jayzhan211 commented on issue #8302: URL: https://github.com/apache/arrow-datafusion/issues/8302#issuecomment-1832926920
> > https://github.com/apache/arrow-datafusion/blob/2a692446f46ef96f48eb9ba19231e9576be9ff5a/datafusion/expr/src/type_coercion/binary.rs#L388-L406 > > > > In `comparison_binary_numeric_coercion`, signed type and u64 is coerced to i64, anyone knows why is this acceptable? what kind of cases are? > > for type u64 and f32, it will return f32 which is smaller than u64, not sure why is this too > > The comment in the code you linked seems to say "some information loss is inevitable": > > ``` > // The following match arms encode the following logic: Given the two > // integral types, we choose the narrowest possible integral type that > // accommodates all values of both types. Note that some information > // loss is inevitable when we have a signed type and a `UInt64`, in > // which case we use `Int64`;i.e. the widest signed integral type. > ``` > > It is likely for "usability" so that you can compare i64 and u64 without an explicit type cast 🤷 I don't think that usability is the good reason :( It might cause data loss. @alamb I would like to change them to the straightforward version, choose the larger type between two. I think it can also be used all the place in datafusion easily (one place is `can_coerce_from`). Is there any concern about this change? -- 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]
