- Ruby
- Rails 7
- ESbuild
- PostgreSQL
- Redis
- Tailwind
- Hotwired
- Railway ( hosting the app )
- Sidekiq
- Digitalocean spaces
Please head over here for more engineering related information.
In order to get the local environment up and running on your machine, you would require the following to be installed :
- Ruby 3.1.1. It is recommended to use a version manager like rbenv or rvm in order to manage your rubies.
- PostgreSQL
- Redis
- Vips -
brew install vips
If installing ruby on M1 architecture, you may have some issues when compiling the Ruby binaries. Try installing it through rbenv using this command:
RUBY_CFLAGS="-Wno-error=implicit-function-declaration" arch -x86_64 rbenv install 3.0.2
We manage api keys and secrets via Rails credentails. This means credentials are all committed to repo via a credentials file. Our setup is the following -
- Global credentails file - config/credentials.yml.emc. You need a
config/master.key
in order to decrypt the secrets run your development server.
For adding a secret to development and test
$ EDITOR="vim" bin/rails credentials:edit
For Staging and production
$ EDITOR="vim" bin/rails credentials:edit --environment staging
To view the existing credentials you could do something like -
$ bin/rails credentials:show
$ bin/rails credentials:show --environment staging
Once the repo is cloned, you need to run the following commands to install all the dependencies and create the database.
$ bundle install
$ yarn install
$ rails db:create db:migrate
$ rails db:seed
The easiest way to start the server is to install the foreman gem and run the server using your procfile.
$ gem install foreman
$ foreman start -f Procfile.dev
Now navigate to localhost:3000 to see the home page of the app. Sidekiq is available at localhost:3000/admin/sidekiq
We use rspec for testing. In order to run your test suite locally simply run from your app directory -
$ bundle exec rspec
For linting our ruby code, we use rubocop. It is recommended to run an autocorrect to keep your code clean and conform with the rubocop rules. You can do so by -
$ bundle exec rubocop -A
We use Honeybadger to log production errors and manage uptime.
We use overcommit to run our pre-commit hooks. Install this by -
$ gem install overcommit
# Inside project's root directory
$ overcommit --install
You're all set 🙌 !!
Mutations
userLogin
mutation is used for logging into the a user's account. In order to make subsequent requests, you would need accessToken
, client
and uid
. So make sure you extract that for the next one.
mutation UserLogin {
userLogin(email: "lyman@breitenberg-jacobson.biz", password: "password") {
authenticatable {
confirmationSentAt
confirmedAt
createdAt
currentSignInAt
currentSignInIp
email
firstName
id
lastName
lastSignInAt
lastSignInIp
purchasesCount
rememberCreatedAt
signInCount
unconfirmedEmail
updatedAt
}
credentials {
accessToken
client
expiry
tokenType
uid
}
}
}
Logout
mutation UserLogout {
userLogout {
authenticatable {
confirmationSentAt
confirmedAt
createdAt
currentSignInAt
currentSignInIp
email
firstName
id
lastName
lastSignInAt
lastSignInIp
purchasesCount
rememberCreatedAt
signInCount
unconfirmedEmail
updatedAt
}
}
}
Queries
For all the queries, you would need to pass the headers you've received from the userLogin
mutation, somewhat like below :
const client = new ApolloClient({
cache,
link: new HttpLink({
uri: 'http://localhost:3000/graphql',
headers: {
"access-token": asyncStorage.getItem('access-token'),
"client": asyncStorage.getItem('client'),
"uid": asyncStorage.getItem('uid')
},
}),
});
query Purchases {
purchases(userId: "65") {
id
product {
ageRating
createdAt
description
fileAttachment
id
issueCover
name
pageCount
price
releaseDate
slug
updatedAt
collaborations {
creatorProfile {
name
formattedBio
skills
avatar
}
}
store {
displayImage
id
name
}
updatedAt
}
}
}