tustvold edited a comment on issue #1108: URL: https://github.com/apache/arrow-rs/issues/1108#issuecomment-1014309178
Yes it should be largely just a case of adding `_binary` versions of the `_utf8` kernels in that module. Typically there are three variants of each operator: * One with fully qualified array types e.g. `lt_eq_utf8` * One with a fully quality array type and a scalar e.g. `lt_eq_utf8_scalar` * One with a dynamic array type and qualified scalar e.g. `lt_eq_dyn_utf8_scalar` Finally there are then two dynamic dispatch functions for each operator * One that takes two dynamic array types - e.g. `lt_eq_dyn` (the dispatch logic is macro-ified) * One that takes a dynamic array and a scalar - e.g. `lt_eq_dyn_scalar` (this doesn't support non-primitive types AFAICT) I'd perhaps recommend starting with the fully qualified kernels and build up from there with successive PRs. The scalar variants are also optimisations and so could definitely be ignored for an initial cut, they were only added in the last month :grinning: -- 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]
