GeorgieMordor posted:

Totally. Request tests should be more than adequate, especially since this is an API app anyway. I was just running through some RSpec refresh and thought I'd give the controller tests a go just for sharpness' sake.

Out of curiosity with controller tests being soft deprecated what's an example of some functionality that would merit a controller test these days?

Tea Bone posted:

I’m trying to set up a relationship from a record to records it isn’t associated with. Imagine the following:

code:

class Reader<ActiveRecord::Base
	belongs_to :collections
	has_many :books, through: :collection
end

class Books<ActiveRecord::Base
	belongs_to :collections
	has_many :readers, through: :collections
end

class Collection<ActiveRecord::Base
	has_many :books
	has_many :readers
end

I want to define an association in reader which will return all the books not in their collection. I could use the following method:

code:

def books_not_in_collection
	Books.where(“id NOT IN (?)”, book_ids)
end

But that causes N+1 queries if it’s run on more than one reader, ideally I want to set it up as an association in reader. Is this possible?

Piano Maniac posted:

Rails gurus, I am in a dire need of your help.
Could somebody break it down Mickey Mouse style.

How the heck can I create a link, something like

code:

<a data-toggle="collapse" href="#XXXX"> Some text </a>

Which once you press on it, it will target a div with id #XXXX (that conveniently matches my file_id)

code:

  <div id="#XXXX">
  </div> 

Which would conveniently insert some sweet ruby-enhanced HTML partial into the div thing,
insert the file_id parameter into the right places (so I can display file_id.pdf file and make it bootstrap collapsible), render it and show it to the user.

I have asked my co-workers but they all say わかんないAJAX使ったら？ and I am like poo poo, I have been writing ruby for like 2 weeks only. What the hell am I supposed to do? I don't get what AJAX and controllers do either...

I have searched for hours now and now I am dying. Please don't ask me to post my controllers or something, I am a mess right now and it's my twelfth hour in the office.

necrotic posted:

Integration tests aren't, which go through the request flow. Testing the controllers directly is indeed deprecated, test the routes.

That is specific to TestUnit, the default test harness in Rails. Dunno if RSpec is changing anything.

ddiddles posted:

This is awesome, thanks for all that info.

Going to read up on requests specs, coming from the front end world, this test environment is already 1000x better to work in.

ddiddles posted:

Another question about rails specs. Switched over to using request specs rather than controller, but I'm repeating myself a lot checking that routes are protected by auth.

code:

require 'rails_helper'

RSpec.describe 'Projects', :type => :request do
  include ApiHelper

  let!(:valid_create_params) do
    { name: 'Test Project' }
  end

  context 'with an unauthorized create request' do
    it 'returns a 401 status' do
      post '/projects', params: valid_create_params
      expect(response).to be_unauthorized
    end
  end

  context 'with an authorized create request' do
    let(:user) { create(:user) }

    it 'returns a project belonging to the authed user' do
      post '/projects', params: valid_create_params, headers: authorize_header(user)
      project = JSON.parse(response.body)
      
      expect(project["name"]).to eq(valid_create_params[:name])
      expect(project["user_id"]).to eq(user.id)
    end
  end

  context 'with an unauthorized get request' do
    it 'returns a 401 status' do
      get '/projects'
      expect(response).to be_unauthorized
    end
  end

  context 'with an authorized get request' do
    let(:user) { create(:user) }

    it 'returns the users projects' do
      user.projects.create!(valid_create_params)
      get '/projects', headers: authorize_header(user)
      user_projects = JSON.parse(response.body)

      expect(user_projects.count).to eq(1)
    end
  end

  context 'with an unauthorized show request' do
    it 'returns a 401 status' do
      get '/projects/1'
      expect(response).to be_unauthorized
    end
  end

  context 'with an authorized show request' do
    let(:user) { create(:user) }
    let(:user2) { create(:user) }
    let(:user_project) { create(:project, valid_create_params.merge(user_id: user.id)) }
    let(:user2_project) { create(:project, valid_create_params.merge(user_id: user2.id)) }
    
    before do
      get "/projects/#{project_id}", headers: authorize_header(user)
    end
    
    context 'with a project_id belonging to the user' do
      let(:project_id) { user_project.id }
      
      it 'returns the project' do
        expect(JSON.parse(response.body)["user_id"]).to eq(user.id)
      end
    end
    
    context 'with a project_id that doesnt belong to the user' do
      let(:project_id) { user2_project.id }

      it 'returns not found' do
        expect(response).to be_not_found
      end
    end
  end

  context 'with an unauthorized delete request' do
    it 'returns 401 status' do
      delete '/projects/1'
      expect(response).to be_unauthorized
    end
  end

  context 'with an authorized delete request' do
    let(:user) { create(:user) }
    let(:user2) { create(:user) }
    let(:user_project) { create(:project, valid_create_params.merge(user_id: user.id)) }
    let(:user2_project) { create(:project, valid_create_params.merge(user_id: user2.id)) }

    before do
      delete "/projects/#{project_id}", headers: authorize_header(user)
    end

    context 'with a project belonging to the user' do
      let(:project_id) { user_project.id }

      it 'should delete authed users project if it exists' do
        expect(response).to be_ok
      end
    end

    context 'with a project belonging to another user' do
      let(:project_id) { user2_project.id }

      it 'returns not found' do
        expect(response).to be_not_found
      end
    end
  end
end

The actual before_action that requires auth on requests is in the application controller, with only a few routes that are going to skip that, such as creating a new user. Should I just have a small application_controller controller spec that checks that before_action is working correctly? Or should I be as verbose as this?

