upgrading mysql adapter, fixes current_database bug

better commenting

working on fixing migration test errors

README.md

@@ -1,16 +1,6 @@
 # Apartment
 *Multitenancy for Rails 3*
 
-> IMPORTANT! While I will do everything possible to get Apartment working for Rails 3.1, I haven't, as of yet, had this opportunity.  
-
-> There have been significant changes in the adapters such as prepared statements that might cause some issues.  
-
-> If anyone is successfully using Apartment with 3.1 please let me know, but please ensure that you're testing data integrity properly (ie. that queries are made in the right schema) as prepared_statements
-> could really throw a wrench in that one.
-
-> I know for a fact that Rails 3.1 missed a patch to make postgresql more schema aware, but it seems to have made it into Rails 3.1.1.  So at the very least, use 3.1.1
-
-
 Apartment provides tools to help you deal with multiple databases in your Rails
 application. If you need to have certain data sequestered based on account or company,
 but still allow some data to exist in a common database, Apartment can help.
@@ -129,11 +119,7 @@ from `Apartment.database_names`
 
 ### Delayed::Job
 
-In Apartment's current state, it doesn't seem to queue jobs properly using DJ.  For whatever reason, DJ jobs are created in the current schema, even though the DJ 
-is part of the ignored models.  I have to look into this further, but until then use `Apartment::Delayed::Job.enqueue` to ensure that queues are placed in the public schema
-
-In order to make ActiveRecord models play nice with DJ and Apartment, include `Apartment::Delayed::Requirements` in any model that is being serialized by DJ.  Also ensure
-that a `database` attribute is set on this model *before* it is serialized, to ensure that when it is fetched again, it is done so in the proper Apartment db context.  For example:
+In order to make ActiveRecord models play nice with DJ and Apartment, include `Apartment::Delayed::Requirements` in any model that is being serialized by DJ.  Also ensure that the `database` attribute (provided by Apartment::Delayed::Requirements) is set on this model *before* it is serialized, to ensure that when it is fetched again, it is done so in the proper Apartment db context.  For example:
 
     class SomeModel < ActiveRecord::Base
       include Apartment::Delayed::Requirements
@@ -151,8 +137,10 @@ that a `database` attribute is set on this model *before* it is serialized, to e
       end
     end
 
-## Contributing
+### Development
 
+* The Local setup for development assumes that a root user with no password exists for both mysql and postgresl
+* Rake tasks (see the Rakefile) will help you setup your dbs necessary to run tests
 * Please issue pull requests to the `development` branch.  All development happens here, master is used for releases
 * Ensure that your code is accompanied with tests.  No code will be merged without tests
 

Rakefile

@@ -17,12 +17,6 @@ namespace :spec do
     end
   end
 
-  namespace :unit do
-    RSpec::Core::RakeTask.new(:adapters) do |spec|
-      spec.pattern = "spec/unit/adapters/**/*_spec.rb"
-    end
-  end
-
 end
 
 task :default => :spec

apartment.gemspec

@@ -18,14 +18,14 @@ Gem::Specification.new do |s|
   s.require_paths = ["lib"]
   s.rubygems_version = %q{1.3.7}
 
-  s.add_dependency 'rails', '~> 3.1.1'   # must be >= 3.0.10 due to poor schema support pre 3.0.10, but < 3.1 because it hasn't been fully tested yet
+  s.add_dependency 'rails', '~> 3.1.2'
   s.add_development_dependency 'rake', '~> 0.8.7'
   s.add_development_dependency 'sqlite3'
   s.add_development_dependency 'rspec', '~> 2.6.0'
   s.add_development_dependency 'rspec-rails', '~> 2.6.1'
   s.add_development_dependency 'capybara', '1.0.0'
   s.add_development_dependency 'pg', '~> 0.11.0'
-  s.add_development_dependency 'mysql2', '0.2.7'
+  s.add_development_dependency 'mysql2', '0.3.7'
   s.add_development_dependency "silent-postgres", "~> 0.1.1"
   s.add_development_dependency 'delayed_job', '~> 2.1.4'
 end

lib/apartment.rb

@@ -13,11 +13,7 @@ def configure
 
     # Be careful not to use `return` here so both Proc and lambda can be used without breaking
     def database_names
-      if @database_names.respond_to?(:call)
-        @database_names.call
-      else
-        @database_names
-      end
+      @database_names.respond_to?(:call) ? @database_names.call : @database_names
     end
 
