GitHub user joeyutong added a comment to the discussion: Parallel Tool Call Execution
Thanks for the detailed follow-up. Your replies to the concerns around duplicate calls, concurrency limits, and tool-level observability all sound reasonable to me. One clarification on failure semantics: based on the later discussion, it sounds like v1 will not expose `durableExecuteAllAsync` as a public `RunnerContext` API, and the Python `tool_call_action` will be updated to match Java's per-tool try/catch behavior. With that scope, should the new parallel tool-call path preserve the existing collect-all behavior of `tool_call_action`? The earlier wording about aligning `durableExecuteAllAsync` with single-call failure semantics made this a bit unclear to me. GitHub link: https://github.com/apache/flink-agents/discussions/855#discussioncomment-17554866 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected]
