temporal-python-testing

Use WorkflowEnvironment.start_time_skipping() from temporalio.testing to create a test environment where workflow sleeps complete instantly. Run a Worker with your workflows and activities inside the environment, then execute the workflow via the client and assert on the result.

Pass a mock function or a test double activity to the Worker's activities list instead of the real activity. This isolates workflow orchestration logic from external services, and you can use dynamic mocks to simulate success, transient failures, or non-retryable ApplicationError cases.

Replay testing re-executes workflow code against recorded event histories using the Replayer class. If the new code makes the same decisions as the recorded history, the change is deterministic and safe to deploy; mismatches indicate breaking changes.

Replay fails when code changes alter the sequence of workflow commands, such as adding or reordering activity calls. Use workflow.get_version() for backward-compatible changes, and avoid non-deterministic calls like random.randint, datetime.now, or direct HTTP requests inside workflows.

Use ActivityEnvironment from temporalio.testing to run an activity directly without a workflow or server. It supports testing heartbeats, cancellation handling, and exception propagation by passing inputs and asserting on outputs or raised errors.

Temporal's recommended target is at least 80 percent coverage for both workflow and activity logic. Combine unit tests with time-skipping, integration tests with mocked activities for critical paths, and replay tests for all workflow versions before deployment.