GraphQL by a Skeptic

There is much hype around GraphQL around these days - like there seems to be for so many other tools lately.

Popularity isn't a guarantee of quality, but that doesn't mean a popular thing can't be good either. In this talk I present the basics ideas behind GraphQL, how those ideas are implemented, and what those decisions mean in a production environment, exploring both their benefits and shortcomings.

Transcript

1. GraphQL por un escéptico 1
2. Hola!
3. Gracias ❤
4. pote.io
5. usecacao.co
6. 6 GraphQL por un escéptico
10. GraphQL
11. Facebook
12. API
13. REST
14. Resources
15. Actions
16. GET /comments HTTP/1.1
17. POST /comments HTTP/1.1 { "text": "<insert angry rant here>" }
18. DELETE /comments/1 HTTP/1.1
19. HTTP
20. Query Language
21. /
22. Let’s fetch a record
23. SQL
24. SELECT name FROM humans WHERE id = 1;
25. GraphQL
26. human(id: 1) { name }
27. human(id: 1) { name }
28. human(id: 1) { name }
29. human(id: 1) { name }
30. { "data": { "human": { "name": "Luke" } } }
31. { "data": { "human": { "name": "Luke" } } }
32. { "data": { "human": { "name": "Luke" } } }
33. human(id: 1) { name friends { name } }
34. human(id: 1) { name friends { name } }
35. human(id: 1) { name friends { name } }
36. { "data": { "human": { "name": "Luke", "friends": [ {"name": "Han” }, {"name": "Leia"}, {"name": "Obi-wan”} ] } } }
37. { "data": { "human": { "name": "Luke", "friends": [ {"name": "Han” }, {"name": "Leia"}, {"name": "Obi-wan”} ] } } }
38. { human(id: 1) { name } droid(id: 1) { name } }
39. { "data": { "human": { "name": "Luke", }, “droid”: { “name”: “R2-D2” } } }
40. { human(id: 1) { name friends { name friends {…} } } droid(id: 1) { name } }
41. Query Language
44. How do languages work?
45. Human to Machine
46. { human(id: 1) { name friends { name } } }
47. Parse
48. { human(id: 1) { name friends { name } } }
49. Abstract Syntax Tree
50. Execute
51. Type System
52. type Human { name: String friends: [Human] }
53. type Human { name: String friends: [Human] }
54. type Human { name: String friends: [Human] }
55. Query definitions
56. human(id: Int!): Human
57. human(id: Int!): Human
58. human(id: Int!): Human
59. human(id: Int!): Human
60. human(id: Int!): Human
61. human(id: Int!): Human
62. Resolver functions
63. human(obj, args, context) { let data = // SELECT * FROM // humans WHERE // id = args.id return new Human(data) }
64. human(obj, args, context) { let data = // SELECT * FROM // humans WHERE // id = args.id return new Human(data) }
65. human(obj, args, context) { let data = // SELECT * FROM // humans WHERE // id = args.id return new Human(data) }
66. human(obj, args, context) { let data = // SELECT * FROM // humans WHERE // id = args.id return new Human(data) }
67. name(obj, args, context) { return obj.name }
68. name(obj, args, context) { return obj.name }
69. Expose it
70. app.use( '/', graphqlHTTP({ schema: MyAPIDefinition, }) );
71. app.use( '/', graphqlHTTP({ schema: MyAPIDefinition, }) );
72. app.use( '/', graphqlHTTP({ schema: MyAPIDefinition, }) );
73. app.use( '/', graphqlHTTP({ schema: MyAPIDefinition, }) );
74. Resources
75. Serialisation
76. Transport
77. Types and resolvers
78. Schema is everything
79. Tradeoffs
80. Single request
83. "slim" requests
84. { "created_at": "2017-11-24 17:42:11 -0300", "updated_at": "2017-11-24 16:52:11 -0300”, ... "text": "this is an angry comment" }
85. { "text": "this is an angry comment" }
86. Optimal network roundtrips
91. Non-optimal database usage
92. Less granularity
93. { "errors": [ { "message": "Variable \"$id\" of required type \"ID!\" was not provided.", "locations": [ { "line": 1, "column": 17 } ] } ] }
94. { "errors": [ { "message": "Variable \"$id\" of required type \"ID!\" was not provided.", "locations": [ { "line": 1, "column": 17 } ] } ] }
95. { "errors": [ { "message": "Variable \"$id\" of required type \"ID!\" was not provided.", "locations": [ { "line": 1, "column": 17 } ] } ] }
96. { "errors": [ { "message": "Variable \"$id\" of required type \"ID!\" was not provided.", "locations": [ { "line": 1, "column": 17 } ] } ] }
97. Caching
98. Less HTTP
99. More complexity
100. More coupling
101. (maybe) Less maintainability
102. Type system
104. (maybe) More patience
105. Tradeoffs
106. ¯\_(ツ)_/¯
110. Gracias!
112. pote.io