Design every step so it is safe to kill and resume.
if retries < 5 { backoff(attempt) }
React Native in depth reels
React Native in depth1 / 2
1 / 4
Runtime may not exist
When the OS wakes your app for background work, it starts native code. Booting the JavaScript runtime costs time and memory the OS may not grant, and iOS offers no general way to run JS in the background at all.
2 / 4
Memory and bytes
Moving a large file through JavaScript means holding it in the JS heap, and a few hundred megabytes is enough to crash the app. Files should cross into native code only as a path.
3 / 4
Split
Keep screens, navigation and UI state in JavaScript. Put scheduling, transfers and ideally the sync engine in native modules. Native code writes progress into SQLite and JavaScript observes it, so both share one source of truth.
Interview answer
Say it in the interview
In React Native, JavaScript owns screens and state, while scheduling, transfers and the sync engine are native. Native code writes progress to SQLite and JavaScript observes it. File bytes never pass through the JS heap; only paths cross into native code.
1 / 4
Runtime may not exist
When the OS wakes your app for background work, it starts native code. Booting the JavaScript runtime costs time and memory the OS may not grant, and iOS offers no general way to run JS in the background at all.
2 / 4
Memory and bytes
Moving a large file through JavaScript means holding it in the JS heap, and a few hundred megabytes is enough to crash the app. Files should cross into native code only as a path.
3 / 4
Split
Keep screens, navigation and UI state in JavaScript. Put scheduling, transfers and ideally the sync engine in native modules. Native code writes progress into SQLite and JavaScript observes it, so both share one source of truth.
Interview answer
Say it in the interview
In React Native, JavaScript owns screens and state, while scheduling, transfers and the sync engine are native. Native code writes progress to SQLite and JavaScript observes it. File bytes never pass through the JS heap; only paths cross into native code.
Why JS can’t own background sync
#react-native
❚❚
React Native in depth2 / 2
1 / 4
Level one: libraries
For an MVP, react-native-background-fetch or expo-background-task runs short JS jobs, a background upload library handles files, and WatermelonDB provides a sync protocol. It ships fast, but tight time budgets and JS start-up cost hold it back.
2 / 4
Level two: native modules
At scale, write Turbo Modules that expose enqueueSync and enqueueUpload. Android uses WorkManager underneath; iOS uses BGTaskScheduler and background URLSession. JavaScript only enqueues work and watches the database.
3 / 4
Level three: a shared core
Larger teams write the sync engine once, in Kotlin Multiplatform or Rust, and use it from the native Android app, the native iOS app and the React Native app. One protocol implementation and one test suite, with no drift between platforms.
In practiceApp Center CodePush was retired in March 2025; EAS Update is the usual replacement.
Interview answer
Say it in the interview
We started with background libraries for speed, then moved to Turbo Modules wrapping WorkManager and BGTaskScheduler as usage grew. With native apps alongside, I would put the sync core in Kotlin Multiplatform from day one to avoid drift between platforms.
1 / 4
Level one: libraries
For an MVP, react-native-background-fetch or expo-background-task runs short JS jobs, a background upload library handles files, and WatermelonDB provides a sync protocol. It ships fast, but tight time budgets and JS start-up cost hold it back.
2 / 4
Level two: native modules
At scale, write Turbo Modules that expose enqueueSync and enqueueUpload. Android uses WorkManager underneath; iOS uses BGTaskScheduler and background URLSession. JavaScript only enqueues work and watches the database.
3 / 4
Level three: a shared core
Larger teams write the sync engine once, in Kotlin Multiplatform or Rust, and use it from the native Android app, the native iOS app and the React Native app. One protocol implementation and one test suite, with no drift between platforms.
In practiceApp Center CodePush was retired in March 2025; EAS Update is the usual replacement.
Interview answer
Say it in the interview
We started with background libraries for speed, then moved to Turbo Modules wrapping WorkManager and BGTaskScheduler as usage grew. With native apps alongside, I would put the sync core in Kotlin Multiplatform from day one to avoid drift between platforms.