CoroutineScope 与生命周期

Every coroutine belongs to a CoroutineScope. The scope decides how long the work may live and when it should stop.

Tie a scope to a screen or session. When that thing ends, the scope cancels its coroutines so no work runs against a dead UI.

You begin a coroutine with launch on a scope. It runs the block and tracks it as part of that scope.

Create one with a Job and a dispatcher. This shared scope can drive async work for a ViewModel both apps use.

A SupervisorJob stops one failed child from cancelling its siblings. One broken request will not kill the whole screen.

Call cancel when the owner goes away. Pending network or timer coroutines stop immediately, preventing leaks and stray updates.

Coroutines check for cancellation at suspension points. Your suspend functions stop cleanly, which is why you rarely need manual flags.

On Android a ViewModel already offers viewModelScope. It cancels automatically when the ViewModel is cleared, so you reuse it.

On iOS there is no automatic scope. You hold your shared scope and cancel it from Swift when the view disappears. 📱

Avoid GlobalScope: it lives forever and ignores lifecycles. Its coroutines leak and may update screens that no longer exist.

Pick the right scope once and cancellation becomes free. Lifecycle-aware scopes keep shared async work safe on both platforms. 🎯

You learned that a CoroutineScope ties async work to a lifecycle: launch starts it, cancel cleans it up, and you avoid GlobalScope. 🎉