     # Default to none

lib/apartment/adapters/postgresql_adapter.rb

@@ -93,7 +93,6 @@ def reset
       # 
       def connect_to_new(database = nil)
         return reset if database.nil?
-        ActiveRecord::Base.connection.clear_cache!
         ActiveRecord::Base.connection.schema_search_path = database
 
       rescue ActiveRecord::StatementInvalid => e

lib/apartment/railtie.rb

@@ -3,8 +3,10 @@
 module Apartment
   class Railtie < Rails::Railtie
 
-    # Set up our default config options
-    # Do this before the app initializers run so we don't override custom settings
+    # 
+    #   Set up our default config options
+    #   Do this before the app initializers run so we don't override custom settings
+    # 
     config.before_initialize do
       Apartment.configure do |config|
         config.excluded_models = []
@@ -18,16 +20,22 @@ class Railtie < Rails::Railtie
     #   Hook into ActionDispatch::Reloader to ensure Apartment is properly initialized
     #   Note that this doens't entirely work as expected in Development, because this is called before classes are reloaded
     #   See the above middleware/console declarations below to help with this.  Hope to fix that soon.
+    # 
     config.to_prepare do
       Apartment::Database.init
     end
 
+    # 
+    #   Ensure rake tasks are loaded
+    # 
     rake_tasks do
       load 'tasks/apartment.rake'
     end
 
-    # The following initializers are a workaround to the fact that I can't properly hook into the rails reloader
-    # Note this is technically valid for any environment where cache_classes is false, for us, it's just development
+    # 
+    #   The following initializers are a workaround to the fact that I can't properly hook into the rails reloader
+    #   Note this is technically valid for any environment where cache_classes is false, for us, it's just development
+    # 
     if Rails.env.development?
 
       # Apartment::Reloader is middleware to initialize things properly on each request to dev

lib/apartment/reloader.rb

@@ -6,9 +6,8 @@ class Reloader
     #   Necessary due to code reload (annoying).  I couldn't figure out how to properly hook into
     #   the Rails reload process *after* files are reloaded, so I've used this in the meantime.
     # 
-    #   Note that this has one MAJOR caveat.  Doing  `reload!` in the console in development WILL NOT run init again
-    #   Thus, excluded models will not be processed again and will be queried from the current_schema rather than public.  
-    #   I hope to fix this soon
+    #   Also see apartment/console for the re-definition of reload! that re-init's Apartment
+    # 
     def initialize(app)
       @app = app
     end

spec/adapters/postgresql_adapter_spec.rb

@@ -48,7 +48,6 @@
       end
 
       it "should yield to block if passed" do
-        Apartment::Test.migrate   # ensure we have latest schema in the public 
         subject.drop(schema2) # so we don't get errors on creation
 
         @count = 0  # set our variable so its visible in and outside of blocks

spec/dummy/config/database.yml

@@ -7,6 +7,7 @@ test:
   min_messages: WARNING
   pool: 5
   timeout: 5000
+
 development:
   adapter: postgresql
   database: apartment_postgresql_development

spec/dummy/db/migrate/20111202022214_create_table_books.rb

@@ -0,0 +1,13 @@
+class CreateTableBooks < ActiveRecord::Migration
+  def up
+    create_table :books do |t|
+      t.string :name
+      t.integer :pages
+      t.datetime :published
+    end
+  end
+
+  def down
+    drop_table :books
+  end
+end