// build.gradle.kts
commonMain.dependencies {
implementation("dev.brewkits:kmpworkmanager:2.5.0")
}Android setup
class MyApplication : Application() {
override fun onCreate() {
super.onCreate()
KmpWorkManager.initialize(
context = this,
workerFactory = AppWorkerFactory() // Must implement AndroidWorkerFactory
)
}
}iOS setup
1. AppDelegate:
@main
class AppDelegate: UIResponder, UIApplicationDelegate {
override init() {
super.init()
// IOSModuleKt.iosModule calls kmpWorkerModule(workerFactory = IosWorkerFactoryGenerated())
KoinInitializerKt.doInitKoin(platformModule: IOSModuleKt.iosModule)
}
func application(_ application: UIApplication,
didFinishLaunchingWithOptions launchOptions: [UIApplication.LaunchOptionsKey: Any]?) -> Bool {
let koin = KoinIOS()
BGTaskScheduler.shared.register(
forTaskWithIdentifier: "kmp_chain_executor_task",
using: nil
) { task in
IosBackgroundTaskHandler.shared.handleChainExecutorTask(
task: task,
chainExecutor: koin.getChainExecutor()
)
}
return true
}
}2. Info.plist:
<key>BGTaskSchedulerPermittedIdentifiers</key>
<array>
<string>kmp_chain_executor_task</string>
<!-- Add other worker bgTaskIds here -->
</array>Full setup: docs/platform-setup.md
// One-time — runs as soon as constraints are met
scheduler.enqueue(
id = "nightly-sync",
trigger = TaskTrigger.OneTime(initialDelayMs = 0),
workerClassName = "SyncWorker",
constraints = Constraints(requiresNetwork = true)
)
// Periodic — every 15 minutes
scheduler.enqueue(
id = "heartbeat",
trigger = TaskTrigger.Periodic(intervalMs = 15 * 60 * 1000L),
workerClassName = "SyncWorker"
)You can implement background logic in your commonMain code, but KMP WorkManager expects platform-specific factory registration. We recommend using kmpworker-ksp to auto-generate this boilerplate.
// commonMain — shared logic
class SyncWorker : Worker {
override suspend fun doWork(input: String?, env: WorkerEnvironment): WorkerResult {
val items = api.fetchPendingItems()
database.upsert(items)
return WorkerResult.Success("Synced ${items.size} items")
}
}// androidMain
import dev.brewkits.kmpworkmanager.annotations.Worker
@Worker(name = "SyncWorker")
class SyncWorkerAndroid : AndroidWorker {
override suspend fun doWork(input: String?, env: WorkerEnvironment) =
SyncWorker().doWork(input, env)
}
// iosMain
import dev.brewkits.kmpworkmanager.annotations.Worker
@Worker(name = "SyncWorker", bgTaskId = "sync_task")
class SyncWorkerIos : IosWorker {
override suspend fun doWork(input: String?, env: WorkerEnvironment) =
SyncWorker().doWork(input, env)
}The name argument must match the workerClassName you pass to scheduler.enqueue(...) ("SyncWorker" above). Set it explicitly so ProGuard/R8 rename of the wrapper class doesn't break factory lookup.
Note: Use AndroidWorkerFactoryGenerated() and IosWorkerFactoryGenerated() in your DI/Initialization if you use KSP. Otherwise, manually implement AndroidWorkerFactory and IosWorkerFactory.
// Multi-step workflows that survive process death.
// If step 47 of 100 was running when iOS killed the app —
// the next BGTask invocation resumes at step 47, not step 0.
scheduler.beginWith(TaskRequest("DownloadWorker", inputJson = """{"url":"$fileUrl"}"""))
.then(TaskRequest("ValidateWorker"))
.then(TaskRequest("TranscodeWorker"))
.then(TaskRequest("UploadWorker", inputJson = """{"bucket":"processed"}"""))
.withId("transcode-pipeline", policy = ExistingPolicy.KEEP)
.enqueue()Most KMP libraries wrap the happy path — iOS BGTaskScheduler is not just "a different API." It has a credit system that punishes apps overrunning their time budget, an opaque scheduling policy, and no recovery mechanism for incomplete work. Getting it wrong means your tasks silently stop running.
