jorisvandenbossche commented on pull request #7604: URL: https://github.com/apache/arrow/pull/7604#issuecomment-653226285
> But if the timestamp is meant for display purposes, I would expect the tz-naive datetime value to reflect the local timezone. Since it displays a "calendar" like object. This isn't clear cut case just stating my expectations. OK, that's indeed what was causing the confusion ;) So my view on this: the datetime.datetime object is also a tz-naive object, so we should "just" convert to the exact same tz-naive timestamp value. This is not only for display, but an actual value to work with when converting the Struct type to native python objects (eg dicts of datetime.datetime values). We already do such conversions right now, and we always preserve the actual timestamp value (and never adjust for system timezone). And for display: datetime.datetime does *not* edit the displayed value related to the system timezone. Personally I like this behaviour, while you would prefer that it would use system timezone. But once we convert to datetime.datetime, it's their display that we use, so we don't really have control over it, although you may not like that. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org