Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
GraphQL and Relay
Search
Sponsored
·
Ship Features Fearlessly
Turn features on and off without deploys. Used by thousands of Ruby developers.
→
Helielson
January 29, 2016
Programming
190
2
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
GraphQL and Relay
Concepts, Spec and how we are using it at jusbrasil.com.br
Helielson
January 29, 2016
More Decks by Helielson
See All by Helielson
Relay @ Jusbrasil
helielson
0
170
Component-based development with React
helielson
1
150
React - A JavaScript library for building user interfaces
helielson
9
2k
Other Decks in Programming
See All in Programming
AI時代の仕事技芸論〜ソフトウェア開発で「遊ぶように働く」職人的熟達のすすめ(スクフェス仙台 2026バージョン)
kuranuki
0
650
Embedded SREと共に達成した会員管理システムのAWS移行 - SRE NEXT 2026 ランチスポンサーセッション
niftycorp
PRO
1
2.7k
AI時代、エンジニアはどう育つのか -未経験エンジニアの成長を間近で見て考えたこと-
thasu0123
0
130
Apache Hive: そしてCloud Native Lakehouseへ
okumin
1
140
なぜ関数型プログラミングで「型」と「証明」が語られるのか #fp_matsuri
kajitack
3
960
OSINT for SRE: 学術論文とポストモーテムから探る システム障害の共通パターン / SRE NEXT 2026
tomoyk
1
3.9k
Hatena Engineer Seminar #37「言語モデルの活用に関する研究」
slashnephy
0
530
Go言語とトイモデルで学ぶTransformerの気持ち / fukuokago23-transformer
monochromegane
0
110
SREは、MCPとSRE Agentをこう使え!
kazumax55
0
150
AI 輔助遺留系統現代化的經驗分享
jame2408
1
1.2k
Generative UI & AI-Assistants for Your Angular Solutions
manfredsteyer
PRO
0
110
1年で人数1.5倍、PR数5.5倍増。 品質とアウトカムはどうなったか、 何が効いたか
ike002jp
0
140
Featured
See All Featured
The Impact of AI in SEO - AI Overviews June 2024 Edition
aleyda
5
1.1k
Java REST API Framework Comparison - PWX 2021
mraible
34
9.5k
Technical Leadership for Architectural Decision Making
baasie
3
440
Have SEOs Ruined the Internet? - User Awareness of SEO in 2025
akashhashmi
0
400
The innovator’s Mindset - Leading Through an Era of Exponential Change - McGill University 2025
jdejongh
PRO
1
220
Heart Work Chapter 1 - Part 1
lfama
PRO
8
36k
Crafting Experiences
bethany
1
220
Hiding What from Whom? A Critical Review of the History of Programming languages for Music
tomoyanonymous
3
980
The Organizational Zoo: Understanding Human Behavior Agility Through Metaphoric Constructive Conversations (based on the works of Arthur Shelley, Ph.D)
kimpetersen
PRO
0
390
Dominate Local Search Results - an insider guide to GBP, reviews, and Local SEO
greggifford
PRO
0
210
Getting science done with accelerated Python computing platforms
jacobtomlinson
2
320
Bridging the Design Gap: How Collaborative Modelling removes blockers to flow between stakeholders and teams @FastFlow conf
baasie
0
610
Transcript
GraphQL and Relay
@helielson @hyetho @rodrigopr @rodrigoropr
Topics - GraphQL What is it? Motivation Spec Implementations How
we use it What is coming
Definition GraphQL is a query language designed to build client
applications by providing an intuitive and flexible syntax for describing their data requirements and interactions.
Why invent something new? What about REST and Ad hoc
endpoints?
REST problems - Multiple round trips to render single views
- Receives unnecessary data - Payloads tend to grow over time for all clients
- Code duplication - API changes with client need -
Hard to scale with multiple clients - Hard to keep backwards compatibility Ad hoc endpoints (view oriented)
How we think about data? - Not tables, joins or
URIs - But: - Objects - Properties - Relationships - What, not How
GraphQL
- Product-centric - Hierarchical - Client-specified queries - Strongly-typed Principles:
Product-centric
Hierarchical
{ post(id: 65) { title, votes, author { name, avatar
} } } Hierarchical
{ post(id: 65) { title, votes, author { name, avatar
} } } Hierarchical { "post": { "title": "Post title", "votes": 100, "author" { "name": "Didi", "avatar": "http://x.jpg" } } }
Strong-Typing
Client-specific Queries
Not tied to: - Language - Framework - Database -
Protocol
Schema - Think as a Graph - Define structure of
nodes (properties) - How they are connected
Schema User { name: String! email: String! followers: [User] isFriendOf(otherEmail:
String!): Boolean }
Schema query { post(postId: Int!): Post! user(email: String!): User! me:
User }
Type System - Interface - Object - Enum - Union
- List - Scalars: - String, Int, Boolean
Query language { user(email: "
[email protected]
") { name followers { name
} } }
query { user(email: "
[email protected]
") { ...userInfo followers { ...userInfo }
} } fragment userInfo on User { name, email lastPosts { ...somePostFrament } }
fragment petInfo on Pet { petName, owner { name }
… on Dog { noiseVolume: barkVolume } … on Cat { noiseVolume: meowVolume } }
Variables query houseTrainedQuery($atOtherHomes: Boolean) { dog { isHousetrained(atOtherHomes: $atOtherHomes) }
}
Fragment directives fragment maybeFragment on Query @include(if: $condition) { me
{ name } }
Fragment directives query myQuery($someTest: Boolean) { experimentalField @skip(if: $someTest) }
Mutation mutation { submitComment(postId: 1, user: "...", content: "") {
createdComment { id, url } } }
Open to optimizations - Batch - Cache - "DataLoader is
the true" http://github.com/facebook/dataloader
Network-GraphQL (how we are using) - Unify object definition -
Unify loader definition - Helper for thrift-style relationships - Add an extra optimization layer - Auto extract selections from query
Network-GraphQL (how we are using) - Define the schema based
on jusbrasil-types - Rethinking data structure - english - relationships - building as we need - Based on network Thrift API
What's coming next? - Open source movement - Improve data
coverage - Provide authorization in schema (acl) - Mutation (done) - Object authorization - Field authorization
Questions?