| Android | iOS | |
|---|---|---|
| Scheduling | Deterministic via WorkManager | Opportunistic — OS decides when |
| Exact timing | ✅ AlarmManager | |
| Chain recovery | ✅ WorkContinuation | ✅ Step-level persistence |
| Time budget enforcement | — | ✅ Adaptive (reserves 15–30% safety margin) |
| Queue integrity | ✅ | ✅ CRC32-verified binary format |
| Thread-safe expiry | ✅ | ✅ AtomicInt shutdown flag |
| Trigger | Android | iOS | Notes |
|---|---|---|---|
OneTime(delayMs) |
WorkManager | BGTaskScheduler | Minimum delay may be enforced by OS |
Periodic(intervalMs) |
WorkManager | BGTaskScheduler | Min 15 min on both platforms |
Exact(epochMs) |
AlarmManager | Best-effort | iOS cannot guarantee exact timing |
Windowed(earliest, latest) |
WorkManager with delay | BGTaskScheduler | Preferred over Exact on iOS |
ContentUri(uri) |
WorkManager ContentUriTrigger | — | Android only |
v2.5.0 is a hardening release driven by a production architecture review for camera-app workloads. Highlights:
- Parallel HTTP download/upload workers —
ParallelHttpDownloadWorkersplits one file into N HTTPRangechunks (default 4, up to 16) with per-chunk.partNresume;ParallelHttpUploadWorkerruns one POST per file under amaxConcurrentsemaphore. - Checksum verification for
HttpDownloadWorker—expectedChecksum+ChecksumAlgorithm(MD5 / SHA-1 / SHA-256 / SHA-512) via Okio'sHashingSource. - DuplicatePolicy on
HttpDownloadConfig—OVERWRITE(default, preserves pre-v2.5 behaviour),SKIP(return Success without network),RENAME(append_1,_2, … to the stem). - iOS background URLSession download —
IosBackgroundDownloadWorkersurvives full app termination; persisted state store ensures cold-launch completion events are delivered to the rightsavePath(the P0 bug fixed in this release). - iOS chain retry honoring —
WorkerResult.Retry(delayMs, attemptCap)is now honored at the chain executor level on iOS viaChainProgress.stepRetryCounts+ChainExecutor.requestedNextBgTaskDelayMs. - Android FGS type configurable —
KmpHeavyWorker.foregroundServiceTypeis overrideable. Companion-object aliases (FGS_DATA_SYNC,FGS_MEDIA_PROCESSING,FGS_CAMERA, …) make camera-app workloads first-class. - Adversarial test coverage — collision proof for
PendingIntentrequest codes (CRC32 vsString.hashCode),BroadcastReceiverlifecycle (Robolectric), iOS per-step retry counter, backward-compat with v2.4.3-shaped JSON files, cold-launch survival for background URLSession state. - Hard-limit docs —
docs/IOS_BGTASK_LIMITS.md,docs/ANDROID_FGS_GUIDE.md,docs/APPLE_APP_STORE_REVIEW_GUIDELINES.md.
Full breakdown: CHANGELOG.md. Upgrade notes for users on v2.4.x:
docs/MIGRATION_V2.5.0.md.
Previously, every task ID had to be declared in BGTaskSchedulerPermittedIdentifiers before scheduling. v2.4.3 removes that constraint: only the single master dispatcher ID needs to be in Info.plist. All other task IDs are routed through an internal AppendOnlyQueue and executed when the OS fires the master dispatcher slot.
// This ID does NOT need to be in Info.plist
scheduler.enqueue(
id = "user-${userId}-daily-sync", // dynamic, per-user ID
trigger = TaskTrigger.Periodic(intervalMs = 24 * 60 * 60 * 1000),
workerClassName = "DailySyncWorker"
)<!-- Info.plist — only one entry needed for all dynamic tasks -->
<key>BGTaskSchedulerPermittedIdentifiers</key>
<array>
<string>kmp_master_dispatcher_task</string>
<string>kmp_chain_executor_task</string>
</array>Added granular control over the first execution of periodic tasks. You can now defer the initial run or set a specific delay, ensuring your app doesn't choke on heavy sync tasks immediately upon startup.
// Run every 1 hour, but defer the very first run by 1 hour
TaskTrigger.Periodic(
intervalMs = 3600_000,
runImmediately = false
)iOS developers can now use idiomatic Double (seconds) instead of Long (milliseconds) for all triggers, making the API feel native to the Apple ecosystem.
