[
https://issues.apache.org/jira/browse/ARROW-6290?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16911306#comment-16911306
]
Andy Grove commented on ARROW-6290:
-----------------------------------
Thanks for the bug report. What should be happening here is that the query
optimizer should be automatically inserting the CAST operation. This example
runs as part of CI so I'm surprised this regression has crept in.
> [Rust] [DataFusion] sql_csv example errors when running
> -------------------------------------------------------
>
> Key: ARROW-6290
> URL: https://issues.apache.org/jira/browse/ARROW-6290
> Project: Apache Arrow
> Issue Type: Bug
> Components: Rust, Rust - DataFusion
> Reporter: Andrew Schoenberger
> Priority: Minor
>
> The current version of
> [https://github.com/apache/arrow/blob/master/rust/datafusion/examples/csv_sql.rs]
> errors when running it.
> {code:java}
> thread 'main' panicked at 'called `Result::unwrap()` on an `Err` value:
> ExecutionError("Cannot compare Float32 with Float64")',
> src/libcore/result.rs:1084:5{code}
> I believe that this is due to column c11 being declared as DataType::Float32
> while the WHERE clause parses the numbers as DataType::Float64. Since the
> comparison_ops only matches when the parsed data types for left and right are
> the same, this errors out.
> I was able to get this running locally through either of the following two
> methods:
> (1) by changing the schema for c11 to be a DataType::Float64.
> (2) by forcing a compute::cast in the case that the left data type is
> different from the right data type in the BinaryExpr (this is likely not the
> best long-term fix, but it works for this case).
> It is also possible that this issue could be due to some mistake or incorrect
> assumption I am making when running it on my end (I am pretty new to the
> project).
>
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
