Rails Ruby on Rails → 3.0.2

Comments and changes to this ticket

• You flagged this item as spam.
  

  phs March 19th, 2010 @ 05:14 AM

  #4032 looks like another issue stomping around in the same neighborhood.

• You flagged this item as spam.
  

  phs March 19th, 2010 @ 07:55 AM

  • Tag changed from “activerecord module migration generator” to “activerecord, generator, migration, module, namespace”

  To choose the new table's name, the migration generator template in AR at lib/generators/active_record/model/templates/migration.rb uses the (gasp) table_name method.

  table_name is defined over in railties:

  # lib/rails/generators/named_base.rb
  def table_name
    @table_name ||= begin
      base = pluralize_table_names? ? plural_name : singular_name
      (class_path + [base]).join('_')
    end
  end
  

  file_name, in the same file, looks like it would work.

  However, the other generator templates in AR also use table_name where one might expect. Given that, and the fact the method is named table_name, it seems more natural to override table_name in ActiveRecord::Generators::Base.

• You flagged this item as spam.
  

  phs March 20th, 2010 @ 07:56 AM

  • Tag changed from “activerecord, generator, migration, module, namespace” to “activerecord, generator, migration, module, namespace, patch”

  This patch effectively renames #table_name to #table_path, and provides a new #table_name that omits any leading namespace components.

  Most generators that use #table_name have been left alone, as they seem to work reasonably in either case. Migrations has been updated to keep the namespace components in their filenames (however, the names of the tables they manipulate have dropped their prefixes.) This keeps their tests happy, which seem to make a point of having the namespaced filenames in various situations.

  • break-out-table-path.diff 5.3 KB

• You flagged this item as spam.
  

  phs March 20th, 2010 @ 07:57 AM

  • Title changed from “AR migration generator includes model's modules in table name. ” to “[PATCH] AR migration generator includes model's modules in table name. ”

• You flagged this item as spam.
  

  phs March 20th, 2010 @ 07:30 PM

  Here's an alternate patch that makes the change to #table_name but doesn't bother introducing #table_path. Generated migrations from the model generator now make un-namespaced files.

  • remove-namespace-components-from-table_name.diff 5.3 KB

• You flagged this item as spam.
  

  phs March 20th, 2010 @ 11:56 PM

  Here's a third option. This one leaves #table_name alone, and slips a self.table_name_prefix = "my_module_" when generating namespaced models.

  • add_table_name_prefix_when_generating_namespaced_models.diff 2.2 KB

• You flagged this item as spam.
  

  Francesc Esplugas April 8th, 2010 @ 07:51 AM

  Same problem here. At this moment I'm fixing this problem by using "set_table_name". Using the table name prefix doesn't seem a good option:

  class Delayed::Job < ActiveRecord::Base
    self.table_name_prefix = 'delayed_'
  end
  
  rails_3b2 → rails console
  Loading development environment (Rails 3.0.0.beta2)
  >> Delayed::Job.table_name
  => "delayed_jobs"
  >> Post.table_name
  => "delayed_posts"
  
  rails_3b2 → rails console
  Loading development environment (Rails 3.0.0.beta2)
  >> Post.table_name
  => "posts"
  >> Delayed::Job.table_name
  => "delayed_jobs"
  >> Post.table_name
  => "posts"
  

  I would go for a set_table_name option ...

• You flagged this item as spam.
  

  Francesc Esplugas April 8th, 2010 @ 07:56 AM

  Another observation is that the usage of self.table_name_prefix on a model would overwrite the system wide table_name_prefix option.

• 

  José Valim April 10th, 2010 @ 12:39 PM

  • Milestone cleared.
  • Assigned user set to “José Valim”
• 

  José Valim April 10th, 2010 @ 12:41 PM

  Can someone please provide a patch that changes self.table_name_prefix to use class_attribute instead of cattr_accessor allowing it to work with inheritance?

