matz / zetabits.com (Yukihiro Matsumoto) writes:

> I think changes 1, 2 and 4 are OK.  But I don't really agree with
> changes 3 and 5, because:
> 
>   * it requires object creation unnecessary for most of the cases.

True, although you _could_ tell at parse time and generate a different
node when the object was required.

>   * UnboundMethod has information of name AND class it belongs, so
>     that changes to functions and statements are bigger than simply
>     retrieve symbol from it.
> 
> Shugo Maeda once proposed making "def" to return symbol name of the
> defined method.

Yup: that would work very nicely too.

So, do we have an RCR?


Dave