andygrove commented on PR #6354: URL: https://github.com/apache/datafusion-comet/pull/6354#issuecomment-5997662278
Thanks @sunchao, @parthchandra and @comphead. You're right about the date-level gap. `resolve_local_datetime` shifts midnight by the length of the gap, which only matches `LocalDate.atStartOfDay` when the gap starts exactly at midnight, so Toronto's 1919-03-31 comes out as 01:00 instead of 00:30. sam-1112's #5956 rewrites the same kernel for literal formats. It already resolves a date-level gap to the transition instant and takes the earlier midnight in an overlap, with Toronto, Asuncion and Havana tests. Rather than fix the same thing twice and keep the two PRs conflicting, I'd like #5956 to land first. After it, the only path that still panics on DST transitions is the per-row format path (`timestamp_trunc_array_fmt_helper`, through `as_micros_from_unix_epoch_utc`). I'll then cut this PR down to that path, reuse the helpers from #5956, and pick up comphead's suggestions there. I'm moving this to draft until then. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
