alamb commented on issue #686:
URL:
https://github.com/apache/arrow-datafusion/issues/686#issuecomment-884524231
> Oh, I see -- this is existing behavior. But then, does existing behavior
define a different result for row 1?
on master, if you execute `to_timestamp('1970-01-01')` on a server running
`DataFusion` in a UTC-5 timezone, you get `5*3600000000000`
> In my view, if the existing behavior has to stay, it has to stay;
I don't think it has to say
I think the choice of where to translate a local time into one with a
specific timezone is probably best left to a higher level than DataFusion.
DataFusion isn't a database system -- it does not have the notion of a client
or server. It is more like a tool kit with which to build such things.
> Note that we're up to four timezones in every call to TO_TIMESTAMP():
I think we have three timezones that are important:
1. The input string timezone offset (may not exist)
2. The local timezone on which datafusion is run (used in the case the user
does not supply a timezone)
2. The target timezone of the arrow type
--
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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
