State container for your JavaScript apps

Blips exposes a simple, GraphQL-like API for managing your application state, which is contained inside a single store.

The store can only be changed through mutations, you can read from it through queries and can also listen for changes with subscriptions.

I developed Blips because I wanted to use GraphQL with every project, regardless of what the API server looks like. So if my aplication is consuming a simple API (whatever kind), I can still have a store that manages the application state and write queries/mutations that would resolve with making requests to that API.

The following examples assume some familiarity with GraphQL. If you haven't used it before, or you have no idea about what it is, you should visit graphql.org and read about GraphQL in detail.

npm install blips

Creating a store requires type definitions, resolvers and an optional object containing the initial state.

// types.js

export default `
  type Todo {
    id: String!
    label: String!
    completed: Boolean!
  }

  type Query {
    allTodos: [Todo]!
    todo(id: String!): Todo!
  }

  type Mutation {
    createTodo(id: String!, label: String!, completed: Boolean)
  }
`

The types.js file contains 3 definitions:

• a basic Todo object type containing three fields: id, label, completed.
• the root Query type
• the root Mutation type

The ! marks the field as required. See graphql.org to read more about the type system.

// resolvers.js

export default {
  Query: {
    allTodos: (obj, args, { store }) => store.data.todos || []
  },
  Mutation: {
    createTodo: (obj, { id, label, completed = false }, { store: { data: { todos } } }) => {
      const newTodo = { id, label, completed }
      todos = [ ...todos, newTodo ]
      return newTodo
    }
  }
}

The resolvers.js file provides definitions for allTodos query and the createTodo mutation. Each resolver function accepts three arguments:

• obj: The previous object.
• args: The arguments provided to the field in the GraphQL query.
• context: Provides access to important information like the currently logged in user, or the store itself.

See graphql.org to read more about resolvers.

// index.js

import { createStore } from 'blips'
import resolvers from './resolvers'
import typeDefs from './types'

const initialState = {
  todos: {
    '3c4a086e-2151-4b54-acb2-13044ea553c1': {
      id: '3c4a086e-2151-4b54-acb2-13044ea553c1',
      label: 'Buy milk',
      completed: false
    },
  }
}
const schemaDef = { typeDefs, resolvers }
const store = createStore(schemaDef, initialState)

You create a store by calling the createStore function provided by blip and passing it your schema definitions ({ typeDefs, resolvers }) and an optional initial state. The returned store instance has the following API:

• data: getter for your entire state.
• schema: getter for your generated schema.
• store: object containing a couple of opinionated helper methods for working with your data. See helper methods
• query: method for executing queries.
• mutate: method for executing mutations.
• subscribe: method for registering subscriptions.
• fetch: method for sending queries to an actual GraphQL API. coming soon

You define operations in the form of GraphQL queries, mutations or subscriptions and use their respective store methods for executing them.

// operations.js

export const todoQuery = `
  query todoQuery {
    todo {
      label
      completed
    }
  }
`

export const createTodoMutation = `
  mutation createTodoMutation($label: String!, $completed: Boolean) {
    createTodo(label: $label, completed: $completed) {
      id
      label
      completed
    }
  }
`

Read more about queries and mutations

import { todoQuery, createTodoMutation } from 'blips'

store.query(todoQuery, { id: '3c4a086e-2151-4b54-acb2-13044ea553c1' }).then(res => {
  console.log(res)
  // {
  //   data: {
  //     todo: {
  //       label: 'Buy milk',
  //       completed: false
  //     }
  //   }
  // }
})

store.mutate(createTodoMutation, { id: '4ecca858-67f8-491e-94cc-48b262061819', label: 'Learn Blips' }).then(res => {
  console.log(res)
  // {
  //   data: {
  //     createTodo: {
  //       id: '4ecca858-67f8-491e-94cc-48b262061819',
  //       label: 'Learn Blips',
  //       completed: false
  //     }
  //   }
  // }
})

If you've used Redux before you can think of execution methods as dispatchers. All executors may receive three arguments:

• operation: can be in the form of documents or documentASTs
• variables: object containing the values for field arguments
• context: object containing any additional data that you might want to pass to the resolvers, like the currently logged in user, tokens etc. (This will extend the default context)

All executors return promises that eventually resolve/reject with an object containing a data prop which holds the requested data. If, for some reason, the operation was not successful, the data prop will hold an error object instead of the requested field.

Managing your state in an asynchronous way might seem scary but unless you're actually fetching data from a web service, hitting an API or performing database queries, your data will be available with the next tick. This behaviour also comes with a few benefits:

• it forces you to introduce checks and error handlers which will eventually result in (I think) better code.
• you can write abstractions or helpers for managing your data and they will apply everywhere.
• it's much easier if you decide to switch from caching some of your state locally to exclusively fetching it from an external service.

