mikamikasuki opened a new pull request, #55245: URL: https://github.com/apache/arrow/pull/55245
### Rationale for this change An exported TableBatchReader can outlive the temporary Ruby Table used to create it. The reader currently holds a reference to the Table, so importing the stream after Ruby garbage collection can segfault. ### What changes are included in this PR? Construct TableBatchReader with its shared_ptr<Table> so it retains the source table for the reader’s lifetime. ### Are these changes tested? Added a Ruby regression that exports a reader from a temporary table, runs GC, imports the stream, and reads all rows. The regression segfaults on current main and passes with this change. The full record-batch-reader test file passes (4 tests, 4 assertions). The Arrow C++ core, GLib bindings, and Ruby extension build succeeded. ### Are there any user-facing changes? This prevents a segmentation fault when an exported TableBatchReader is used after its source Ruby table wrapper is garbage-collected. **This PR contains a "Critical Fix".** It prevents a crash when the source table is collected before the exported reader stream is used. ### Was AI used for this PR? OpenAI Codex assisted with root-cause analysis, implementation, and regression-test generation. The contributor reviewed the code and regression output. **PR code and description written by:** - [ ] Human - [x] AI **Reviewed before submission by:** - [x] Human - [ ] AI - [ ] Not reviewed Closes #54689 -- 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]
