sm-Fifteen commented on issue #36558: URL: https://github.com/apache/arrow/issues/36558#issuecomment-1802476715
> > > > I don't think Arrow (nor any dataframe library, as far as I'm aware) supports columns with mixed timezones (barring `object` dtype in pandas, which can store anything, but then time series algorithms don't work on it) @MarcoGorelli : Oh, obvously working with mixed timezone or mixed offset series is going to prove pretty much unworkable for most uses if you don't normalize it first, and I'm also pretty sure Arrow doesn't have a column format that could handle mixed zone timestamps. I'm just saying that It's not a problem that's exclusive to fixed time offsets, that the problem can pose itself even if you're working with IANA zones (such as a vehicle moving across timezones while measuring something, and the logs adjusting the zone consequently) and that it's not a good reason for Arrow to not support offset timezones. If you're dealing with a column that has a uniform, fixed offset time, that shouldn't be forbidden. Mixed offsets and/or mixed timezones are, I believe, a separate problem. In fact, I would go as far as to suggest Arrow might want to add a fourth timestamp format, which would see the internal representation normalized to UTC (as with the other two tz-aware formats) and the tz string in the column metadata pointing to a separate column containing the timezone names or fixed offsets. -- 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. To unsubscribe, e-mail: github-unsubscr...@arrow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
