[
https://issues.apache.org/jira/browse/ARROW-12722?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17342834#comment-17342834
]
Neal Richardson commented on ARROW-12722:
-----------------------------------------
Technically, in Arrow the names of fields in a Schema (and thus RecordBatch and
Table) don't need to be unique. We may want to raise a warning in R when
creating one, though. But I think the issue is that {{GetFieldByName}} errors
if they're not unique. So that should error clearly.
> [R] Raise error when attemping to create table with duplicated naming
> ---------------------------------------------------------------------
>
> Key: ARROW-12722
> URL: https://issues.apache.org/jira/browse/ARROW-12722
> Project: Apache Arrow
> Issue Type: Improvement
> Reporter: Nic Crane
> Assignee: Nic Crane
> Priority: Major
>
> It is currently possible to create a table (and possibly RecordBatch) with
> duplicated column names, which leads to later issues when trying to analyse
> the data - see the below example.
> An appropriate error should be raised instead of letting duplicate names to
> be used
>
> {code:java}
> > Table$create(iris, iris) %>% filter(Species == "versicolor")
> Error in schm$GetFieldByName(name)$ToString() : attempt to apply non-function
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
