Flirting with Flutter @ Getaround

Ishan Khanna presents a case study into integrating flutter into the native iOS and Android apps at Getaround.

Transcript

1. FLIRTING WITH FLUTTER AT GETAROUND ISHAN KHANNA ANDROID ENGINEER 09-DEC-2019
2. WHAT I WILL TALK ABOUT?
3. MOTIVATION
4. WORLD CLASS USER EXPERIENCE
5. WORLD CLASS USER EXPERIENCE WORLD CLASS DEV EXPERIENCE
6. SCRATCHING THE SURFACE
7. WHAT DO WE NEED
8. WHAT DO WE NEED • ARCHITECTURE
9. WHAT DO WE NEED • ARCHITECTURE • DEVELOPMENT STRATEGY
10. WHAT DO WE NEED • ARCHITECTURE • DEVELOPMENT STRATEGY • ADVOCACY
11. HOW TO FIND OUT?
12. BUILD A MOVIES DB APP...
13. BUILD A MOVIES DB APP… BECAUSE IT’S THE WAY BEST WAY TO PROVE, A TECHNOLOGY WORKS!
14. SO WE BUILT IT
16. NOT JUST ONCE
17. NOT JUST ONCE, BUT MULTIPLE TIMES!
18. WHY, THOUGH?
19. THIS WAS PRETTY EASY! AND IT WAS LOVE AT FIRST BYTE :)
20. NEXT, PUSHING THE BOUNDARIES
21. ALL THIS WHILE ACTIVELY DOCUMENTING KEY LEARNINGS
22. WHAT DO WE NEED • ARCHITECTURE • DEVELOPMENT STRATEGY
23. HOW TO USE FLUTTER?
24. • COMPLETE REWRITE
25. • COMPLETE REWRITE • PLATFORM VIEWS
26. Flutter Popup over Native View
27. Flutter Popup over Native View Flutter Bottom Sheet over Native View
28. Flutter Popup over Native View Flutter Bottom Sheet over Native View Native listview with one Flutter cell in list.
29. • COMPLETE REWRITE • PLATFORM VIEWS • ADD TO APP
30. GIT HAPPENS!
31. • GIT SUB-TREES • GIT SUB-MODULES • GIT MONOREPO
32. HOW COMPLEX IS IT TO? • CLONE THE REPO
33. HOW COMPLEX IS IT TO? • CLONE THE REPO • PULL CHANGES FROM FLUTTER
34. HOW COMPLEX IS IT TO? • CLONE THE REPO • PULL CHANGES FROM FLUTTER • PUSH CHANGES TO FLUTTER
35. HOW COMPLEX IS IT TO? • CLONE THE REPO • PULL CHANGES FROM FLUTTER • PUSH CHANGES TO FLUTTER • PULL CHANGES FROM NATIVE
36. HOW COMPLEX IS IT TO? • CLONE THE REPO • PULL CHANGES FROM FLUTTER • PUSH CHANGES TO FLUTTER • PULL CHANGES FROM NATIVE • PUSH CHANGES TO NATIVE
37. MONOREPO FOR THE WIN!!
38. ALL SET FOR DEVELOPMENT, ARE WE?
39. LET’S TALK CI
40. STRAIGHT FORWARD SETUP ON CIRCLE CI
41. BUILDS RAN CONDITIONALLY
42. SHIPPING AN EASTER EGG!
44. WHAT SHOULD WE BUILD, THAT IS VALUABLE FOR CUSTOMERS
45. WE SHORTLISTED SOME FEATURES BASED ON EFFORT/IMPACT RATIO
46. WE WANTED TO PICK, LOW EFFORT & HIGH IMPACT
47. MEDIUM EFFORT & MEDIUM IMPACT
48. WITHOUT FURTHER ADO, I CAN TELL YOU!
49. WE SHIPPED IT, AS WELL!
54. WHAT DO WE NEED • ARCHITECTURE • DEVELOPMENT STRATEGY • ADVOCACY
55. WE ARE A DISTRIBUTED TEAM ACROSS SFO & AMS
56. ON SITE TRAININGS IN SF
57. THE RESPONSE AT THE END?
58. :| NEUTRAL
59. OOPS, IS IT REALLY FASTER?
60. LET’S FIND OUT
61. FRAMING THE HYPOTHESIS
62. THE BAKE-OFF
63. TEAM SETTING
64. THE RESULTS?
65. FLUTTER WAS ONLY 1.4X FASTER*
66. THE GOOD, THE BAD, THE UGLY
67. ADD TO APP INSTABILITY WHILE DEVELOPMENT
68. DART VM FAILED TO INITIALIZE
69. HOT RELOAD GOES OUT OF SYNC
70. BUILDS WERE BROKEN DUE TO CATALINA UPGRADE ON MAC OS
71. ANDROID STUDIO KILLED iOS SIMULATORS WHEN DOING HOT RELOADS
72. PROBLEM WITH REMOVING FOCUS FROM A TEXTFIELD
73. FLUTTER FRAGMENT ON ANDROID SHOULD BE PRESENTER IN `TEXTURE` RENDER MODE
74. NATIVE DEVELOPERS WERE OFTEN AFFECTED BECAUSE FLUTTER HAD TO UPGRADED OR INSTALLED ON THEIR MACHINES
75. THE DECISION
76. WE HAVE DECIDED TO PULL THE PLUG ON IT
77. NOT USING IT ANYMORE ON THE EXISTING APP
78. NEXT STEPS FOR US, WOULD BE TO REMOVE FLUTTER COMPLETELY AND BUILD THE OWNER RESERVATIONS NATIVELY
79. THE FUTURE
80. FLUTTER IS A GREAT TECHNOLOGY
81. SERIOUSLY, I AM NOT KIDDING!
82. FLUTTER IS A GREAT TECHNOLOGY
83. LOOKING FOR TOOLING TO IMPROVE FOR PROGRESSIVE MIGRATION
84. SUMMARY OF RISKS:
85. 1. TECHNICAL
86. 1. TECHNICAL 2. ORGANIZATIONAL
87. 1. TECHNICAL 2. ORGANIZATIONAL 3. ECOSYSTEM
88. SPECIAL MENTION ABOUT THE BLOG POST FROM AIRBNB ABOUT SUNSETTING REACT NATIVE
89. RELATIONSHIP STATUS:
90. WE DON’T TALK ANYMORE, LIKE WE USED TO DO...
91. QUESTIONS? @droidchef