// Swift
let trigger = createTaskTriggerPeriodicSeconds(
intervalSeconds: 3600,
initialDelaySeconds: 600
)The host application no longer needs to copy and maintain 150+ lines of Swift boilerplate to handle iOS background tasks. The library now provides a native Kotlin API that handles the entire lifecycle:
// AppDelegate.swift — now just 3 lines to handle any task
BGTaskScheduler.shared.register(forTaskWithIdentifier: taskId, using: nil) { task in
IosBackgroundTaskHandler.shared.handleSingleTask(
task: task,
scheduler: koin.getScheduler(),
executor: koin.getExecutor()
)
}This handler automatically:
- Sets up the
expirationHandlerfor graceful shutdown. - Resolves worker metadata (
workerClassName,inputJson) from file storage. - Executes the worker with timeout protection via
SingleTaskExecutor. - Auto-reschedules periodic tasks and the chain executor if the queue is not empty.
- Performs deadline checks for windowed tasks.
We've overhauled the iOS storage engine to ensure industrial-grade reliability:
- Okio Streaming: All file operations (Events, History, Queue) now use Okio. Peak RAM usage is now constant (O(1)) regardless of file size, preventing OOM kills on older devices.
- Race Condition Fixes: Critical fix in
IosFileCoordinatorusingAtomicIntto ensure background blocks are executed exactly once, even during high-concurrency stress or timeouts. - Self-Healing Queue: We now detect CRC32 checksum mismatches and automatically recover/reset corrupted records to prevent persistent job stalls.
- UTF-8 Safety: Guaranteed safety against multi-byte character corruption (Emoji/CJK) at chunk boundaries.
Every chain execution is persisted locally. Collect, upload, clear:
lifecycleScope.launch {
val records = scheduler.getExecutionHistory(limit = 200)
if (records.isNotEmpty()) {
analyticsService.uploadBatch(records)
scheduler.clearExecutionHistory()
}
}Each ExecutionRecord carries chainId, status (SUCCESS / FAILURE / ABANDONED / SKIPPED / TIMEOUT), durationMs, step counts, error message, retry count, and platform. Up to 500 records kept; older ones pruned automatically.
Route task lifecycle events to Sentry, Crashlytics, or Datadog:
KmpWorkManagerConfig(
telemetryHook = object : TelemetryHook {
override fun onTaskFailed(event: TelemetryHook.TaskFailedEvent) {
Sentry.captureMessage("Task failed: ${event.taskName} — ${event.error}")
}
override fun onChainFailed(event: TelemetryHook.ChainFailedEvent) {
analytics.track("chain_failed", mapOf(
"chainId" to event.chainId,
"failedStep" to event.failedStep
))
}
}
)Six events: onTaskStarted, onTaskCompleted, onTaskFailed, onChainCompleted, onChainFailed, onChainSkipped. All have default no-op implementations.
LOW, NORMAL, HIGH, CRITICAL. On Android, HIGH/CRITICAL map to expedited work. On iOS, the queue is sorted by priority before each BGTask window:
scheduler.beginWith(
TaskRequest(workerClassName = "PaymentSyncWorker", priority = TaskPriority.CRITICAL)
).enqueue()KmpWorkManagerConfig(minBatteryLevelPercent = 10) // defer when < 10% and not chargingDefault 5%. Works on both platforms.
