[Raju Ahmed (raju-opti)]

Summary

• localStorage check: new isLocalStorageAvailable() in lib/utils/env/browser.ts. It returns false when reading the localStorage global throws (a ReferenceError where it isn't defined, such as SSR, or a SecurityError where storage is blocked, such as disabled cookies or sandboxed iframes) or when the global is null.
• Browser batch event processor: createBatchEventProcessor only builds its default LocalStorageCache-backed EventStore when localStorage is available. Otherwise it runs without an event store and adds a Warn startup log: localStorage is not available, events will not be persisted. A user-provided eventStore is used exactly as before, whether or not localStorage is available.
• Startup logs from platform factories: getBatchEventProcessor accepts a startupLogs option and puts those logs before its own, without changing the caller's array. The option is on the internal BatchEventProcessorFactoryOptions only; the public BatchEventProcessorOptions is unchanged.

Before this change, the browser factory always created a LocalStorageCache event store. Where localStorage can't be accessed, every store operation threw. With no event store, events are still batched and dispatched; failed events just aren't persisted for retry across page loads.

Test plan

• New lib/utils/env/browser.spec.ts: localStorage that is accessible, null, not defined (ReferenceError), and blocked (SecurityError).
• event_processor_factory.browser.spec.ts: no event store and a startup warning when localStorage is unavailable; no warning when it is available; a provided eventStore is wrapped with getPrefixEventStore whether or not localStorage is available, with no warning.
• event_processor_factory.spec.ts: provided startupLogs are included with the processor's own, and the input array isn't changed.
• Ran the three spec files locally (41 tests pass), plus tsc --noEmit, eslint on the changed files and validate-platform-isolation.

Issues

• FSSDK-13343
• [BUG] Unhandled SecurityError rejection from BatchEventProcessor.start() when localStorage is inaccessible #1182

🤖 Generated with Claude Code

[Coveralls (coveralls)]

coverage: 77.554% (-0.07%) from 77.624% — raju/local-storage-availability into master

[Copilot]

🟢 Approval recommended

The focused fallback preserves existing optional-store behavior, regression tests cover the changes, and no blocking issues were identified.

No open issues 🎉

What changed in this PR

This PR addresses #1182 by skipping the browser’s default persistent event store when localStorage is inaccessible, while preserving event batching, dispatch, and custom event stores.

Changes:

• Adds a guarded localStorage availability check and a startup warning when persistence is disabled.
• Allows platform factories to supply startup logs without mutating the caller’s array.
• Adds regression tests for storage availability, custom stores, and startup-log propagation.

File	Description
lib/​utils/​env/​browser.ts	Adds the guarded storage availability check.
lib/​utils/​env/​browser.spec.ts	Tests accessible, null, missing, and blocked storage.
lib/​message/​error_message.ts	Defines the persistence-disabled warning.
lib/​event_processor/​event_processor_factory.ts	Copies supplied startup logs before adding processor logs.
lib/​event_processor/​event_processor_factory.spec.ts	Tests log propagation and input-array preservation.
lib/​event_processor/​event_processor_factory.browser.ts	Conditionally creates the default store and supplies the warning.
lib/​event_processor/​event_processor_factory.browser.spec.ts	Tests fallback behavior and custom-store preservation.

_{Review effort: Balanced}

[Md Junaed Hossain (junaed-optimizely)]

Looks good overall. Should we add a catch in lib/event_processor/batch_event_processor.ts:275 ? The localStorage check covers the default store, but a user-provided eventStore that rejects would still cause the unhandled rejection!

[Raju Ahmed (raju-opti)]

Looks good overall. Should we add a catch in lib/event_processor/batch_event_processor.ts:275 ? The localStorage check covers the default store, but a user-provided eventStore that rejects would still cause the unhandled rejection!

In that case, I would say the responsibility to ensure that the store that is passed in works is on the user.

Md Junaed Hossain (@junaed-optimizely)