GeorgieMordor posted:

I'm using FactoryBot in my test cases. How do I create a related record that can be referenced multiple times in a single factory? For instance, instead of using association :owner in a :car factory, I'd create the :owner record first and then explicitly define it's properties in the :car factory. Example of what I thought might work:

code:

FactoryBot.define do
	factory :car do
		_owner = create(:owner)
		
		make { "Honda" }
		model { "Civic" }
		year { [1980...2019].sample
		owner_name { _owner.name }
		owner_age { _owner.age }
	end
end

I realized this was a problem because there could be some environment conflicts between :test and :development environments. For instance, foreign key restrictions seem to step on each others toes when I try to boot up a local server.

So, I'm either going about building my factories wrong, OR I have my environments somehow misconfigured?

GeorgieMordor posted:

Also my thoughts, though this is a legacy application I've been brought on board for so I'm at the mercy of the current functionality. At least for now.

For the sake of supporting these convoluted models in tests / factories though, you indicated some ways to do this -- what are they?

Awesome Animals posted:

You could also probably use a let!

code:

let!(:car) { FactoryBot.create(:car) }

That's assuming you've defined that specific FactoryBot.create

https://relishapp.com/rspec/rspec-core/docs/helper-methods/let-and-let

DaTroof posted:

Overriding an existing method is heinous because it can lead to incorrect assumptions and elusive bugs. Adding new methods to an existing class isn't as bad. Given that Rails is already a quilt sewn from monkeypatches, adding a couple String methods seems okay; but my first preference would still be to encapsulate those features in a dedicated class or module, especially if the implementation requires internal state.

GeorgieMordor posted:

I'm working through a Rails update now. Pre-existing Rails 4.2.* app to 5.2.3.

The application runs against Postgres, and is utilizing some json and jsonb fields.

While testing the upgrade we learned that some of these fields are being saved as strings instead of accessible JSON, and then this will break subsequent lookups to said fields when they can't access it as hashed JSON. Further reading found this is a known thing with Rails upgrades but I'm not totally understanding where the disconnect is happening.

Trying to debug, I see that params come through the request looking like this:

code:

"json"=>"{\"widgets\":[{\"Widget3000\":\"WG9240\"}]}"

What I don't understand is where are those escape slashes being added? Post-request by Rails? By the original requester and sent over with the payload?

What's tripping me up is if I try to post the same structure in a debug tool like Paw or Postman, I get a response from Rails that it's unparsable JSON. Yet, if the original requester sends it over in that shape, the payload is saved to the database without any error responses, though it is saved incorrectly as a string.

The Journey Fraternity posted:

I would love to, if it weren't such a behemoth undertaking even to get it to 5.0. This codebase predates every single engineer, and was originally written by PHP devs writing their first Rails app.

duck monster posted:

Unpopular opinion, but yes. Ditto with Node JS. The jobs have been drying up as companies move across to python, and a LOT of pretty important libraries seem abandoned now.

Edgar Allan Pwned posted:

so im working on a coding practice, using an api for a random name and another api for a random joke. im supposed to be able to create an api that generates this new joke on create. it will be tested by using curl.

so i gotten all of the parts working, creating the new joke, gettinn CRUD to work for my api. the only thing is, im trying to get my method (for creating the new joke) to run when i create a new instance of the model. it is not working.

is running methods in the create method common? is there a different way i should be thinking about this?

should i be having the user run the method on create manually? this seems wrong but im not finding a ton on google for running a method for params on create.

Tea Bone posted:

Thanks guys. I ended up flipping the relationship (so that suit belonged to card) as it made more sense for my use case and that ended up solving the issue.

Jaded Burnout posted:

I've been working with Rails for a very long time, and I have a new side project I'd like to do that's very data input heavy, and I absolutely cannot face setting up all the controllers and such. Is there a nicer way to do it than either grinding through it or using the scaffolding generators? Anything nice and dynamic that doesn't immediately fall apart?

I'm just so tired.

Gmaz posted:

I use stimulus for non SPA purposes with my rails projects and I must say I really enjoy it. It's very simple to use, and anyone familiar with JS can pick it up super quickly. Even someone primarily backend oriented like me.

Tea Bone posted:

I'm trying to track the number of live users on my web app.

When a user connects to the app they subscribe to a "live users" action cable channel. I have a "Live Users" index on the admin back end.

Initially, I had it so when the admin went to the live user index the server would get the number of connections on the live users channel from Redis and use that to populate the index. Then any new subscriptions to the live user channel would broadcast to an admin channel and get added to the index. After running it like this for a few hours on my dev server things would start slowing down, and since I don't know too much about Redis or what's going on under the hood I switched to a different method.

My current solution, I have a live user model, on subscription to the live user channel a new record is created (and then removed on unsubscribe). Then the index page simply queries the database for current live users. This is working great on my dev server and has the added benefit of being able to add extra data about the connection (IP, browser language, current page etc) but it feels like there is a lot of database overhead here and I'm not sure how well it will work for production.

Does anyone have any suggestions how I can better handle this?

Slimy Hog posted:

Steely Dad posted:

I'm working on a solo project using Rails with Docker, deploying to Heroku, mostly so I can get my feet wet with containers. My dev setup is basically having rvm and whatnot running natively on my Mac so I can run bundle and yarn to set up dependencies there, but only having a database instance inside my local container setup and only running tests and my dev server that way. I have no idea if this is a good way to do things. Any wisdom to share on developing locally with Docker?

A MIRACLE posted:

is docker sync still a thing