From: Gary Wright <gwtmp01@...>
Date: 2007-12-08T13:39:10+09:00
Subject: Re: Clarification of retry change


On Dec 7, 2007, at 10:54 PM, Charles Oliver Nutter wrote:
> Michal Suchanek wrote:
>> Hello
>> When talking about the places where retry is allowed .. what does  
>> this
>> imply for retry inside if, case, and while?
>> I do not really use retry nor am I familiar with the grammar trees  
>> you
>> keep talking about. Just want to remind that there are branching
>> constructs inside which the retry might make sense (when placed  
>> inside
>> a rescue).
>
> The scary thing is that what you're saying sounds totally sane, but  
> retry has never worked that way. retry outside of a rescue body has  
> always been a cross-method affair, causing the stack to unwind as  
> many calls as necessary to get back to a call that accepts a block  
> (and I think, unwind all the way back to the call that produced the  
> block where retry is happening...whew!).
>
> So the places you're talking about would all be forbidden locations  
> for retry in 1.9. What you want is redo, which already works for  
> looping constructs and is quite normal and predictable.

You lost me there.  retry is certainly legal and unambiguous
within an if/case statement:

begin
   puts "running.."
   raise
rescue
   if rand(2)%2 == 0
     retry
   end
end

Using retry within a while loop (within a rescue) raises the block/ 
iterator
question yet again but if/case shouldn't be problematic at all, right?

Gary Wright