From: David Vallner <david@...>
Date: 2006-01-14T06:58:04+09:00
Subject: Re: Range#member? Oddity

On Fri, 13 Jan 2006 22:16:14 +0100, Matthew Desmarais <desmarm@gmail.com>  
wrote:

> James Edward Gray II wrote:
>> I'm not understanding what I am seeing here.  Can anyone please explain  
>> why the last line of this session gives *false* as an answer?
>>
>> >> range = ("1".."10")
>> => "1".."10"
>> >> range.to_a
>> => ["1", "2", "3", "4", "5", "6", "7", "8", "9", "10"]
>> >> range.member?("1")
>> => true
>> >> range.member?("2")
>> => false
>>
>> James Edward Gray II
>>
>>
>>
> Hi,
>
> There was some discussion about this in the recent past.  If my memory  
> serves me right (certainly an infrequent happening), the issue that  
> you're running into is that Range#member?  is implemented as:
> class Range
>     def member?(val)
>        if self.exclude_end?
>            (self.first <= val) and (val < self.last)
>        else
>           (self.first <= val) and (val <= self.last)
>        end
>     end
> end
>
> You should find this in both 1.8.2 and 1.8.4 I think.
>
> There's a previous thread on ruby-talk about it, here's a link to  
> somewhere near the conclusion of the discussion:
> http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/167194
>
> Matthew
>


This would seem to make it an oddity of String#succ, that behaving  
automagically, not generating successors with respect to String comparison.

E.g. for any Integers i, i.succ > i. For some strings, that does't hold  
true.

Bottom line: Don't use strings when you're really using numbers. Like in  
mathemathical contexts. D'oh. You could possibly hack around that in Range  
code to provide for data types where generating successors is inconsistent  
with comparison, but I wouldn't like to see that.

David Vallner