You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@calcite.apache.org by "Rui Wang (Jira)" <ji...@apache.org> on 2019/11/21 18:20:00 UTC
[jira] [Commented] (CALCITE-3530) TIMESTAMPADD/TIMESTAMPDIFF with
microsecond/nanosecond unit lost precision
[ https://issues.apache.org/jira/browse/CALCITE-3530?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16979477#comment-16979477 ]
Rui Wang commented on CALCITE-3530:
-----------------------------------
If need to make Calcite SqlFunction supports beyond millisecond, we will have to discuss how to pass around values with micros or nanos. A long won't be enough and it might be a (long, int). To support this kind of format, as to function parameters, it's relative easy. But I didn't figure it our how to return (long int) from functions.
> TIMESTAMPADD/TIMESTAMPDIFF with microsecond/nanosecond unit lost precision
> --------------------------------------------------------------------------
>
> Key: CALCITE-3530
> URL: https://issues.apache.org/jira/browse/CALCITE-3530
> Project: Calcite
> Issue Type: Bug
> Components: core
> Reporter: Zhenghua Gao
> Priority: Minor
>
> Since the TimestampAddConvertlet and TimestampDiffConvertlet tread TIMESTAMP as long (with millisecond precision), they lost precision even if the downstream can support microsecond or nanosecond.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)