bug: [v4] `batchTriggerAndWait` cached values are returned as success, no matter what their original value was

[cachho]

Provide environment information

System:
OS: Linux 5.15 Ubuntu 20.04.6 LTS (Focal Fossa)
CPU: (12) x64 AMD Ryzen 5 5600X 6-Core Processor
Memory: 6.49 GB / 15.59 GB
Container: Yes
Shell: 3.7.1 - /usr/bin/fish
Binaries:
Node: 18.20.8 - ~/.local/share/nvm/v18.20.8/bin/node
npm: 10.8.2 - ~/.local/share/nvm/v18.20.8/bin/npm
pnpm: 8.6.3 - ~/.local/share/pnpm/pnpm

Describe the bug

In the SDK, these runs don't appear as child runs at all.

Reproduction repo

not available

To reproduce

not at the moment

Additional information

How can I reliably show my users, how many of their sub tasks failed, while in progress/before I get the final output? Without idempotency I can check the status of the run.relatedRuns.children, but tasks that are cached do not appear in this array.

[cachho]

as far as I can tell, the run object returned by runs.retrieve contains no info about any cached child runs. So there's no way to show progress until it's completed and you can validate the output.

[matt-aitken]

@cachho the actual data you get back in your run should be correct I think, success or failure. But on the left side of the run page it's green when it should be red (these are "logs" and are a separate system than the operational data on the right). We can easily fix this.

Are you seeing the actual result data that you use in your run coming back incorrectly?

[matt-aitken]

@cachho we now don't cache failed runs when you use an idempotency key. It doesn't make sense for 99% of cases.

If you do want to have a "failed" result from a run you should return a Result type.