manuzhang opened a new pull request, #17299: URL: https://github.com/apache/iceberg/pull/17299
## Summary - add optional `data-sequence-number` to REST-planned data files - preserve manifest-entry data sequence numbers for Java data files while leaving delete-file JSON and file sequence numbers unchanged - add content-file, scan-task, and REST scan-response regression coverage ## Motivation REST-planned scan tasks currently omit manifest-entry data sequence numbers. Row-lineage clients need the data file value to inherit a missing or null `_last_updated_sequence_number` and must not substitute a file or snapshot sequence number. The field remains optional for rolling compatibility. Servers SHOULD include it when known for data files with a non-null `first-row-id`, while clients MUST NOT substitute a file or snapshot sequence number when it is absent. Scoping and emitting the new field only for row-lineage data files avoids changing ordinary data-file and delete-file payloads that do not need it for lineage inheritance. This incorporates feedback from the broader content-file approach in https://github.com/apache/iceberg/pull/17234 by @wgtmac. ## Compatibility - New servers add an optional field only to row-lineage data-file payloads with non-null `first-row-id`; old clients are expected to ignore unknown fields. - New clients continue to parse responses from old servers that omit the field and report unsupported lineage synthesis when the value is required. - Delete-file JSON remains unchanged. --- **AI Disclosure** - Model: GPT-5 - Platform/Tool: Codex - Human Oversight: reviewed - Prompt Summary: Add backward-compatible REST data sequence number support for row-lineage data files. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
