fixnum instance does not have a eigenclass

fixmum_eigenclass.rb

# -*- coding: utf-8 -*-
# $ ruby -vw foobar.rb 
# ruby 1.9.2p0 (2010-08-18 revision 29036) [x86_64-linux]
# foobar.rb:1: syntax error, unexpected tINTEGER
# def 1.hello
#      ^
# foobar.rb:3: syntax error, unexpected keyword_end, expecting $end

class Object
  def metaclass
    class << self; self; end
  end
end

# Fixnum objects do not have metaclass
begin
  binding.eval("def 1.hello;puts \"hello #{self}\";end") #Error: defining hello on 1 did not work
rescue Exception
  puts "Error: defining hello on 1 did not work" #noop
end

puts 1.class #Fixnum
puts 1.metaclass rescue nil #Error, fixnum instance does not have a metaclass. can't define singleton (TypeError)
puts 1.class.metaclass # #<Class:Fixnum>
puts 1.class.metaclass.object_id == 1.class.object_id #false
puts 1.class.metaclass.object_id == 1.class.class.object_id #false 

class Fixnum
  def hello
    puts "hello #{self}"
  end
end

1.hello #hello 1

1.instance_eval { @x = 10 }
puts 1.instance_eval { @x } #10

__END__

dkannan>	why cannot i define a method in the eigenclass of a instance of a Fixnum? 	11:16 AM
 <cdcarter>	fixnums don't have eigenclasses	11:17 AM
 <dkannan>	really. let me check	11:19 AM
 <cdcarter>	TypeError: no virtual class for Fixnum	11:20 AM
 <banisterfiend>	cdcarter: what did you type to get that error?	11:21 AM
 <banisterfiend>	and what ruby version are you on?	11:21 AM
 <cdcarter>	class<<3;self;end	11:21 AM
 <cdcarter>	on 1.8, i should note	11:21 AM
 <banisterfiend>	yeah	11:21 AM
 <banisterfiend>	1.9 has a diff error msg	11:21 AM
 <dkannan>	can't define singleton (TypeError) in 1.9.2-p0	11:22 AM
 <banisterfiend>	dkannan: fixnums can have ivars though	11:22 AM
 <dkannan>	banisterfiend: 1.instance_eval { @x = 10 }; 1.instance_eval { @x };	11:23 AM
 <dkannan>	banisterfiend: how does this work, meaning where is it stored 	11:24 AM
 <banisterfiend>	dkannan: in an ivar table	11:24 AM
 <banisterfiend>	dkannan: where did you think? :P	11:26 AM
 <banisterfiend>	but ok	11:26 AM
 <banisterfiend>	ordinary T_OBJECTs and T_CLASSes and so on have pointers to an ivar table in their structs	11:26 AM
 <banisterfiend>	immediate values (like fixnums) dont, instead their ivar tables are associated to them through a hash table	11:27 AM
 <banisterfiend>	called a 'generic_ivar_table' or some such	11:27 AM
 <banisterfiend>	you pass it the immediate value, and it returns the ivar table for that immedate	11:27 AM
 <banisterfiend>	immediate	11:27 AM
 <banisterfiend>	same with T_DATA objects	11:27 AM
 <banisterfiend>	they all have teh EXIVAR flag set	11:27 AM
 <banisterfiend>	the*	11:27 AM
 <banisterfiend>	which means 'this object has an external ivar table, look in the generic ivar table to get it'	11:28 AM
 <cdcarter>	T_DATAs don't have their own ivar table?	11:28 AM
 <banisterfiend>	im pretty sure they dont	11:28 AM
 <cdcarter>	maybe I don't remember my C API fully	11:29 AM
 <cdcarter>	i thought they were VALUEs that came from Data_Wrap_Struct	11:29 AM
 <banisterfiend>	yeah	11:29 AM
 <cdcarter>	oh right	11:30 AM
 <cdcarter>	they wrap the struct	11:30 AM
 <cdcarter>	so the struct doesn't have an ivar member	11:30 AM
 <cdcarter>	right right right	11:30 AM
→ dhruvasagar (~dhruvasag@118.102.165.194) joined	11:30 AM
 <dkannan>	is this because these values are interned - the datastructure is in a hashmap rather than in that instance	11:30 AM
 <dkannan>	1.object_id == 1.object_id => true	11:31 AM
 <banisterfiend>	dkannan: the values dont have structs associated with them, they're burnt into the VALUE itself, so there's no room for an iv table	11:31 AM
 <dkannan>	what if i want to send a message to a fixnum - but i want it to behave differently based on the value	11:33 AM
 <banisterfiend>	then define the method on Fixnum	11:33 AM
 <banisterfiend>	and have a condition inside the method	11:33 AM
 <banisterfiend>	if self == blah	11:33 AM
 <dkannan>	oh ok	11:34 AM