DanielLeens commented on PR #11829: URL: https://github.com/apache/seatunnel/pull/11829#issuecomment-5379751595
@goutamadwant Good question — and to be direct: this is ultimately your and the maintainers' call, not something I can dictate as a reviewer. But since you're asking, here's the concrete evidence I have to decide with. **What I already said in my last review** (Issue 1): either resolution is acceptable — (a) move the discussion back to the STIP-36 issue only, or (b) relocate the design content to a less end-user-facing part of `docs/`, gated with a clearer "proposal, not implemented" banner. The problem flagged by both nzw921rx and me was never the content (I verified it line-by-line against `MultiTableSinkWriter`, `TransformFlowLifeCycle`, `TableEvent`, etc. and found it accurate) — it was the current location: `docs/en(zh)/introduction/configuration/`, directly linked from the live `schema-evolution.md` page that users read to configure an already-shipped feature. **New data point since then**: I checked how a directly comparable case was actually resolved on this repo. #11734 (STIP-33, "Add task failure history design" — same author, same pattern: an unimplemented-feature design doc with an "It does not describe an implemented API yet" disclaimer) placed its doc at `docs/en/engines/zeta/task-failure-history.md` — an engine-feature-docs folder, not `introduction/configuration/` — and I did not raise a placement objection across the review rounds there; I approved it in that location. So the working precedent right now is: a STIP design doc *can* live in the docs tree, it just needs to sit somewhere that isn't the actionable end-user configuration reference. That's what makes `introduction/configuration/` specifically risky here, not `docs/` in general. I also checked: there is no `STIP`-named directory or convention anywhere under `docs/` on `dev` — every STIP today lives purely as a labeled GitHub issue. So "keep the design entirely in the issue" (nzw921rx's original suggestion) is also a legitimate, more conservative option; it's just not the only path other STIPs on this repo have actually taken. Concretely, my recommendation (not a mandate): keep this PR, move `schema-evolution-timeline-design.md` out of `introduction/configuration/` into something like `docs/en/architecture/` or an engine-docs location mirroring #11734 (e.g. `docs/en/engines/zeta/`), keep STIP-36 (#11790) as the source-of-truth discussion, and resolve the dev conflict on top of that relocation. If you'd rather fold everything into the issue and close this PR instead, that's equally consistent with repo precedent — that call belongs to you and the maintainers, not to me. -- 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]
