In Ruby on Rails, if you declare accepts_nested_attributes_for in your model, autosave is set to true for the child association. Is this necessary?
According to my understanding, Rails will already validate all new and changed children without declaring autosave: true. It seems that would cover all cases where you are accepting nested attributes for the child association. However, with autosave: true the child now gets validated every time the parent is saved, even if the child is unchanged.
This can have major unintended consequences, especially if, for instance, you modify the child model in such a way that a large amount of your records are invalid.
Try setting validate: false on the association. You can see from http://api.rubyonrails.org/classes/ActiveRecord/AutosaveAssociation.html that activating autosave (through accepts_nested_attributes_for in our case) always validates the record unless you use validate: false
Good question! I just ran into this surprise myself.
I think Rails will only save/validate new child records when autosave is nil (the default).
In save_belongs_to_association(reflection), it looks like it only saves the associated record if it is a new record or autosave is enabled.
saved = record.save(:validate => !autosave) if record.new_record? || (autosave && record.changed_for_autosave?)
I've always found the autosave option to be slightly confusing/inconsistent... But now we're probably stuck with it for backwards compatibility reasons...
Related
I want to write a method that creates a bunch of almost-duplicate records, just with one or two parameters changed. I'll make a form to control those parameters, I'm just wondering about how best to write the method, and where do keep it.
Presently in my document.rb I've written this:
def self.publish(brand, components, template)
brand.users.each do |user|
Document.create(:component_ids => components, :message => 'Message.', :template_id => template.id, :user_id => user.id)
end
end
It doesn't feel right though. Is there a better way to do this?
This code is fine if your security model allows all these fields to be bulk assignable by mention in attr_accessible in the model. If it doesn't then you're better off using the block form of create. Also, if Document, Template and User are ActiveRecord instances, you should let Rails manage the details of ids.
def self.publish(brand, components, template)
brand.users.each do |user|
Document.create do |doc|
doc.component_ids = components,
doc.message 'Message.',
doc.template = template,
doc.user = user
end
end
end
One final note is that component_ids must be serialized to store a list. This is probably a flaw in your model design. The better way is (probably) to specify Component belongs_to User and also User has_many Components. I.e. Component contains a foreign key to User. If it's necessary for a Component to belong also to many users, then you'll need either has_and_belongs_to_many or has_many ... through. The Rails guide on relations describes all this in more detail.
With the right relations set up, the code will become:
def self.publish(brand, components, template)
brand.users.each do |user|
Document.create do |doc|
doc.components = components, # Components is now a list of active records.
doc.message 'Message.',
doc.template = template,
doc.user = user
end
end
end
The resulting SQL will get all the foreign keys and (if necessary) relation tables filled in correctly.
I just inherited a project. That project has upgraded to Rails3, and is lacking an application_model. I'm trying to implement it. I've extended all other models from it. The problem is that for some reason, when trying to determine the table name, ActiveRecord is using ApplicationModel, rather than the child model so I'm getting:
Table "your_db.application_models" doesn't exist.
Any thoughts? Why would it be pulling from the app model? Any idea on how to fix?
Thanks!
Apparently, here's the issue (according to what I found on this page: rails - root model or application model):
ActiveRecord normally detects when you subclass an already subclassed ActiveRecord::Base and uses this to turn STI (single table inheritance) on.
ApplicationModel will be an actual model that is expected to have a table in your database. This can lead to problems down the line.
To fix these two problems, you have to set abstract_class to true for ActiveRecord to properly function.
class ApplicationModel < ActiveRecord::Base
self.abstract_class = true
end
In contrast to an abstract ActionController, abstract_class must set to true which means the developer must know they cannot remove this line from ApplicationModel. With ApplicationController you can do pretty much whatever you want to it.
How can I validate uniqueness of an attribute with a custom or virtual scope? I thought of using a virtual attribute, but it keeps trying to query audit_year in the database. I would rather not create another database column just for the purpose of this uniqueness constraint.
Each location can only have one audit scheduled per year, so I need to extract the year from the scheduled attribute and validate uniqueness over that scope.
class Audit
attr_accessible :location_name, :scheduled_date, :completion_date ...
validates :location_name, :presence => true, :uniqueness => { :scope => :audit_year }
...
def audit_year
scheduled_date.year
end
end
I may not even be on the correct path with my virtual attribute attempts. What would be the "right" way to do this in rails?
I know this is a bit late, but I figured I'd pitch in. I'm doing this from memory so you may need to screw with this a bit.
