Compose Adoption at Pinterest: A Postmortem
summary
Company tries Jetpack Compose for core Android UI and measures actual throughput and latency against a decade of hand-tuned views, finds performance regression on the home feed where code uses custom drawables and micro-optimizations. Experiments run with concrete metrics every release or quarterly and results show Compose isn’t yet parity, so a firm exploding deadline plus continued quarterly experiments decides whether to adopt or rebaseline — outcome: postpone full migration until metrics and performance match.
AI-written summary · about
chapters (11)
- 0:00Kotlin Conversations intro
- 1:47Christina Lee’s Kotlin journey
- 3:54Compose adoption attempt
- 5:16Initial Compose challenges
- 7:08Legacy codebase performance
- 8:53Balancing Compose and business
- 10:18Compose adoption deadlines
- 12:23External Compose pressures
- 13:26Enthusiasm for Compose
- 14:50Most exciting Compose feature
- 16:36Watch full talk
related talks
01h 08mAdopting Jetpack Compose @ Scale
34mRetrofitting Legacy Code with Jetpack Compose at Scale
37mUsing Performance Management Tools: Firebase Performance Monitoring and Macrobenchm…
21mPerformance best practices for Jetpack Compose
36mCoupling Compose- Interoperability in a Legacy Codebase
43mBranching out to Jetpack Compose
17mWhat's new in Jetpack Compose
46mOptimizing Compose: Issues, Experience, Tactics, Strategy