benibus commented on PR #14552: URL: https://github.com/apache/arrow/pull/14552#issuecomment-1297341786
So, I ended up changing my approach from the original PR. In this version, field ordering/omission is [optionally] handled by the generators in `test_common.h`, as this allows it to be recursively applied to nested objects. It also makes the interaction between null value probability and missing field probability more robust. As for the benchmarks, I added combinations for input with/without a schema primarily because it's an easy way to tell if ordered field optimizations are working properly for inferred schemas (plus, there does seem to be a performance difference as is). FWIW, I'm not entirely convinced that testing both possibilities for sparse fields is necessary. -- 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]
