SEZ9 commented on issue #12268:
URL: https://github.com/apache/seatunnel/issues/12268#issuecomment-5851885029

   Thanks for posting the dev-list link — that closes the first part of the 
gate from my last comment. On your direct question about D1–D6, here is where I 
land on what you summarized:
   
   - Identity as `(jobId, nodeId)` with `name` as a display label only: 
confirmed. That is the right stable identity and matches the concern CryoThrust 
raised about retries, renamed aliases, and parallel subtasks.
   - Point-in-time snapshot with no capture timestamp: mild push-back. I accept 
that `JobDAGInfo` carries none, but the contract should still make freshness 
explicit to the caller. If a server-side serialization time is not something 
you want to add in V1, the "Snapshot freshness and completeness" section should 
state that omission as a deliberate decision and the response should still 
expose whether the snapshot is complete or partial, so the 503 / 
partially-available outcome is distinguishable from a full graph.
   - Whole-response 413 `LINEAGE_GRAPH_TOO_LARGE`, no truncation or pagination 
in V1: confirmed as a V1 decision, provided it is listed in the route/error 
mapping alongside 404/409/503 and the structural/byte limits it is derived from 
are the provisional values already in the STIP.
   - Opaque dataset IDs deferred, no config-derived IDs: confirmed. This keeps 
the endpoint separate from the OpenLineage work, which is what I asked for.
   
   Two things I cannot confirm from this thread alone: the 
history-versus-active precedence for a reused ID, and the access/overlap 
boundary. Please paste the resolved wording for those here (or point to the 
exact section headings) so the record in this issue is self-contained.
   
   Remaining asks before implementation:
   1. Let the [DISCUSS] thread run and summarize its outcome here, including 
any push-back and the final D1–D6 decisions.
   2. Apply the freshness/completeness clarification above to the STIP body.
   3. No implementation PR, cache warming, DAG reconstruction, connector 
loading, or new persistent state until those decisions are recorded.
   
   <!-- streview-comment:1337 -->


-- 
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]

Reply via email to