correlation wiht finally isn't documented.
http://www.digitalmars.com/d/statement.html#try
# A FinallyStatement may not exit with a goto, break, continue, or
# return; nor may it be entered with a goto.
Thus the code below is illegal, yet compiles:
# int test(){
# scope(exit) return 0;
# }
# void test(){
# scope(exit) return 0;
# }

Fixed 0.161