As an experiment, I wrote some code that looks like
class MyClass
@var = 100
def hello
p "hello"
end
end
I know that if I do MyClass.new that @var is not defined on that object, rather I think this would define @var on the class MyClass.
Is there a practical usage to this?
It does have a use: Class variables. The normal Ruby class variable implementation, @@, shares the same variable between a superclass and its subclasses:
class A
@@v = 0
def self.v; @@v; end
def self.v=(val); @@v=val; end
end
class B < A; end
A.v #-> 0
A.v= 3 #-> 3
B.v #->3
B.v= 42 #-> 42
A.v #-> 42
Obviously, this is pretty useless (except for the fact that, unlike class instance variables, class variables are also accessible directly in instance methods, as opposed to through self.class). But the same example with class instance variables:
class A
@v = 0
def self.v; @v; end
def self.v=(val); @v=val; end
end
class B < A; end
A.v #-> 0
A.v= 3 #-> 3
B.v= 42 #-> 42
A.v #-> 3
Also, class instance variables can harness all of the metaprogramming already written for instance variables, like so:
class Foo
class << self
attr_accessor :v #Uses a class instance variable
end
end
If you make accessor methods for @var
MyClass.singleton_class.class_eval { attr_accessor :var }
then you can treat @var like a class variable. The difference is that it won't be inherited by subclasses of MyClass.
_why takes advantage of this in Dwemthy's Array.
