Zanie Blue

Hi! This is because you are using the SQLite backend with an ephemeral API. When using the ephemeral API, each task run starts its own copy of the API server...

If you run tasks sequentially yeah you're unlikely to run into database locking issues. You could still have problems running flow runs concurrently though.

Thanks for the additional details and reproduction! I'm pretty confused about there being any overhead here, these objects should all be safely shared in memory. There must be something weird...

I believe before we can support running subflows concurrently, we will need to establish the `Flow.submit` interface #6689 — we could in theory add a `map` operator that behaves the...

Thanks so much for digging into this! To clarify, the broken case here is `prefect orion start` without `--detach` cannot be stopped by Ctrl-C on Windows? All other cases are...

:D by > Do you see a clear path to retaining the async process implementation while returning before the process exists? I mean, can we use the previous async `open_process`...

Sweet! This is coming along nicely. Regarding Windows, I'm fine with users needing Python 3.10+ to not see the error. Especially since it does not actually impact the server. I'm...

In v1 we have automations for UI driven notifications and complex event-based triggers but still allow notification on state change via state handlers in code. I think a similar pattern...

After further thought this is much easier if we require it to be a global setting rather than at the flow or task level. This is still quite challenging though...

Here's a small POC design that would allow this to apply per flow / task by patching the `print` built-in. Stdout capturing would need to be global. https://github.com/PrefectHQ/prefect/compare/poc/log-print-statement