h-vetinari opened a new issue, #44455: URL: https://github.com/apache/arrow/issues/44455
### Describe the bug, including details regarding any error messages, version, and platform. While [testing](https://github.com/conda-forge/arrow-cpp-feedstock/pull/1432) arrow 18.0.0rc0, I'm getting a new batch of failing tests on windows ``` FAILED pyarrow/tests/test_compute.py::test_strftime - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'CET': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_extract_datetime_components - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'UTC': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_assume_timezone - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'UTC': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_round_temporal[nanosecond] - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'Asia/Kolkata': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_round_temporal[microsecond] - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'Asia/Kolkata': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_round_temporal[millisecond] - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'Asia/Kolkata': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_round_temporal[second] - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'Asia/Kolkata': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_round_temporal[minute] - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'Asia/Kolkata': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_round_temporal[hour] - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'Asia/Kolkata': oops: bad dow name: il FAILED pyarrow/tests/test_compute.py::test_round_temporal[day] - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'Asia/Kolkata': oops: bad dow name: il FAILED pyarrow/tests/test_convert_builtin.py::test_sequence_timestamp_from_int_with_unit - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'UTC': oops: bad dow name: il FAILED pyarrow/tests/test_scalars.py::test_timestamp_scalar - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'UTC': oops: bad dow name: il FAILED pyarrow/tests/test_scalars.py::test_cast_timestamp_to_string - pyarrow.lib.ArrowInvalid: Cannot locate timezone 'UTC': oops: bad dow name: il ``` The "cannot locate timezone" seems to be coming from https://github.com/apache/arrow/blob/16ad97938f48c9a12833f61f7553fbcf8dbaf9ca/cpp/src/arrow/compute/kernels/temporal_internal.h#L51 and hasn't changed recently. In particular, my understanding would be that https://github.com/apache/arrow/blob/16ad97938f48c9a12833f61f7553fbcf8dbaf9ca/cpp/src/arrow/compute/kernels/temporal_internal.h#L20 doesn't yet refer to C++20's specification that includes the timezones (but rather some [vendored](https://github.com/apache/arrow/blob/main/cpp/src/arrow/vendored/datetime/tz.h) variant of https://github.com/HowardHinnant/date, c.f. https://github.com/apache/arrow/commit/641c6990cdc702c736921c78448cf3a175c97863, https://github.com/apache/arrow/commit/2bf44214bf6ce9d3e275cd52b03e178b2a478105, #29200, etc.) AFAIU, the change is most likely due to the tests getting a new/updated decorator: https://github.com/apache/arrow/blob/16ad97938f48c9a12833f61f7553fbcf8dbaf9ca/python/pyarrow/tests/test_compute.py#L2508 which is ultimately determined by https://github.com/apache/arrow/blob/16ad97938f48c9a12833f61f7553fbcf8dbaf9ca/python/pyarrow/tests/util.py#L425-L437 It's not exactly described _what_ needs to be under that path, but it does seem to get forwarded to the vendored date here https://github.com/apache/arrow/blob/16ad97938f48c9a12833f61f7553fbcf8dbaf9ca/cpp/src/arrow/config.cc#L93 which will then [iterate](https://github.com/HowardHinnant/date/blob/dd8affc6de5755e07638bf0a14382d29549d6ee9/src/tz.cpp#L3046) through the contents. As such, conda-forge (which ships its own `tzdata`) should be able to set `"PYARROW_TZDATA_PATH=%PREFIX%\share\zoneinfo"` and be done with it. I'm mainly opening this issue now to document the chasing down of rabbit holes for this, which might serve someone else (and if I'm wrong, we'll be able to discuss further as well). ### Component(s) C++, Packaging -- 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: issues-unsubscr...@arrow.apache.org.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
