# Redefining @my\_attr= **URL:** \ **Category:** Ruby\ **Created:** [February 11, 2009, 5:02am UTC](https://www.ruby-forum.com/t/redefining-my-attr/160029 "2009-02-11T05:02:37Z")\ **Posts on this page:** 3\ **Page:** 1
**Author:** ![mikec](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/977dab/32.png) [@mikec](https://www.ruby-forum.com/u/mikec)\ **Post date:** [February 11, 2009, 5:02am UTC](https://www.ruby-forum.com/t/redefining-my-attr/160029/1 "2009-02-11T05:02:37Z")
I want to write a generic validator that ensures an attr value is in a list before assigning, and if not, assigns a default. It works for self.attrib=‘value’, and Dummy.new.attrib=‘value’, but not @attrib=‘value’ within the class. Any way to do that? Here’s the code: module Validator def validate\_is\_member\_of( attrib, list, default ) original\_method = instance\_method( “#{attrib}=”.to\_sym ) define\_method( “#{attrib}=”.to\_sym ) do |val| instance\_variable\_set( “@#{attrib}”, (list.include?( val ) ? val : default )) end end end class Dummy extend Validator attr\_accessor :name, :type validate\_is\_member\_of :type, [:fruit, :veggie, :dairy], :fruit def initialize( name, type ) @name = name @type = type end end good = Dummy.new(‘good’, :veggie ) # =\> … @type=:veggie good.type = :ice\_cream # =\> … @type=:fruit. i know i overwrote a valid value. it’s ok. bad = Dummy.new(‘bad’, :chocolate ) # =\> … @type=:chocolate; I want :fruit here. ---
**Author:** ![mikec](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/977dab/32.png) [@mikec](https://www.ruby-forum.com/u/mikec)\ **Post date:** [February 11, 2009, 5:37am UTC](https://www.ruby-forum.com/t/redefining-my-attr/160029/2 "2009-02-11T05:37:02Z")
Mike C. wrote: > I want to write a generic validator that ensures an attr value is in a > list before assigning, and if not, assigns a default. It works for > self.attrib=‘value’, and Dummy.new.attrib=‘value’, but not > @attrib=‘value’ within the class. The connection between a method foo=(val) and some instance variable @foo is essentially coincidental. When you use a class method such as attr\_accessors it dynamically creates code that defines accessor methods, and in those methods uses an instance variable with a matching name. But there’s no reason the instance variable in those accessors methods could not called something else. It just makes a certain sense to use an obvious naming convention; it’s so much easier to track what your code is doing. And it creates (for better or worse) the illusion of public “properties” (as, for example, what Java has). There’s nothing to stop other methods from manipulating those instance variables; they have no intrinsic connection to any particular methods, no matter what they are named. # Runs, but pedantic def foo=(x); @bar=x;end def baz; @bar; end def foo; 47; end When you do @attrib = 47 you are working directly with the instance variable, not with a method that might just happen to have a matching name. – James B. [www.happycamperstudios.com](http://www.happycamperstudios.com) - Wicked Cool Coding [www.jamesbritt.com](http://www.jamesbritt.com) - Playing with Better Toys [www.ruby-doc.org](http://www.ruby-doc.org) - Ruby Help & Documentation [www.rubystuff.com](http://www.rubystuff.com) - The Ruby Store for Ruby Stuff ---
**Author:** ![mikec](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/m/977dab/32.png) [@mikec](https://www.ruby-forum.com/u/mikec)\ **Post date:** [February 11, 2009, 4:42pm UTC](https://www.ruby-forum.com/t/redefining-my-attr/160029/3 "2009-02-11T16:42:20Z")
On Tue, Feb 10, 2009 at 11:01 PM, Mike C. