You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@arrow.apache.org by "David Li (Jira)" <ji...@apache.org> on 2021/06/03 20:55:00 UTC

[jira] [Commented] (ARROW-12951) [C++] Refactor StringTransform

    [ https://issues.apache.org/jira/browse/ARROW-12951?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17356711#comment-17356711 ] 

David Li commented on ARROW-12951:
----------------------------------

Is this the case? MaxCodeunits and Transform do depend on the concrete type since they use offset_type. And while you could abstract between the large and regular types, that would preclude kernels specializing on binary vs string types (though admittedly, no kernel does that).

> [C++] Refactor StringTransform
> ------------------------------
>
>                 Key: ARROW-12951
>                 URL: https://issues.apache.org/jira/browse/ARROW-12951
>             Project: Apache Arrow
>          Issue Type: Task
>          Components: C++
>            Reporter: Antoine Pitrou
>            Priority: Minor
>
> In {{compute/kernels/scalar_string.cc}, currently there is a non-trivial inheritance relationship between {{StringTransform}} and its derived implementation classes. Also, the derived classes have to be templated on the actual string type even though their code doesn't care, making generated code size probably bigger than it should be.
> Instead, there could be a {{StringTransformExec}} class that delegates to an independent class for per-kernel specifics (mostly: the {{MaxCodeunits}} and {{Transform}} methods).



--
This message was sent by Atlassian Jira
(v8.3.4#803005)