From: mengx@...
Date: 2002-01-02T23:22:33+09:00
Subject: [ruby-talk:30041] Re: A Bug or something else?

> 
>  Well, in ruby a private method is a method that can be called only in a
>  function form (i.e. it can have only self as a receiver). Here you are in
>  a special where ruby want to convert a value to a string, and in this case
>  it can call any methods (public, private, ...)
> 

Thanks. I guess this is a special case when nil is used in something like string
substitution. Below are some examples:

class NilClass
	private :to_s
end
nil.to_s   => undefined method.  Yet
"#{nil}"   => ""

I'd consider it against POLS. Apparently, the C program does not use "to_s"
method. Secondly, as I said before, it would be nice that "nil" does not
respond to "to_s, to_i, to_a" at all.

> m> about the reasoning for the existence of methods for "nil", such as "to_a",
> m> "to_i", "to_s". Can an exception  be raised for this scenario and the one
> m> below?
> 
> m> a = nil
> 
> m> a == 1    => false  (even after the undef)
> 
>  Well here you have written something like
>    
>    a.== 1
> 
>  this mean that if you undef #== for NilClass it will work
> 
> pigeon% ruby
> class NilClass
>    undef ==
> end
> a = nil
> a == 1
> ^D
> -:5: undefined method `==' for nil (NameError)
> pigeon%
> 
>  it's still work with
> 
>   1 == a
> 

That is exactly what I meant. It is impossible to achieve what a static
typed language can do for this purpose, but for "nil", it would not be too
hard to conduct a check for "nil", a special and straitforward case in
either C interpreter or Ruby itself. Does it sound reasonable? I am a little
obsessed with this since it is a great danger to silently corrupt data due to
common program errors. 

Thanks

-Ted