Re: retry semantics changed

Yukihiro Matsumoto wrote:
> But how about retry being inside of the block?

In the simple case, it's ok as far as I can tell, since it's being 
explicitly used:

CODE:

public
def go(x, y = (print :c; 1))
   (print :d; 1).times { retry }
end
(print :a; self).go((print :b; self))

OUTPUT:

abcdddddddddd...

However there are implicit cases that make it just as dangerous:

CODE:

public
def go(x, y = (print :c; 1))
   p = proc { retry }
   (print :d; 1).times &p
end
(print :a; self).go((print :b; self))

OUTPUT:

abcdddddddddd...

CODE:

public
def go(x, y = (print :c; 1), &p)
   (print :d; 1).times &p
end
(print :a; self).go((print :b; self)) { retry }

OUTPUT:

abcdabcdabcdabcdabcd....

CODE:

public
def go(x, y = (print :c; 1), &p)
   (print :d; 1).times &p
end
(print :a; self).go((print :b; self), &proc{retry})

OUTPUT:

abcdabcdabcdabcdabcd....

None of these behaviors seems particularly useful, and they're certainly 
not predictable without experimentation. I'm sure there's other similar 
cases when you have nested blocks/procs interacting. At this point I 
don't see retry being useful (or at least, useful enough to counter its 
danger) outside of a rescue. If its behaviors were clearly specified, 
perhaps it could be improved, but in its current state I wouldn't 
recommend anyone use it outside rescue.

- Charlie