Plus, using async/await will still make your code look synchronous and badass.

1. GraphQL subscriptions RFC
2. Proposal for GraphQL Subscriptions by Apollo

In addition to polling queries or scheduling them to execute at different moments throughout your application life-cycle to keep you data up-to-date, you can also subscribe to changes through GraphQL subscriptions:

const allTodosSubscription = `
  subscription allTodosSubscription {
    allTodos {
      id
      label
      completed
    }
  }
`

const allTodosObservable = await store.subscribe(allTodosSubscription)
const sub = allTodosObservable.subscribe(
  res => {
    // { data: { allTodos: [ ... ] } }
  },
  err => {
    // { data: { error: { ... } } }
  }
)

// remember to dispose of any unused subscriptions
// sub.unsubscribe()

Blips uses Apollo's graphql-subscriptions PubSub implementation, where any query or mutation would publish data over a specific topic and have your subscription resolvers subscribe to one or more topics.

In order to use subscriptions with Blips, you need to have access to the store's PubSub instance. We can achieve that by passing a resolvers function instead of an object to createStore. This function accepts as first argument an object containing the PubSub instance and the withFilter method.

// resolvers.js

export default ({ pubsub, withFilter }) => ({
  Query: { ... },
  Subscription: {
    allTodos: {
      resolve: (obj, args, { store }) => store.data.todos || [],
      subscribe: () => pubsub.asyncIterator([ 'TODO_UPDATED', 'TODO_CREATED', 'TODO_DELETED', ]),
    }
  }
  Mutation: {
    createTodo: (obj, { id, label, completed = false }, { store: { data: { todos } } }) => {
      const newTodo = { id, label, completed }
      todos = [ ...todos, newTodo ]

      pubsub.publish('TODO_CREATED')
      return newTodo
    }
  }
})

In the above example, the createTodoMutation publishes over TODO_CREATED as soon as the todos get updated with the new entry. It doesn't need to publish any data because the allTodos subscription will return all the todos and it's resolve method will do just that.

When publishing data to subscribers, we need to make sure that each subscribers get only the data it needs.

To do so, we can use withFilter helper from this package, which wraps AsyncIterator with a filter function, and let you control each publication for each user.

// types.js

export default {
  ...
  type Subscription {
    todo(id: String!): Todo!
  }
  ...
}

// resolvers.js

export default ({pubsub, withFilter}) => ({
  Query: { ... },
  Subscription: {
    todo: {
      subscribe: withFilter(
        () => pubsub.asyncIterator('TODO_UPDATED'),
        (payload, variables) => payload && payload.todo.id === variables.id
      )
    }
  },
  Mutation: {
    updateTodo: (obj, { id, label, completed }, { store }) => {
      const modified = store.patch(id, {
        label,
        completed,
      })('todos')
      pubsub.publish('TODO_UPDATED', { todo: modified })
      return modified
    }
  }
})

The updateTodoMutation will publish the updated todo over the TODO_UPDATED topic. If we have subscriptions registered for each todo, only those who's ID variable match to the updated todo's ID will emit.

Head over to graphql-subscriptions for a more detailed documentation on how to use this PubSub implementation.

Your access is not limited to the default context, nor limited to working with only the local state. Resolvers can completely ignore any context and just make async requests to external API servers.

const resolvers = {
  Mutation: {
    createTodo: async () =>
      await fetch('api/v1/todos', {
        method: 'post',
        body: JSON.stringify(args)
      }).then(res => res.json())
  }
}

const createTodoMutation = `
  mutation createTodoMutation($label: String!, completed: Boolean) {
    createTodo(label: $label, completed) {
      id
      label
      completed
    }
  }
`

const newTodo = await store.mutate(createTodoMutation, { label: 'Buy milk' })

import { createStore , createNetworkInterface } from 'blips'
import resolvers from './resolvers'
import typeDefs from './types'

const initialState = {}
const networkInterface: createNetworkInterface({ uri: 'http://localhost:8080/graphql' })

const store = createStore({ typeDefs, resolvers }, initialState, networkInterface)

const allTodosQuery = `
  query allTodosQuery($first: Int) {
    allTodos(first: $first) {
      id
      label
      completed
    }
  }
`

const todos = await store.fetch(allTodosQuery, { first: 10 })

• Props to the awesome developers at Facebook for giving us GraphQL
• Props to the guys at Apollo for their awesome work. Using apollo-client is what inspired me to start working on this library, and they also provide some of the tools it depends on.