Fix continueAsNew dropping fire-and-forget actions (sendEvent, signal…

[YunchuWang]

Fix #124
Problem
When an orchestration calls sendEvent() or signalEntity() before continueAsNew(), those fire-and-forget actions are silently dropped.

File: packages/durabletask-js/src/worker/runtime-orchestration-context.ts

In getActions() (line 243), when the completion status is CONTINUED_AS_NEW, only the single continue-as-new completion action is returned, dropping all other pending actions (including fire-and-forget actions like sendEvent). This is inconsistent with setComplete() (line 195) and setFailed() (line 217), which both preserve pending fire-and-forget actions.

Both setComplete and setFailed have explicit comments:

Note: Do NOT clear pending actions here - fire-and-forget actions like sendEvent must be preserved and returned alongside the complete action

Root Cause
setContinuedAsNew() stores only flags (_newInput, _saveEvents) without adding the completion action to _pendingActions. Then getActions() creates a fresh completion action and returns [action] as the sole item, discarding everything in _pendingActions.

Fix Available
Branch: copilot-finds/bug/continue-as-new-drops-fire-and-forget-actions

The fix modifies getActions() to include all pending actions alongside the continue-as-new completion action. A new unit test verifies that sendEvent actions scheduled before continueAsNew are preserved and delivered.

Summary

What changed?

Why is this change needed?

Issues / work items

• Resolves #
• Related #

---

Project checklist

• Release notes are not required for the next release
  • Otherwise: Notes added to CHANGELOG.md
• Backport is not required
  • Otherwise: Backport tracked by issue/PR #issue_or_pr
• All required tests have been added/updated (unit tests, E2E tests)
• Breaking change?
  • If yes:
    • Impact:
    • Migration guidance:

---

AI-assisted code disclosure (required)

Was an AI tool used? (select one)

• No
• Yes, AI helped write parts of this PR (e.g., GitHub Copilot)
• Yes, an AI agent generated most of this PR

If AI was used:

• Tool(s):
• AI-assisted areas/files:
• What you changed after AI output:

AI verification (required if AI was used):

• I understand the code and can explain it
• I verified referenced APIs/types exist and are correct
• I reviewed edge cases/failure paths (timeouts, retries, cancellation, exceptions)
• I reviewed concurrency/async behavior
• I checked for unintended breaking or behavior changes

---

Testing

Automated tests

• Result: Passed / Failed (link logs if failed)

Manual validation (only if runtime/behavior changed)

• Environment (OS, Node.js version, components):
• Steps + observed results:
  1.
  2.
  3.
• Evidence (optional):

---

Notes for reviewers

• N/A

[Copilot]

Pull request overview

Fixes a core orchestration runtime behavior where continueAsNew() previously caused pending fire-and-forget actions (e.g., sendEvent, signalEntity) to be dropped by getActions(). This aligns the continue-as-new completion path with the existing setComplete() / setFailed() behavior in the Durable Task JS SDK.

Changes:

• Update RuntimeOrchestrationContext.getActions() to return all pending actions plus the continue-as-new completion action.
• Add an in-memory backend regression test validating that sendEvent() is preserved across continueAsNew().

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
packages/durabletask-js/src/worker/runtime-orchestration-context.ts	Ensure continue-as-new returns pending actions (fire-and-forget) alongside the completion action.
packages/durabletask-js/test/in-memory-backend.spec.ts	Add regression coverage verifying sendEvent is not dropped when continuing-as-new.