When you use Ruby on Rails with ActiveRecord, you can get used to having separate transaction on each request. This is valid also when using Trailblazer (when inside of a request scope), however Trailblazer on its own does not provide such functionality. It means that when you're using it from the console and/or when you process stuff inside background workers, you no longer have an active transaction wrapping around an operation.
This behavior is good most of the time. Since background tasks can run for a long period of time, there might be a risk of unintentional locking a big part of your database. However, sometimes you just need to have transactions.
In order to provide this feature for each operation, we will use a concern that will include that logic. We will also make it configurable, so if we inherit from a given operation, we will still have an option to disable/enable transaction support based on the operation requirements.
The code itself is pretty simple - it will just wraps around a #run method of the operation class with a transaction (as long as transaction support is enabled). Note, that by default transactional flag is set to false.
module Concerns module Transactional extend ActiveSupport::Concern included do class_attribute :transactional self.transactional = false end def run if self.class.transactional self.class.transaction do super end else super end end class_methods do def transaction ActiveRecord::Base.transaction do return yield end end end end end
In order to use it, just include it into your operation:
class ApplicationOperation < Trailblazer::Operation # Including on its own won't turn transactions on include Concerns::Transactional end class DataOperation < ApplicationOperation # This operation will have a single transaction wrapping it around self.transactional = true end
February 7, 2018 — 11:05
It should be `include Concerns::Transactional` right?