Clean way to detect change using callbacks

rubyonrails-talk
1

I would like in my 'after_update' callback to detect if an attribute has been changed, and if so change some related objects. Is the best way something like this

class Thing < ActiveRecord::Base

def before_update   @old = Thing.find(self.id) end

end after_update   if @old.val != self.val      do_stuff   end end

Or is there a better way? I wish I could avoid the 'find' call if possible.

Thanks in advance, Don Mc

2

What I’ve done is handle this from the controller, something like this:

class TheController < ActionController def update @thing = Thing.find(params[:id]) rescue nil redirect_to X unless @thing # Where X = named route or routing hash [controller, action, etc]

old = {}
old[:name] = @[thing.name](http://thing.name)
old[:body] = @thing.body
# Process @thing
if @[thing.name](http://thing.name) != old[:name]
  # Do whatcha gotta do...

end
if @thing.body != old[:body]
  # You get the picture...
end

end end

Don’t just make a dup of @thing, though. 'Cause when @thing.update_attributes(params[:thing]) gets called it will change old as well due to Ruby’s shallow copying. You could marshal the whole object but I’ve found it easier to just copy specific attributes. I like the old[:attribute] method but it’d work the same with separate variables for each, like old_name = @ thing.name, etc.

Hope that helps.

RSL

3

I prefer the OP's instinct to handle the dependencies within the model. It's centralized for all users of the model (i.e., controllers), easier to test and better encapsulated. That said, I think the Rails way to deal with this is to override the default setter method for the attribute.

class Thing < ActiveRecord::Base   def val=(v)     write_attribute(:val, v)     do_stuff # Operations that depend on new val   end end

This is also independent of whether the Thing is ever written to the DB, which is generally what you want.

HTH,

Colin

4

Will this work for things like calling a sweeper if the url of a page changes? It’s been a while but I remember having one heck of a time getting ActiveRecord models to communicate with their controllers that way. But A) it’s been a while and B) I only half know what I’m doing know and probably less back then.

Oh yeah, and what’s the OP?

RSL

5

Thanks for responding!

In this particular case, I really don't want to update the related objects unless the object has been saved, since the related objects are in the database also.

Thats a good point about the shallow copying issue.I would like to leave the operation in the model, to handle updates from different sources.

6

I just wanted to thank you for drawing my attention to this methodology. I’ve done some further looking into it and it will indeed do exactly the kind of things I need. What I ended up doing was something like

class Thing < ActiveRecord::Base

attr_reader old_name

def name=(val) @old_name = (val == name) ? nil : name super val end

def vitals_changed? @old_name ? true : false end end

then checking @ thing.vitals_changed? which works marvelously. I seriously cannot thank you enough. This really helps clean up some rather sticky, messy, and misplaced code fragments in my app. Thanks, thanks, thanks.

RSL