I haven't had any issues so far but also
I haven't had any issues so far but also haven't played with hibernation much so far 😅 ... but I want to know if my assumption is correct.
Rules of workflows stating that you should not depend on state define solely outside of a step, and workflows can use outputs of steps across their execution.
So... would somethint like this be OK?
2 Replies
or maybe sleep/hibernation coming last in the loop (after we update someState) 🤔
the TLDR for this is:
Love agents and the work Sunil & team has put there, but observability DX/UX of workflows is really good also, so trying to see how to combine them better for agent execution-loops 🙂
If I undestand it correctly, when a Workflow re-initializes after sleep, retries or any other reason, the promises of your
await step.do
calls that were already successfully executed would just instantly resolve.
That means that re-defining state should be fine and your workflow should work as expected!