You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@arrow.apache.org by "Bernhard J. (Jira)" <ji...@apache.org> on 2021/05/23 16:59:00 UTC
[jira] [Comment Edited] (ARROW-7242) [C++] Coerce Arrow
half-precision float to float32
[ https://issues.apache.org/jira/browse/ARROW-7242?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17350081#comment-17350081 ]
Bernhard J. edited comment on ARROW-7242 at 5/23/21, 4:58 PM:
--------------------------------------------------------------
Would be really nice if we could extend the parquet spec to support half-precision floats instead of float32 coercion.
Any reason why this is not intended?
was (Author: häussler):
Would be really nice if we could extend the parquet spec to support half-precision floats instead of float32 coercion.
> [C++] Coerce Arrow half-precision float to float32
> --------------------------------------------------
>
> Key: ARROW-7242
> URL: https://issues.apache.org/jira/browse/ARROW-7242
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++, Python
> Reporter: Naor Volkovich
> Priority: Major
>
> When trying to save a Pandas DF using "to_parquet" when that DF has a column with a dtype of float16, I get the error: "pyarrow.lib.ArrowNotImplementedError: Unhandled type for Arrow to Parquet schema conversion: halffloat"
--
This message was sent by Atlassian Jira
(v8.3.4#803005)