From: Rob Biedenharn <Rob@...>
Date: 2009-11-11T22:35:51+09:00
Subject: Re: Roman Numerals (Arrgh!)


On Nov 10, 2009, at 10:30 PM, Gennady Bystritsky wrote:
> On Nov 10, 2009, at 6:43 PM, Rob Biedenharn wrote:
>> On Nov 10, 2009, at 8:49 PM, Gennady Bystritsky wrote:
>>> On Nov 10, 2009, at 5:37 PM, Rob Biedenharn wrote:
>>>> On Nov 10, 2009, at 6:56 PM, Rick Barrett wrote:
>>>>> I have a homework assignment where I have to convert an inputted
>>>>> integer
>>>>> into Roman Numerals. I was doing fine building upon the previous
>>>>> steps
>>>>
>>>> P.S. I saw Gennady's post, but I'd argue that no one would really
>>>> want
>>>> to solve the problem that way. And Nik's way doesn't give any help
>>>
>>> Why not that way? It works and it is short ;-)
>>>
>>> Gennady.
>>
>> Not that it doesn't work, just that it might be easier/quicker to
>> rewrite than to understand. Short isn't *always* a virtue and if
>> there's a reason (performance?) to be terse, then there ought to be a
>> comment by way of explanation.
>>
>> There are few truly necessary uses of Enumerable#inject and this  
>> isn't
>
> I respectfully disagree. inject() is a very powerful facility that  
> allows quick elegant implementation of many tasks at hand. I usually  
> start with an inject()-based implementation (where appropriate) and  
> then may refactor it later if performance is not satisfactory. The  
> latter does not happen very often, though. As for the virtue, the  
> less code you write the fewer bugs get in -- especially when you use  
> well tested standard facilities like inject().

Well, I won't argue the fact that inject is a powerful abstraction. I  
*will* argue that it would be much easier to inject a bug (pun  
intended ;-) into your original terse code than something that is a  
bit longer, but tries to do less at each step. Changing/swapping array  
subscripts, for example.

>
>> one of them, but if it were, I'd make it at least slightly more
>> readable:
>>
>>  def roman_numeral(number)
>>    $numerals.inject(["", number]) { |(answer, number), (arabic,
>> roman)|
>
> For that thanks a lot -- somehow I missed the fact it is possible in  
> ruby.
>
>>      [ answer + roman * (number / arabic), number % arabic ]
>>    }.first
>>  end
>>
>> And with an appropriately formed $numerals array, the "subtraction
>> rule" is already covered. ;-)
>
> Would you please elaborate? You mean same algorithm and "better"  
> $numerals would cover mapping of 4 => "IV", 9 => "IX", etc.?
>
>>
>> -Rob


Yes, that's it exactly. Testing/using 9 before 5, you find IX (and  
stop) before you go to VIIII. You can't repeat one of the "subtraction  
groups" the way that you can repeat the individual numerals: 8 is  
always VIII and never IVIV or IIX

Rob Biedenharn		http://agileconsultingllc.com
Rob@AgileConsultingLLC.com