SEZ9 commented on issue #11616: URL: https://github.com/apache/seatunnel/issues/11616#issuecomment-5381235515
Thanks @DanielLeens and @goutamadwant — this is exactly the discussion I was hoping for, and I agree with the consensus that has formed: - **Sub-issues:** I'll split the Stage 1 items into focused sub-issues following the proposed template (scope, code pointer, benchmark validation, compatibility notes), and keep this issue as the umbrella for roadmap discussion. - **Benchmark discipline:** adding a small held-out/paraphrased regression subset is a good safeguard against optimizing for known prompts, and I'll ask that accuracy PRs report task-level pass↔fail transitions alongside aggregate deltas. - **Self-healing (2.5):** agreed — confidence should be derived from deterministic evidence (recognized error category, allowlisted remediation, idempotency/dedup, successful revalidation), not model self-confidence, and remain strictly gated on the 2.1 security model. - **Skill Packs:** connector metadata stays the source of truth for option definitions; packs carry the semantic layer (prerequisites, topology rules, verified examples) to avoid a second option registry. @goutamadwant, you're very welcome to pick up a Stage 1 item — the Doris/StarRocks option pack (1.1) or the PostgreSQL-CDC prerequisites pack (1.3) would be great starting points, with the shape DanielLeens outlined (small reviewable change, loading pointer, before/after benchmark evidence, note on which tasks moved). Comment on the sub-issue once it's up to claim it. <!-- streview-comment:443 --> -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