• You flagged this item as spam.
  

  phs April 11th, 2010 @ 12:36 AM

  #4032 might be another way out. If we could find a canonical place to put the module-specific table_name_prefix declaration, then the model generator could ensure it is present.

  I don't have a justifiable opinion for preferring either #4032 or class_attribute :table_name_prefix over the other, but the module-specific table_name_prefix is already in master.

• You flagged this item as spam.
  

  phs April 11th, 2010 @ 03:33 AM

  Here's the class_attribute patch.

  • make_table_name_prefix_a_class_attribute_instead_of_cattr_accessor.diff 6.1 KB
• 

  José Valim April 11th, 2010 @ 06:40 AM

  Awesome! I talked with Jeremy Kemper and he proposed us to follow the pattern in #4032. That said, when you run:

  rails g model my_module/post
  

  It should create a file in app/models/my_module.rb with the following contents:

  module MyModule
    def self.table_name_prefix
      'my_module_'
    end
  end
  

  Can you provide a patch for this? I will apply both this new patch and the class_attribute one once they are done. :)

  Thanks!

• 

  Andrew White April 11th, 2010 @ 09:33 AM

  José, should the patch try and load the parent constant and check for table_name_prefix or should it just check for the presence of the 'app/models/my_module.rb' file and don't create it if it already exists.

  The attached patch is a naive implementation that doesn't have any kind of intelligence. Is this the kind of thing you're looking for or a more sophisticated patch that checks for the parent constant first and then adjusts the migration name and the table name appropriately.

  The patch is just a proof of concept - if it's acceptable I'll add another later with docs and tests.

  • 0001-generate-module-with-table-name-prefix.diff 1.9 KB
• 

  José Valim April 11th, 2010 @ 10:24 AM

  Andrew, the patch looks great. We just need tests and there is no need to check if a file is already there, since Rails does it for us and delegates the decision to the user.

• 

  Andrew White April 11th, 2010 @ 11:22 AM

  Test and docs added to USAGE

  • 0002-generate-module-with-table-name-prefix.diff 4.4 KB
• 

  Andrew White April 11th, 2010 @ 11:28 AM

  Forgot to squash commits - updated patch as single commit

  • 0003-generate-module-with-table-name-prefix.diff 4.2 KB
• 

  Andrew White April 11th, 2010 @ 01:03 PM

  Third time lucky - removed check for existing file.

  • 0004-generate-module-with-table-name-prefix.diff 3.9 KB
• 

  Repository April 12th, 2010 @ 10:10 AM

  • State changed from “new” to “resolved”

  (from [788d9238931bb36ed2af78d34c84562dc3a4848c]) Generate module file for namespaced models [#4230 state:resolved]

  Signed-off-by: José Valim jose.valim@gmail.com
  http://github.com/rails/rails/commit/788d9238931bb36ed2af78d34c8456...

• 

  Repository April 12th, 2010 @ 10:10 AM

  (from [bab1f910c7399fcfe9f031a1ce3a1f36bf5fd277]) table_name_prefix and table_name_suffix are class_attributes instead of cattr_accessors. [#4230]

  Signed-off-by: José Valim jose.valim@gmail.com
  http://github.com/rails/rails/commit/bab1f910c7399fcfe9f031a1ce3a1f...

• 

  José Valim April 12th, 2010 @ 10:11 AM

  Thanks guys, applied! Can someone please tell me if #2965 is still an issue in Rails master?

• 

  Andrew White April 12th, 2010 @ 02:34 PM

  Yes, it's still an issue. Is it something you'd like me to investigate?

• 

  José Valim April 12th, 2010 @ 02:35 PM

  @Andrew, if you can, it would be awesome!

• 

  Andrew White April 12th, 2010 @ 02:51 PM

  I'll get on it. One immediate consequence of getting it working is that the fixture file path will have to switch from matching the table name to matching the underscored class name - is this okay?

• 

  Jeremy Kemper October 15th, 2010 @ 11:01 PM

  • Milestone set to “3.0.2”
  • Importance changed from “” to “High”

  [bulk edit]