laskoviymishka opened a new issue, #1048: URL: https://github.com/apache/iceberg-go/issues/1048
Parent: #589. Puffin spec `deletion-vector-v1` requires `snapshot-id` and `sequence-number` on blob metadata to both be `-1`. The writer in `puffin.Writer.AddBlob` enforces this on the way out. The reader's `validateBlobs` does not enforce the same invariant on the way in, so a malformed DV puffin written by a third-party tool would be accepted and its blob bytes would flow into `dv.DeserializeDV` without complaint. The bitmap CRC still catches byte-level corruption, but the metadata invariant is silently violated. Extend `validateBlobs` with a per-blob check: when `Type == BlobTypeDeletionVector`, both `SnapshotID == -1` and `SequenceNumber == -1`, else return a clear error. One unit test with a hand-built puffin carrying the wrong values is sufficient; the envelope-shape fixtures used by `TestDeletionVectorPuffinWireFormat` are a good starting point. Touches `puffin/` only — no dependency on any other deletion-vector work in flight. -- 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]
