jaideeppyne commented on PR #10835: URL: https://github.com/apache/arrow-rs/pull/10835#issuecomment-5446878782
Added a test that round-trips through `to_ffi`/`from_ffi` instead of hand-building the ArrayData. I checked the layout against pyarrow 25.0.1: exporting `arr.slice(1, 5)` of a 6-element struct gives `length=5 offset=1` on the parent and `length=6 offset=0` on the child, and the test asserts the exported `FFI_ArrowArray` matches that before importing and slicing it. It panics without the fix. On the IPC question, it doesn't happen there. There's no per-array offset in the IPC format, so the writer applies the parent offset and the reader always gives back offset 0 with truncated children, which is why this only shows up over the C data interface. -- 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]