// iosMain
@Worker("SyncWorker", bgTaskId = "com.example.sync-task")
class SyncWorker : IosWorker { ... }
// kmpWorkerModule() automatically validates bgTaskId against Info.plist at startup
kmpWorkerModule(workerFactory = IosWorkerFactoryGenerated())Add to build.gradle.kts:
plugins { id("com.google.devtools.ksp") }
dependencies {
ksp("dev.brewkits:kmpworker-ksp:2.5.0")
commonMain.implementation("dev.brewkits:kmpworker-annotations:2.5.0")
}| Worker | Status | Notes |
|---|---|---|
HttpRequestWorker |
Stable | One-shot HTTP with configurable method, headers, body. SSRF-validated. |
HttpDownloadWorker |
Stable (v2.5+) | Resumable download via HTTP Range. <savePath>.partial survives process kill; a process kill resumes from last byte. Supports SHA-256/SHA-1/SHA-512/MD5 checksum verification and DuplicatePolicy (overwrite / skip / rename). |
ParallelHttpDownloadWorker |
New in v2.5 | Splits a single file into N (1..16, default 4) HTTP Range chunks downloaded concurrently with per-chunk .partN resume. Automatic sequential fallback when the server does not advertise Accept-Ranges: bytes. Same checksum verification surface as HttpDownloadWorker. |
HttpUploadWorker |
Streaming multipart upload. No resumable / chunked upload yet (see ParallelHttpUploadWorker for multi-file uploads). |
|
ParallelHttpUploadWorker |
New in v2.5 | One POST per file with per-host maxConcurrent limit (1..16, default 3) and per-file retry on 5xx / network errors (maxRetries 0..5). Per-file outcomes exposed via WorkerResult.Success.data.fileResults. |
IosBackgroundDownloadWorker |
iOS-only, experimental (v2.5+) | Hands the download to URLSessionConfiguration.background so the transfer survives full app termination. Host AppDelegate must wire application(_:handleEventsForBackgroundURLSession:completionHandler:) — see docs/IOS_BACKGROUND_URL_SESSION.md. |
HttpSyncWorker |
Stable | Fetch-and-persist data sync. |
FileCompressionWorker |
✅ Android · 🚧 iOS | iOS has no ZIP codec in Kotlin/Native. The default behavior on iOS is to fail fast with an explicit error. Set FileCompressionConfig.allowIosUncompressedFallback = true to accept an uncompressed copy at the output path (useful for demo chains; the output is not a real ZIP). For real iOS compression, integrate ZIPFoundation via cinterop. |
Camera / media-app advisory. For burst upload (50 photos at once), use
ParallelHttpUploadWorkerinstead of one chain step per file. For RAW / video downloads over cellular, preferIosBackgroundDownloadWorkeron iOS so the transfer survives swipe-to-quit.HttpUploadWorkeris the only stable worker without resumable/chunked semantics — pin those uploads to Wi-Fi (Constraints(requiresUnmeteredNetwork = true)) until v2.6.
SSRF protection — all built-in worker HTTP calls are validated before dispatch. Blocked:
169.254.169.254 AWS/GCP/Azure IMDS
fd00:ec2::254 AWS EC2 (IPv6)
100.100.100.200 Alibaba Cloud metadata
localhost, 0.0.0.0/8, [::1], 10.x, 172.16–31.x, 192.168.x
100.64.0.0/10 CGNAT (Tailscale, carrier-grade NAT)
fc00::/7, fe80::/10
RFC 3986 UserInfo bypass and multi-@ authority attacks are both handled. DNS rebinding is a known limitation — use certificate pinning or an egress proxy for high-trust environments.
Input size validation — inputs exceeding WorkManager's 10 KB Data limit throw IllegalArgumentException at enqueue time.
600+ tests across commonTest, iosTest, androidInstrumentedTest
QA_PersistenceResilienceTest— 100-step chain killed at step 50, resumes at exactly step 50V236ChainExecutorTest— time budget, shutdown propagation, batch loop correctnessIosExecutionHistoryStoreTest— save/get/clear, auto-pruning, all status variantsAppendOnlyQueueTest— CRC32 corruption detection, truncation recovery, concurrent accessSecurityValidatorTest— SSRF, IPv6 compressed loopback, multi-@UserInfo bypass
| Quick Start | Running in 5 minutes |
| Platform Setup | Android & iOS configuration |
| API Reference | Full public API |
| Task Chains | Chain API and recovery semantics |
| Built-in Workers | Worker reference and input schema |
| Constraints & Triggers | All scheduling options |
| iOS Best Practices | BGTask gotchas and recommendations |
| iOS BGTask Hard Limits | Opportunistic scheduling, time budget, headless DI |
| App Store Review Compliance | §2.5.4 — what gets rejected and how to ship safely |
| Android FGS Type Guide | mediaProcessing / camera / dataSync setup |
| iOS Background URLSession | Surviving app termination during long downloads |
| Troubleshooting | Common issues |
| CHANGELOG | Release history |
Migration: v2.2.2 → v2.3.0 · v2.3.3 → v2.3.4 · v2.4.x → v2.5.0
| Kotlin | 2.1.0+ |
| Android | 8.0+ (API 26) |
| iOS | 13.0+ |
| Gradle | 8.0+ |
./gradlew :kmpworker:allTests # all platforms must pass before opening a PRCommit messages follow Conventional Commits.
Apache 2.0. See LICENSE.