manuzhang opened a new pull request, #1120: URL: https://github.com/apache/iceberg-cpp/pull/1120
Snapshot-backed metadata views need to retain their source metadata and FileIO, derive schemas from captured metadata, and reuse partition and manifest-reading logic. Add the common support needed by the focused FilesTable (#876), ManifestsTable, and PartitionsTable follow-ups. The scan builder obtains its schema from the captured metadata, and `schema()` returns shared ownership by value. Generic snapshot tasks retain metadata, projection, FileIO, and a row-reader callback; the task reader opens them without depending on concrete table classes. Shared helpers resolve snapshot IDs/references/timestamps, unify partition specs, project partition values, and read live files one manifest at a time. The fallible row-stream overload validates its schema, source, callback, and batch size before exporting a C stream. Errors and exhaustion are terminal. Shared utilities are exported for both static and shared library consumers. The common scan fixture and focused regressions cover input errors, lazy batching, repeated failures, task ownership/reopening, projection validation, and partition evolution. Validation: all 61 metadata tests and the C++20 compatibility test passed in Debug for both static and shared library builds with Arrow/Parquet 23.0.1. Pre-commit and diff checks passed. The complete split composition also passed 132 tests with one expected format-version skip, plus C++20 compatibility. -- 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]
