[jira] [Commented] (ARROW-8282) [C++/Python][Dataset] Support schema evolution for integer columns

["Joris Van den Bossche (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ARROW-8282?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17247513#comment-17247513 ] 

Joris Van den Bossche commented on ARROW-8282:
----------------------------------------------

Do we need a separate Fragment type? We could also do the cast when scanning (eg we already do some edits at that point, like projection, adding null columns, etc)

cc [~bkietz]

> [C++/Python][Dataset] Support schema evolution for integer columns
> ------------------------------------------------------------------
>
>                 Key: ARROW-8282
>                 URL: https://issues.apache.org/jira/browse/ARROW-8282
>             Project: Apache Arrow
>          Issue Type: Improvement
>          Components: C++
>            Reporter: Uwe Korn
>            Priority: Major
>              Labels: dataset
>             Fix For: 3.0.0
>
>
> When reading in a dataset where the schema specifies that column X is of type {{int64}} but the partition actually contains the data stored in that columns as {{int32}}, an upcast should be done.



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