dwsmith1983 commented on PR #5867: URL: https://github.com/apache/datafusion-comet/pull/5867#issuecomment-5800016194
> Could those use `-0.0D` instead, and `CAST(-0.0D AS FLOAT)` for the float case, so the fixtures exercise the normalization they describe? Done. The four lookups now use `-0.0D`, with `CAST(-0.0D AS FLOAT)` for the float key and `array(-0.0D)` for the array key, and the comments say why the suffix matters. `typeof(-0.0)` is `decimal(1,1)` and `CAST(-0.0 AS DOUBLE)` renders as `0.0`, while `-0.0D` renders as `-0.0`. The old form was hiding the answer check as well as the comment. With the float decline in `MapKeySupport` switched off, the old queries still returned 7 on the native path and only the dispatch assertion caught them. The new queries return null natively against Spark's 7 for both `element_at` and `m[...]`, so the fixtures now fail on results when the guard is missing. Spark itself returns 7 for both forms, so the expected answers are unchanged. The other map fixtures and the aggregate fixtures already use `-0.0D`, and the signed zero tip in the contributor guide now names that form too. -- 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]
