From: Robert Klemme <bob.news@...>
Date: 2004-03-15T03:54:34+09:00
Subject: Re: float allocator


"Anders Borch" <spam@deck.dk> schrieb im Newsbeitrag
news:t1Z4c.112164$jf4.6624704@news000.worldonline.dk...
> I have been testing ruby 1.8 on my windows box before I dare adopt it on
> my linux box. I came across this odd difference:
>
> In ruby 1.6 I can dup a Float:
>
> $ ruby -v
> ruby 1.6.8 (2002-12-24) [i586-linux-gnu]
> $ irb
> irb(main):001:0> a = 3.14
> => 3.14
> irb(main):002:0> a.dup
> => 3.14
>
> In ruby 1.8 I cannot:
>
> $ ruby -v
> ruby 1.8.1 (2003-12-25) [i386-cygwin]
> $ irb
> irb(main):001:0> a = 3.14
> => 3.14
> irb(main):002:0> a.dup
> NoMethodError: allocator undefined for Float
>          from (irb):2:in `dup'
>          from (irb):2
>
> I did a bit of testing and found out that neither Fixnum, Float or
> Bignum can be dup'ed. Is this intentional, and if so then how do I go
> about copying the descendents of Numeric?

You don't.  You gain nothing by dup'ing Integer, Float etc. because these
are immutable anyway.  You don't gain anything by copying these.

OTOH, for a uniform invocation these instance's #dup methods could be
defined to return self.

Regards

    robert