djnavarro commented on PR #12154:
URL: https://github.com/apache/arrow/pull/12154#issuecomment-1169578666
Another bug report for you @rok, sorry 😢
It looks like the temporal rounding functions in acero only respect the
timezone when the rounding unit is `"week"` or smaller. For rounding units at
the `"month"`, `"quarter"` and `"year"` level, it doesn't
An example based on New Years in Sydney. The timestamp `ts` corresponds to
00:00:01 January 1, 2018 AEDT, which maps to the same instant as 13:00:01
December 31, 2017 UTC. Whenever I call `floor_temporal()` for a rounding unit
of a week or smaller, I get the correct answer: the boundaries are expressed in
Sydney time. The New Year in Sydney corresponds to `2017-12-31 13:00:00.000000
UTC` and that's what we get (the print method doesn't show it, but the timezone
is properly recorded in the metadata and `year()` correctly returns `"2018"`
for this instant.
For the larger rounding units, it looks to me like the UTC offset (+11 in
this case) needs to be subtracted from the stored instant? They all look
superficially like the right answer but they're all 11 hours in the future
``` r
# one second after New Year in 2018 (happens to be Monday!)
ts <- arrow::Scalar$create(as.POSIXct("2018-01-01 00:00:01",
tz="Australia/Sydney"))
# rounding units for week and below correctly respect timezone
arrow::call_function("floor_temporal", ts, options = list(unit = 7L)) # day
#> Scalar
#> 2017-12-31 13:00:00.000000
arrow::call_function("floor_temporal", ts, options = list(unit = 6L)) # day
#> Scalar
#> 2017-12-31 13:00:00.000000
arrow::call_function("floor_temporal", ts, options = list(unit = 5L)) # hour
#> Scalar
#> 2017-12-31 13:00:00.000000
arrow::call_function("floor_temporal", ts, options = list(unit = 4L)) #
minute
#> Scalar
#> 2017-12-31 13:00:00.000000
# rounding units for month and above don't
arrow::call_function("floor_temporal", ts, options = list(unit = 10L)) # year
#> Scalar
#> 2018-01-01 00:00:00.000000
arrow::call_function("floor_temporal", ts, options = list(unit = 9L)) #
quarter
#> Scalar
#> 2018-01-01 00:00:00.000000
arrow::call_function("floor_temporal", ts, options = list(unit = 8L)) #
month
#> Scalar
#> 2018-01-01 00:00:00.000000
```
<sup>Created on 2022-06-29 by the [reprex
package](https://reprex.tidyverse.org) (v2.0.1)</sup>
--
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]
