#1151 ✓not-applicable
Carl Porth

Problem with associations when types don't match

Reported by Carl Porth | December 12th, 2009 @ 04:48 PM

Association doesn't work when foreign key type doesn't match primary key type. (I think a typecasting issue.)

This feature is especially desirable when using cross repository associations.


Comments and changes to this ticket

  • Carl Porth

    Carl Porth December 18th, 2009 @ 01:27 PM

    To clarify further:

    Let's say I'm using a 3rd party library which provides an adapter and some default models for a webservice for listing products (like google products). The webservice calls for a product_id with type String. On my db, I have on my own product model with product_id defined as an Integer. I would like to associate the two.

    It might look something like this:

      require 'rubygems'
      gem 'dm-core', '0.10.2'
      gem 'do_sqlite3', '0.10.0'
      require 'dm-core'
      module Webservice
        class Product
          include DataMapper::Resource
          def self.default_repository_name
          property :id, String, :key => true
          property :name, String
          property :product_id, String
          belongs_to :product, 'Store::Product', :repository => :default
      module Store
        class Product
          include DataMapper::Resource
          property :id, Serial
          property :name, String
          has 1, :webservice_product, 'Webservice::Product', :repository => :webservice
      DataMapper.setup(:default, 'sqlite3::memory:')
      DataMapper.setup(:webservice, 'sqlite3::memory:')
      Webservice::Product.create(:id => 'x', :name => 'foo', :product => Store::Product.create(:name => 'foo'))
      # in 0.10.1:
      Webservice::Product.first.product # => #<Store::Product @id=1 @name="foo">
      # in 0.10.2:
      Webservice::Product.first.product # => nil

    In 0.10.1, this type of association worked.
    In 0.10.2, this no longer works.

    I'm not sure it is a regression because the behavior is undefined.

    Is this a weird edge case not worth handling?

  • Dan Kubb (dkubb)

    Dan Kubb (dkubb) February 5th, 2010 @ 02:13 AM

    • State changed from “new” to “not-applicable”

    Actually, in the future it will only be possible to join on properties that have the same type and the same constraints (like :min/:max for Integer, :length for String, etc). I recently reverted a commit that did implicit coercion between types in a join.

    I may add a way to explicitly coerce the types to another before joining them, but that will likely be after DM 1.0, and after we rework the query/adapter API to be built off veritas or something similar.

Please Sign in or create a free account to add a new ticket.

With your very own profile, you can contribute to projects, track your activity, watch tickets, receive and update tickets through your email and much more.

New-ticket Create new ticket

Create your profile

Help contribute to this project by taking a few moments to create your personal profile. Create your profile »

People watching this ticket