From: devesh <devesh.agrawal@...>
Date: 2006-11-21T05:40:13+09:00
Subject: Fast 32 bit ints in ruby ?

Hi Folks,

I want an efficient implementation of an Ipaddress class. The ones I
have seen (for eg: lib/ipaddr.rb) all represent things internally in
the form of a Bignum. And that is turning out to be a bottleneck for
me. Ideally I would want to create some kind of a class, in which I can
wrap in a normal C int and then operate with it directly. Does anyone
know of how to do this ?. Another way would be to create a struct
wrapping this int and then access that from my C extension (Like it is
done with the Jukebox example in the Pickaxe book). I am new to ruby
and to mixing C and ruby. What would be a good way to achieve the
following interface:

class IPAddress

	# Some code to declare @magicCInt as a C type int
	# magicCInt is a good'ol C type int (of full 32 bits)
	# It is not a Bignum/Fixnum

	def initizialize
		@magicCInt = 0
	end
	def & (other)
		p = self.clone
		p.magicCInt &= other.magicCInt
	end
	def == (other)
		return (other.magicCInt == self.magicCInt)
	end
	alias_method :== :eql?
	#.. And other methods
end

Can I achieve this using RubyInline. It would be great if the more
experienced of you could provide me with a code sketch to get me
started ?

PS: My apologies if this question has been asked before, I would be
grateful if you could point me to some other thread which seeks to
answer this.

PPS: I am writing some networkingish code and comparing two ipaddress's
is sadly my biggest bottleneck (Bignum#==)

Thanks, 
-Devesh