[
https://issues.apache.org/jira/browse/ARROW-5712?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16871673#comment-16871673
]
TP Boudreau commented on ARROW-5712:
------------------------------------
Sorry, was AFK unit just now – I think you've figured it out – UTC for time
since midnight seems nonsensical.
It may be that we can just change the UTC flag to true here (and below for
TIME64 types), declaring that Arrow times types are UTC adjusted, and you'll
get TIME_MILLIS / TIME_MICROS for converted types and type preserving
roundtrips:
[https://github.com/apache/arrow/blob/master/cpp/src/parquet/arrow/schema.cc#L706]
(I don't have any notes on why I thought "false" was correct for this type.)
Arrow TIME64::NANOS will have to be dealt with differently; I'll have to look
at the code to make a recommendation on that. I'll also test the change I
mentioned above to see if there's anything I'm missing/forgetting.
> [C++][Parquet] Arrow time32/time64/timestamp ConvertedType not being restored
> properly
> --------------------------------------------------------------------------------------
>
> Key: ARROW-5712
> URL: https://issues.apache.org/jira/browse/ARROW-5712
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++, Python
> Reporter: Wes McKinney
> Assignee: Wes McKinney
> Priority: Major
> Fix For: 1.0.0
>
>
> These values are currently being written as raw INT32 without a logical or
> converted type set
> Example statistics for {{time32('ms')}} with {{version='2.0'}} set
> {code}
> (Pdb) stats
> <pyarrow._parquet.Statistics object at 0x7f6a9dca9f30>
> has_min_max: True
> min: 37800001
> max: 55800001
> null_count: 0
> distinct_count: 0
> num_values: 2
> converted_type: NONE
> physical_type: INT32
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
