[
https://issues.apache.org/jira/browse/ARROW-13247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17373366#comment-17373366
]
Joris Van den Bossche commented on ARROW-13247:
-----------------------------------------------
Yeah, I was just typing a comment that this could also be done with Cast, and
that this actually already works today:
{code}
In [5]: pa.array([1, 2, 3], pa.timestamp("us",
"Europe/Brussels")).cast(pa.timestamp("us", "Europe/London"))
Out[5]:
<pyarrow.lib.TimestampArray object at 0x7f7558ca3340>
[
1970-01-01 00:00:00.000001,
1970-01-01 00:00:00.000002,
1970-01-01 00:00:00.000003
]
In [6]: _.type
Out[6]: TimestampType(timestamp[us, tz=Europe/London])
{code}
However, a drawback of cast is that you need to know/specify the unit of your
input timestamp (assuming you typically want to preserve the unit for this
operation), which makes this more verbose to express, and difficult to do in a
generic query execution / expression interface where you might not necessarily
know the exact unit when defining the operation.
> [C++] Kernel to convert timestamp with timezone to another timezone
> (metadata-only change)
> ------------------------------------------------------------------------------------------
>
> Key: ARROW-13247
> URL: https://issues.apache.org/jira/browse/ARROW-13247
> Project: Apache Arrow
> Issue Type: Improvement
> Reporter: Joris Van den Bossche
> Priority: Major
> Labels: timestamp, timezone
>
> Related to ARROW-13033, but assuming you already have a tz-aware "timestamp
> with timezone" type, and you want to change it to have another timezone.
> In practice, this is a trivial, metadata-only change since in both cases the
> actual values are stored in UTC, and so it's only changing the tz
> parametrization of the timestamp type. But, for convenience, I think it would
> still be useful to have it as an actual kernel to easily integrate this
> operation in this framework.
> References:
> [tz_convert|https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DatetimeIndex.tz_convert.html]
> in pandas, or
> [with_tz|https://lubridate.tidyverse.org/reference/with_tz.html] in R's
> lubridate package,{{withZone}} in Joda or
> [withZoneSameInstant|https://docs.oracle.com/javase/8/docs/api/java/time/ZonedDateTime.html#withZoneSameInstant-java.time.ZoneId-]
> in Java
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