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid)wrote: > I want to write a generic validator that ensures an attr value is in a > list before assigning, and if not, assigns a default. It works for > self.attrib=‘value’, and Dummy.new.attrib=‘value’, but not > @attrib=‘value’ within the class. Any way to do that? No, not in general. @attrib = value involves a primitive assignment operator which isn’t a method invocation and therefore can’t be overriden. The best you can do is to impose a discipline and avoid direct iv assignment to the varlable(s) you want to validate within the methods of that class. Here’s the code: If I may have to temerity to offer some critique: > module Validator > > def validate\_is\_member\_of( attrib, list, default ) > original\_method = instance\_method( “#{attrib}=”.to\_sym ) > define\_method( “#{attrib}=”.to\_sym ) do |val| > instance\_variable\_set( “@#{attrib}”, (list.include?( val ) ? val : > default )) > end > end Not sure why you are doing with the original\_method variable since it’s never used. In the code below the getter method for the type attribute generated by attr\_accessor :name :type is just discarded. > @name = name > @type = type > end > end > > good = Dummy.new(‘good’, :veggie ) # =\> … @type=:veggie > good.type = :ice\_cream # =\> … > @type=:fruit. i know i overwrote a valid value. it’s ok. @type here is not an instance varlable of an instance of Dummy, it’s an instance variable of the top-level object, so this line is moot. > bad = Dummy.new(‘bad’, :chocolate ) # =\> … @type=:chocolate; I > want :fruit here. Now, if I were to approach this I might change the dsl a bit and have the class method take on the job of attr\_accessor and generate the getter and setter methods, For clarity I’d change the name validate\_is\_a\_member\_of. Here’s another swing at this: module Validator def validated\_attr( attrib, list, default=nil) attr\_reader attrib define\_method( “#{attrib}=”.to\_sym ) do |val| instance\_variable\_set( “@#{attrib}”, (list.include?( val ) ? val :default )) end end end class Dummy extend Validator attr\_accessor :name validated\_attr :type, [:fruit, :veggie, :dairy], :fruit def initialize( name, type ) @name = name # here is an example of the discipline I mentioned, since initialize is an instance method, # it should use the setter method. self.type = type end end good = Dummy.new(‘good’, :veggie ) # =\> #\ good.type # =\> :veggie good.type = :ice\_cream good.type # =\> :fruit bad = Dummy.new(‘bad’, :chocolate ) # =\> #\ “I want :fruit here:” # =\> “I want :fruit here:” bad.type # =\> :fruit “No chocolate fo you kid!” # =\> “No chocolate fo you kid!” Note that this still doesn’t prevent someone from sending :instance\_variable\_set and bypassing this, using #send or # **send** to get around the privacy… Here’s a slightly more complicated version which closes that hole, but IMHO this is really going a bridge too far. module Validator module ClassMethods def validated\_setters @validated\_setters ||= {} end ``` def validated_attr( attrib, list, default=nil) attr_reader attrib module_eval( "def #{attrib}=val;@#{attrib} = ``` #{list.inspect}.include?(val) ? val : #{default.inspect};end") self.validated\_setters[“@#{attrib}”.to\_sym] = :“#{attrib}=” end end def self.included(other\_mod) other\_mod.extend ClassMethods end def send(symbol, \*args, &block) if symbol == :instance\_variable\_set && setter = self.class.validated\_setters[args.first.to\_sym] send(setter, args[1], &block) else super end end alias : **send** :send end class Dummy include Validator attr\_accessor :name validated\_attr :type, [:fruit, :veggie, :dairy], :fruit def initialize( name, type ) @name = name # here is an example of the discipline I mentioned self.type = type end end good = Dummy.new(‘good’, :veggie ) # =\> #\ good.type # =\> :veggie good.type = :ice\_cream good.type # =\> :fruit bad = Dummy.new(‘bad’, :chocolate ) # =\> #\ “I want :fruit here:” # =\> “I want :fruit here:” bad.type # =\> :fruit “No chocolate fo you kid!” # =\> “No chocolate fo you kid!” good.send(:instance\_variable\_set, :@type, :dairy) good.type # =\> :dairy good.send(:instance\_variable\_set, :@type, :arsenic) good.type # =\> :fruit good. **send** (:instance\_variable\_set, :@type, :arsenic) good.type # =\> :fruit – Rick DeNatale Blog: [http://talklikeaduck.denhaven2.com/](http://talklikeaduck.denhaven2.com/) Twitter: [http://twitter.com/RickDeNatale](http://twitter.com/RickDeNatale)