[PATCH] AR migration generator includes model's modules in table name.

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

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.

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

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 ...

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

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?

#4032 Module specific table_name_prefix 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 Module specific table_name_prefix or class_attribute :table_name_prefix over the other, but the module-specific table_name_prefix is already in master.

Here's the class_attribute patch.

Awesome! I talked with Jeremy Kemper and he proposed us to follow the pattern in #4032 Module specific table_name_prefix. 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!

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.

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.

Test and docs added to USAGE

Forgot to squash commits - updated patch as single commit

Third time lucky - removed check for existing file.

(from [bab1f910c7399fcfe9f031a1ce3a1f36bf5fd277]) table_name_prefix and table_name_suffix are class_attributes instead of cattr_accessors. [#4230 [PATCH] AR migration generator includes model's modules in table name. ]

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

Thanks guys, applied! Can someone please tell me if #2965 Advanced / foxy fixture features doesn't work well with models in modules is still an issue in Rails master?

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

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

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?