[
https://issues.apache.org/jira/browse/PARQUET-1545?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16796229#comment-16796229
]
Tim Swast commented on PARQUET-1545:
------------------------------------
Thanks [~nkollar]. I see now in the [logical types
specification|https://github.com/apache/parquet-format/blob/master/LogicalTypes.md]
that the {{isAdjustedToUTC}} parameter is even documented.
This issue can be closed. I'll dig deeper to see why this isn't being set when
I convert from pandas DataFrame -> Arrow -> Parquet in the Python code I'm
using.
> Logical Type for timezone-naive timestamps
> ------------------------------------------
>
> Key: PARQUET-1545
> URL: https://issues.apache.org/jira/browse/PARQUET-1545
> Project: Parquet
> Issue Type: New Feature
> Reporter: Tim Swast`
> Priority: Major
>
> {{In many systems there is a difference between a timezone-naive timestamp
> column (called DATETIME in BigQuery, 'logicalType': 'datetime' in Avro) and a
> timezone-aware timestamp (called TIMESTAMP in BigQuery and always stored in
> UTC). It seems from [this
> discussion|https://github.com/apache/parquet-format/pull/51#discussion_r119911623]
> and the [list of logical
> types|https://github.com/apache/parquet-format/blob/master/LogicalTypes.md]
> that parquet only has the timezone-aware version, as all timestamps are
> stored according to UTC.}}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