My first thought is in your audit_year method, you could query the database for all years. Something like:
def audit_year
!self.class.select('selected_date').map{ |a| a.selected_date.year }.include?(selected_date.year)
# or using Rails 3.2 and Ruby 1.9.3:
# !self.class.pluck('created_at').map(&:year).include?(selected_date.year)
end
My assumption of the unique method is if it returns false, it will fail validation. This code selects just that one column from your class (I used self.class instead of Audit so it's more reusable), then maps out the years to an array. If it's included (true), return the opposite (!). You could probably optimize the query with uniq, but it depends on how large the table is whether it's necessary or not.
The other option would be to roll your own validator for the attribute, which really wouldn't be that difficult. The only difference is you'd add a line that conditionally checks for selected_date.present? in addition to the above. A great resource is the Rails Guides for callbacks and errors if you don't know how: http://guides.rubyonrails.org/active_record_validations_callbacks.html
Hope that helps.
I need to apply validation to a Model so that 2 integer values in the record, minimum and maximum, form an inclusive range (ex. 2 and 3 are ok, but 4 and 1 are not). From what I understand, since I need to validate 2 values against each other in the same record, I have to use ActiveModel::Validator (and not ActiveModel::EachValidator). So, I try the following:
Model (app/models/position.rb):
class Position < ActiveRecord::Base
validates_with InclusiveRangeValidator
end
Validator (app/lib/validators/inclusive_range_validator.rb):
class InclusiveRangeValidator < ActiveModel::Validator
def validate(record)
record.errors[:base] << "#{record.minimum} and #{record.maximum} do not form a valid inclusive range." unless record.minimum < record.maximum
end
end
I've read that Rails 3.0.5 doesn't automatically load the lib directory anymore, so I added this line in config/application.rb:
config.autoload_path += %W({config.root}/lib)
And then I reset the rails server so it'll take the change to config/application.rb. I run my unit tests, which tries to exercise the model to prove this validation works. I get this error:
uninitialized constant: Position::InclusiveRangeValidator (NameError)
What I think is happening is that Rails is not recognizing/finding my custom validator class, and so it assumes InclusiveRangeValidator is a constant in the class that it's referenced in. But, I thought the change I made to config/application.rb would put my validator in the load path so that it would be available.
I've gone through the other posts on StackOverflow and didn't come up with a solution, and I've read the API docs on validators, to no avail. I've got to be doing something simple and stupid, but I can't see what the issue is. Any help?
EDIT:
After more searching, I discovered that I don't need a custom validator at all, as I can accomplish the same goal with this:
validates :minimum, :numericality => {:greater_than_or_equal_to => 0 }
validates :maximum, :numericality => {:greater_than => :minimum }
However, the question still remains as to why Rails can't locate the custom validation class.
Once, I changed the line in application.rb to:
config.autoload_paths += %W[#{config.root}/lib/validators/]
Rails was able to find the right path to load my custom validator. I made the mistake of assuming Rails would automatically recurse the directory structure, this is evidently not the case.
Apologies for the long title, but this is bothering me. I'm new to Rails, so this is my first project. Rails 3.0.3.
In my model, a User may or may not have read many Entries; this is tracked in a model called ReadEntries. This many-to-one relationship is properly defined in the code, I think.
User.rb:
has_many :read_entries
Entry.rb:
has_many :read_entries
ReadEntry.rb:
belongs_to :entry
belongs_to :user
This table has to be populated at some point. If I try to do this:
user.read_entries.find_or_create_by_entry_id(entry.id, :read => false)
I get the error Unknown key(s): read. Leave out trying to set :read, and it works.
However, if I create the same row with this, it works:
ReadEntry.find_or_create_by_entry_id_and_user_id(entry.id, user.id, :read => false)
Logically, these methods should be identical, right? Thanks.
I've also had weird experiences with find_or_create. I would love it if it worked, but it seems inconsistent.
I'm currently having the same issue as you, and I think it may be due to calling find_or_create on an association as opposed to the model directly. Here's my example:
permission_assignments.find_or_create_by_role_id(:role_id => role_id, :is_allowed => false)
This works to create the assignment, except the "is_allowed" field gets set to it's default of "true". This code works for me (in the Permission model, hence the self reference)
PermissionAssignment.find_or_create_by_permission_id_and_role_id(:permission_id => self.id, :role_id => role_id, :is_allowed => false)
It's more verbose, unfortunately, but it works. The only problem that I still notice is that the object that is returned has no id assigned (the record does get created in the database, however, but if I wanted to update any more attributes I wouldn't be able to without the id). Don't know if that's a separate issue or not.
Rails 3.0.4 here with Postgres 8.4
You cannot pass in other fields like that as Rails will assume they are options for the find. Instead, you will need to make your method call longer:
user.read_entries.find_or_create_by_entry_id_and_read(entry.id, false)
Or alternatively use a shorter, custom syntax for that.
For your final example, my thoughts are that Rails will take the second argument and use that as options. Other than that, I am not sure.