Azure / Azure/azure-functions-durable-python
Add regression coverage for None-returning orchestrations
- Dominant language
- Python
- Stars
- 157
- Forks
- 70
- Avg merge
- 2d 10h
- Merged PRs (30d)
- 2
Description
💡 **Feature description**
Add explicit regression tests for all supported `None`-return paths in orchestration replay:
- A sub-orchestrator whose completion event has `Result=None`.
- A top-level generator orchestrator that completes without an explicit return value.
Activity completion with `Result=None` is already covered by `test_call_activity_with_none_return`. The additional tests should verify that replay completes successfully, produces the expected `None` task value where applicable, and does not call JSON deserialization with `None`.
💭 **Describe alternatives you've considered**
Relying on the shared guard in `TaskOrchestrationExecutor.parse_history_event` and the existing activity test leaves the separate `SUB_ORCHESTRATION_INSTANCE_COMPLETED` branch without direct regression coverage. A top-level no-return test would also preserve the behavior that originally made issue #260 difficult to diagnose.
**Additional context**
Issue #260 reported the original failure. The root cause was isolated in #454 and fixed by #466, released in `azure-functions-durable` 1.2.9. Current code handles both task and sub-orchestration completion events with null results, but only the activity path has a focused regression test.
Contributor guide
Assessment
This issue has not been assessed yet.