yyanyy opened a new pull request, #19034: URL: https://github.com/apache/iceberg/pull/19034
Alternative to #18676 that models `VectorType` as a `NestedType` without an element field or element ID. Element type and dimension remain type parameters, and schema JSON stays `vector(E, D)`. `Nested-only:` comments identify the extra handling for visitors, whole-vector projection, count statistics and metrics eligibility, compact serialization, and schema compatibility/union-by-name merging. [The final commit](https://github.com/yyanyy/iceberg/commit/9d5dc97e8932d119fbba22002c21d9c3aaea5126) isolates the changes relative to the primitive implementation in #18676. This draft is for comparing the two designs; data-file I/O remains outside its scope, as in #18676. Test plan: verify that vectors expose no synthetic element fields, IDs, names, or parent mappings; schema rewrites and projection preserve complete vectors; compatibility and union-by-name reject dimension/type changes; and serialization, statistics, and metrics retain vector semantics. Validation: the full API suite (1,315 tests, 1 skipped), 259 relevant Core tests (7 skipped), and API/Core `spotlessCheck`, `checkstyleMain`, and `checkstyleTest` pass. --- **AI Disclosure** - Model: GPT-6 - Platform/Tool: Codex - Human Oversight: unreviewed - Prompt Summary: Implement the nested alternative on top of #18676 without element IDs, annotate the extra integration work compared with primitive vectors, and add regression tests. The nested implementation, comparison comments, and new tests were generated by Codex. -- 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]
