raghav-reglobe commented on PR #2203: URL: https://github.com/apache/iceberg-rust/pull/2203#issuecomment-4754199824
**MoR (deletion-vector) extension to this RowDelta action — validated, branch ready** Thanks for this RowDelta foundation, @wirybeaver. I've built the **merge-on-read** half on top of this PR's head (`0d651e50`): writing V3 Puffin **deletion vectors** + a `content=Deletes` manifest (`Operation::Delete`) — the path that backs `MERGE INTO`/`UPDATE`/`DELETE` on V3 tables (it replaces the current `FeatureUnsupported` stub in `add_delete_files`). Branch (stacked on this PR's head): https://github.com/raghav-reglobe/iceberg-rust/tree/rowdelta-dv-write **Adds** - `DeleteVector::{to_puffin_blob, from_puffin_blob, write_to_puffin_file}` — serialize/deserialize `deletion-vector-v1` (magic `0xD1D33964`, BE-length + portable 64-bit roaring + BE-CRC32, `snapshot-id`/`sequence-number` = -1, `fields` = `[]`) and produce the `PositionDeletes` `DataFile`. - `RowDeltaAction.add_delete_files` MoR commit path (writes a `content=Deletes` manifest). **Validation** - **Byte-identical to Apache Iceberg-Java** — `to_puffin_blob` output equals the Java-produced golden fixtures from `apache/iceberg` core test resources (3 vectors: empty, `{1,3,5,7,9}`, and small+large spanning two roaring containers). - Cross-impl parity with `apache/iceberg-go` (`puffin/dv_golden_test.go`, `TestSerializeDV*`). - **Real data, cross-engine** — wrote a DV to a real V3 table via a REST catalog on S3; an independent engine (Doris) read it back correctly (10 → 7 rows, exact positions deleted). - 14 `delete_vector` tests + the `transaction` suite (incl. the RowDelta MoR path) pass; `fmt` + `clippy` clean. Happy to (a) fold this into this PR, (b) open a follow-up PR once this lands, or (c) help rebase this branch onto current `main` (this PR is currently flagged conflicting) so the combined CoW + MoR RowDelta can move forward. Whatever works best for you. -- 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]
