C++ Bug or (my) Interpretation error

[John Fine]

I've never heard of any rule specifying whether the left side of == is 
evaluated before or after the right side.  I'm pretty sure there is no 
such rule.  That means the sequence is at the compiler's convenience.

I would expect (but would never rely on) the compiler to place the 
harder side first in most cases.  That is what you are seeing.   The 
side of the == containing the ++ was evaluated first in each example.

Arthur Schwarz wrote:
> For the Short Code Snippet we have the Following Results. I am using gcc-version 3.4.4 under Cygwin. 
>
> I would have expected the 2nd result to have been equal with the post-increment occurring after the comparison operation. Does the precedence of the closing ')' force the unary post-increment to execute before the comparison operation.
>
> I would have expected that the last result would have been not equal. What seems to be happening is either the original 'num' is being compared to itself or the pre-increment 'num' is being compared to itself. At first blush, this appears wrong. Have I misunderstood the semantics?
>
> skidmarks
>
> /************** Short Code Snippet *************/
>
> int main() {
>
>         int       num   = 10;
>
>         if (  num++ == num  ) { printf("(  num++ == num  )\n"); }
>                        else   { printf("(  num++ != num  )\n"); }
>         if (  num   == num++) { printf("(  num   == num++)\n"); }
>                        else   { printf("(  num   != num++)\n"); }
>         if (++num   == num)   { printf("(++num   == num  )\n"); }
>                        else   { printf("(++num   != num  )\n"); }
>         if (  num   == ++num) { printf("(  num   == ++num)\n"); }
>                        else   { printf("(  num   != ++num)\n"); }
> }
> /************** Following Results *************/
> (  num++ != num  )   OK
> (  num   != num++)   Not so OK
> (++num   == num  )   OK
> (  num   == ++num)   Not So OK
>
>