sfc-gh-gfrere commented on issue #43252: URL: https://github.com/apache/arrow/issues/43252#issuecomment-2822965772
Hi @khwilson — are you referring to this check? https://github.com/apache/arrow/blob/0cc4855de73777524f1419a7e11ab15ba2abd645/cpp/src/gandiva/decimal_type_util.cc#L59. This is intentional to safely account for carry in the product. Or were you meaning something else? Not exactly your question, but has there been discussion around allowing higher-precision products/quotients? For example, allowing decimal128(p=38) operands to yield higher-precision results (when explicitly permitted), even if no further PyArrow operations are supported on those results. Clearly this is not safe and strict type abiding which I assume is why it has not been implemented. The idea would be: as long as input operands are within spec, allow pc.multiply / pc.divide to produce out-of-bound precision — while attempting some safety by restricting follow-up operations or requiring user-managed handling. This avoids forced pre-casting operands which potentiates precision loss. -- 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: github-unsubscr...@arrow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
