From: Morton Goldberg <m_goldberg@...>
Date: 2006-12-06T13:00:25+09:00
Subject: Re: Turtle Graphics (#104)

On Dec 5, 2006, at 9:00 PM, Dema wrote:

> Thanks for pointing that out. I don't even know how all the sample
> drawings were right with that huge bug in the code.

A bug in Turtle#toward has no effect on reproducing the sample  
designs because toward is not used in any of the sample turtle  
scripts. Turtle#toward is not part of core turtle graphics -- it is  
part of the optional non-local-geometry extensions found in some, but  
by no means all, turtle graphic packages. Implementing toward is can  
be considered an extra-credit exercise.

> Morton Goldberg wrote:
>
>> Your solution passes all the unit tests I supplied and is certainly
>> good enough to reproduce all the sample designs. So you have good
>> reason to think it's completely correct. However, one of the optional
>> methods has a problem.
>>
>>>    # Turn to face the given point.
>>>    def toward(pt)
>>>      @heading = atan(pt[0].to_f / pt[1].to_f) / DEG
>>>    end

IMO, Turtle#toward is the most difficult turtle command to get fully  
right. The version I posted as part of my solution is correct, but  
it's not the best that can be done. A better implementation would  
have been:

    # Turn to face the given point.
    def toward(pt)
       x2, y2 = pt
       must_be_number(x2, 'pt.x')
       must_be_number(y2, 'pt.y')
       x1, y1 = xy
       set_h(atan2(x2 - x1, y2 - y1) / DEG)
    end

But that's not what I had when I wrote the quiz -- this version  
incorporates an improvement I saw in Matthew Moss' solution.

Regards, Morton