Android lifecycle and flow collection

Android Lifecycle and Flow Collection Deep Dive¶

Overview¶

Lifecycle-aware Flow collection prevents wasted work, leaks, and duplicate collectors after configuration or navigation transitions.

Core Concepts¶

repeatOnLifecycle for active-window collection
viewModelScope for producer lifetime
UI collector ownership in Fragment/Compose lifecycle scopes
restartable collection semantics

Internal Implementation¶

repeatOnLifecycle launches a child collector when lifecycle reaches target state and cancels it when lifecycle falls below that state. This makes collection restartable and avoids background collection when UI is stopped.

Threading Model¶

Collection often starts on main for UI updates, while upstream may move work with flowOn. Keep rendering lightweight and move expensive transforms upstream.

Coroutine / Flow Behavior¶

Restartable collectors must tolerate re-subscription. Hot sources with replay need careful configuration to avoid duplicate event handling on re-entry.

Code Examples¶

viewLifecycleOwner.lifecycleScope.launch {
    viewLifecycleOwner.repeatOnLifecycle(Lifecycle.State.STARTED) {
        viewModel.uiState.collect { state ->
            render(state)
        }
    }
}

Common Interview Questions¶

Q: Why is plain launch { flow.collect {} } risky in Fragments? A: Start from delivery semantics: use StateFlow for durable state, SharedFlow or Channel for transient events, and lifecycle-aware collection to prevent duplicate work.
Q: How does repeatOnLifecycle differ from one-time collection? A: Start from delivery semantics: use StateFlow for durable state, SharedFlow or Channel for transient events, and lifecycle-aware collection to prevent duplicate work.
Q: How do you avoid duplicate event consumption on recreation? A: Answer with correctness first and throughput second: cancellation model, dispatcher choice, bounded parallelism, and contention or latency measurements.
Q: What should be hot vs cold in ViewModel streams? A: Answer with correctness first and throughput second: cancellation model, dispatcher choice, bounded parallelism, and contention or latency measurements.

Production Considerations¶

tie collection to screen visibility state
keep event channels separate from persistent state
audit collectors during navigation backstack changes
test lifecycle transitions explicitly

Performance Insights¶

Stopping unnecessary collection in background states improves battery, network usage, and reduces hidden memory churn.

Senior-Level Insights¶

Staff-level discussion should include lifecycle ownership contracts between presentation and domain/data layers, not only API usage.