Adding an associated Table to an Existing Table (Book > Revision)

Question

Hello, newbie here... I have the following models:

class Instance < ActiveRecord::Base 
 has_many :users 
 has_many :books 
end 
class User < ActiveRecord::Base 
 belongs_to :instance 
end 
class Book < ActiveRecord::Base 
 belongs_to :instance 
end 

I now want to add a BookRevision table, that has the following columns (id, user_id, version # (auto increment), diff (old book copy), timestamps) Logic:

• When a book is Created, a record is added to the BookRevision table, so we know who created the book in the first place
• When a book is Updated, a record is added with the user_id (could be a different user), and a new version #, and the old book text, to serve as an archive.

Given that I have the Instance, User, Book table implement in my rails add, are these the correct steps to make the above come to life? - Add a migration for the BookRevision table.... rails generate migration AddTableBookRevision user_id:integer version:integer diff:text - Then update the models as follows:

class Instance < ActiveRecord::Base 
 has_many :users 
 has_many :books 
end 
class User < ActiveRecord::Base 
 belongs_to :instance 
end 
class Book < ActiveRecord::Base 
 belongs_to :instance 
 has_many :BookRevisions 
end 
class BookRevision  < ActiveRecord::Base 
 belongs_to :Book 
end 

Then in my controller, when adding a new book? Right now I have:

@book = Book.create(params[:book].merge(:instance_id => 
current_user.instance_id)) 

How do I update that to account for the BookRevision association? Thanks for helping me out!

Answer 1

You might want to check out something like acts_as_versioned instead of rolling your own. This works in the fashion you've described here, where modifications are saved into a separate but related table.

Keep in mind you will have to apply migrations to your Book and BookRevision table in parallel from that point forward. They must be schema compatible for revisioning to work.

I have built a version tracking system of this sort that used serialized models to avoid having to maintain migrations, as the intent was to preserve the exact state of the model regardless of future modifications via migrations. This has the disadvantage of not being able to roll back to an arbitrary older version because there may be a schema mis-match.