This message was imported from the Ruby/Rails Modularity Slack server. Find more info in the import thread.
Message originally sent by slack user U71V6G2PDJS
How do you think about dependencies at the data level? For instance ActiveRecord ID’s. Example: A user has some notes, so note has a user_id. They are in different packages. Would you have a dependency from the note package to the user package?
How would you deal with cascading deletes in this case? Make note an observer of user deletions somehow?
How would you deal with cascading deletes in this case?
Correct me if i’m wrong, but if you’re using packs-rails, aren’t the ActiveRecord objects all loaded into the same namespace? Any cascade deletes can be setup with :dependent.
I’d love to know more about this. What we’ve done at Bloom & Wild is add a “Reference” ApplicationRecord module.
Let’s say that we have user_id spread around the system.
We create a ‘User reference’ object:
class UserReference < ApplicationRecord
include ReferenceObject::ActiveRecord
end
And change things that refer to it to use it:
class SomeClass < ApplicationRecord
belongs_to :user, class_name: "UserReference"
end
The ReferenceObject::ActiveRecord is a bit complex, but the core of it is this:
module ReferenceObject
module ActiveRecord
# Rails supports "ignored_columns", but doesn't have the inverse - something like "allowed_columns"
# would be very useful here.
def ignored_columns
@ignored_columns ||= begin
# Generate a list of ignored_columns that include all of the actual table
# columns, excluding the ones in the classattr `allowed_columns`
cached_names = connection.schema_cache.columns_hash(table_name).map { |name, _| name.to_s }
(cached_names - ["id"]).freeze
end
end
end
end
The reason I’ve not done this is because by using a ReferenceObject I can find all places that use that Reference object nice and easily. As opposed to an integer that happens to be called ‘user_id’, say
Hey all