Relay
Topics - Relay What is it? Motivation Spec Implementation How
we use it What is coming
Definition A javascript framework for building data-driven react applications
Dispatcher View Store Action
Dispatcher View Store Action Server
Let's make an App !! Will be easy !!!!
None
How to load data from server?
<TopicTag /> <Feed /> <Document />
A lot of trips between server and client /feed #
get feed (documents ids) /documents/{id} # get a document /topics/{id} # get a topic Reuse the endpoint
A big payload /feed_with_topics_max_tree_documents One custom endpoint that loads everything
/feed?topics=true&max_documents=3 Multiple custom endpoints
And what about: Pagination? Ordering? Error handling? Retry? Caching?
Don't forget about writes: Conflicts? Write Coordination? Latency Compensation? State
Sync?
Nah... We can solve this! Let's combine X + Y
+ Z and solve it!
None
Want to introduce changes ??? First learn about the entire
project … :(
- How to fetch data - How to coordinate requests
- How to coordinate writes - How to handle retries/errors - How to keep server/app in sync - How to update view with changes - How to cache data - How to keep that all in your head - How to …. - Your App What you had to think about:
- What data to fetch - Your App What you
want to think:
Focus on what matters most! Relay make easy for developers
to focus on the product and forget about data access nuances
Declarative Colocation Mutations Principles
Declarative The component describe what it want
// Story.js export default class Story extends React.Component { render()
{ var story = this.props.story; return ( <View> <Image uri={story.author.profilePicture.uri} /> <Text>{story.author.name}</Text> <Text>{story.text}</Text> </View> ); } }
fragment on Story { text, author { name, profilePicture {
uri } } }
export class Story extends Component { … } export default
Relay.createContainer(Story, { fragments: { story: () => Relay.QL` fragment on Story { author { name profilePicture { uri } } text } `, }, });
export class Story extends React.Component { ... } export default
Relay.createContainer(Story, { fragments: { story: () => Relay.QL` fragment on Story { author { ... } } `, }, }); Colocation
Easy to compose
export class NewsFeed extends Component { … } export default
Relay.createContainer(NewsFeed, { fragments: { feed: () => Relay.QL` fragment on Feed { stories(first: $count) { // fetch viewer's stories edges { // traverse the graph node { ${Story.getFragment('story')} // compose child fragment } } } } `, }, });
class FeedRoute extends Relay.Route { static routeName = "FeedRoute"; static
queries = { feed: () => Relay.QL` query { userFeed(uid: $userID) } `, }; } var feedRoute = new FeedRoute({userID: '123'}); Query Routes
ReactDOM.render( <Relay.RootContainer Component={Feed} route={feedRouter} renderLoading={...} // optional renderFailure={...} // optional
renderFetched={...} // optional />, domElement ); Root Containers
A common problem: Pagination
export default Relay.createContainer(NewsFeed, { initialVariables: { count: 3 }, fragments:
{ feed: () => Relay.QL` fragment on Feed { stories(first: $count) { ... } } `, }, });
class NewsFeed extends React.Component { render() { ... } loadMore()
{ var count = this.props.relay.variables.count; this.props.relay.setVariables({ count: count + 5, }); } }
followers(first: Int, last: String) { edges { node // the
actual follower object, cursor // node identifier in the edge }, pageInfo { hasNextPage, lastCursor } } Pagination: How it works?
export default Relay.createContainer(Story, { fragments: { story: () => Relay.QL`
fragment on Story { author { name followers(first: $limit) @include(if: $isDesktop) { ... } } text } `, }, }); Directives
Mutations
getMutation() { ... } getFatQuery() { ... } getConfigs() {
... } getVariables() { ... } Defining a mutation
export default class LikePostMutation extends Relay.Mutation { getMutation() { ...
} getVariables() { ... } getConfigs() { ... } getFatQuery() { ... } } Defining the mutation
getMutation() { return Relay.QL`mutation{ likePost // `likePost` mutation must be
defined on the server }`; } getMutation()
getVariables() { return { id: this.props.post.postId, like: this.props.like }; }
getVariables()
getConfigs() { return [{ type: 'FIELDS_CHANGE', fieldIDs: { post: this.props.post.id
} }]; } getConfigs()
getFatQuery() { return Relay.QL` fragment on LikePostPayload { post {
id, likesCounter }, } `; } getFatQuery()
export class Story extends React.Component { handleEditPost() { const onFailure
= () => {}; const onSuccess = () => {}; Relay.Store.commitUpdate(new LikePostMutation({ post: this.props.post, like: true }), {onFailure, onSuccess}); } } Sending the mutation
getOptimisticResponse() { return { post: { id: this.props.post.id, likesCounter: this.props.post.likesCounter
+ 1 } } } getOptimisticResponse()
interface Node { id: ID! } type Post extends Node
{ id: ID!, … // post fields } Object Tracking:
- Server-side rendering support - React Router integration - Babel
Plugin - Active development: - Subscription (live) on the work - Client state is being discussed Important Details
- Work well with others tools - Flux, Redux, etc
- Patterns are still emerging Not a silver bullet
Question?