[akleshchev]

Looks like new chat service dies on a (planned) exception before the work it posted via runStorage completes.

I'm not sure if I should be changing WaitForResult to not crash or changing chat service to not block inside a finicky coroutine or both. Blocking is a poor choice for a service that does not control it's own lifetime.

[Copilot]

🔵 Needs a closer look

Cross-thread completion after coroutine teardown lacks regression coverage and needs human validation.

1 open issue

• Add regression test for late work completion after coroutine unwind · New

What changed in this PR

This PR addresses dangling promises when queued work outlives a waiting coroutine and stops chat-history processing during application errors.

Changes:

• Share promise ownership with queued callbacks for value and void results.
• Exit the chat-history manager loop when the application reports an error.

File	Description
indra/​newview/​llchatservicehistory.cpp	Stops manager processing on application errors.
indra/​llcommon/​workqueue.h	Keeps promises alive through queued callback completion.

_{Review effort: Balanced}

[akleshchev]

I guess I will just not use waitForResult and poll instead.

[Copilot]

🟡 Changes recommended

Three unresolved moderate issues affect shutdown safety, object lifetime, and queue-failure handling.

3 open issues

• Queued work can dereference dangling stack references after logout · New
• Manager wait loop can hang during error shutdown · New
• Add regression test for late work completion after coroutine unwind

_{Review effort: Lite}

[github-actions]

Thank you for your submission, we really appreciate it. Like many open-source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution. You can sign the CLA by just posting a Pull Request Comment same as the below format.

---

I have read the CLA Document and I hereby sign the CLA

---

1 out of 2 committers have signed the CLA.
✅ (akleshchev)[https://github.com/akleshchev]
❌ @copilot
_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[Copilot]

🔵 Needs a closer look

Address storage-operation generation fencing and move the queued callable to avoid copying large work data.

No open issues 🎉

3 resolved since last review

• Manager wait loop can hang during error shutdown
• Queued work can dereference dangling stack references after logout
• Add regression test for late work completion after coroutine unwind

_{Review effort: Lite}

[Copilot]

🔵 Needs a closer look

Cross-thread lifetime and logout ordering need human validation, and unnecessary copying of large queued payloads remains unresolved.

1 open issue

• Avoid copying staged data when capturing publication work · New

_{Review effort: Balanced}