Fix warnings in CI

[Illviljan]

Noticed in #151.

=============================== warnings summary ===============================
tests/test_core.py: 22 warnings
  /home/runner/work/flox/flox/flox/aggregate_flox.py:110: RuntimeWarning: divide by zero encountered in divide
    out /= nanlen(group_idx, array, size=size, axis=axis, fill_value=0)

tests/test_core.py: 207 warnings
  /home/runner/work/flox/flox/flox/aggregations.py:250: RuntimeWarning: invalid value encountered in divide
    finalize=lambda sum_, count: sum_ / count,

tests/test_core.py: 22 warnings
  /home/runner/work/flox/flox/flox/aggregate_flox.py:118: RuntimeWarning: divide by zero encountered in divide
    out /= nanlen(group_idx, array, size=size, axis=axis, fill_value=0)

tests/test_core.py: 531 warnings
tests/test_xarray.py: 88 warnings
  /home/runner/work/flox/flox/flox/aggregations.py:259: RuntimeWarning: invalid value encountered in divide
    finalize=lambda sum_, count: sum_ / count,

tests/test_core.py: 2772 warnings
  /home/runner/work/flox/flox/flox/aggregations.py:268: RuntimeWarning: invalid value encountered in divide
    result = (sumsq - (sum_**2 / count)) / (count - ddof)

tests/test_core.py: 1068 warnings
  /home/runner/work/flox/flox/flox/core.py:877: RuntimeWarning: All-NaN slice encountered
    result = getattr(np, combine)(array, axis=axis_, keepdims=True)

tests/test_core.py: 18 warnings
  /home/runner/work/flox/flox/tests/test_core.py:209: RuntimeWarning: Mean of empty slice
    expected = getattr(np, func)(array[..., ~nanmask], axis=-1, **kwargs)

tests/test_core.py: 8 warnings
tests/test_xarray.py: 12 warnings
  /home/runner/work/flox/flox/flox/aggregate_flox.py:118: RuntimeWarning: invalid value encountered in divide
    out /= nanlen(group_idx, array, size=size, axis=axis, fill_value=0)

tests/test_core.py: 18 warnings
  /home/runner/work/flox/flox/tests/test_core.py:209: RuntimeWarning: Degrees of freedom <= 0 for slice.
    expected = getattr(np, func)(array[..., ~nanmask], axis=-1, **kwargs)

tests/test_core.py: 1944 warnings
  /home/runner/micromamba-root/envs/flox-tests/lib/python3.8/site-packages/dask/array/numpy_compat.py:41: RuntimeWarning: invalid value encountered in divide
    x = np.divide(x1, x2, out)

tests/test_core.py: 200 warnings
  /home/runner/micromamba-root/envs/flox-tests/lib/python3.8/site-packages/dask/array/reductions.py:1062: RuntimeWarning: All-NaN slice encountered
    vals = func(x, axis=arg_axis, keepdims=True)

tests/test_core.py: 18 warnings
  /home/runner/micromamba-root/envs/flox-tests/lib/python3.8/site-packages/numpy/lib/nanfunctions.py:1878: RuntimeWarning: Degrees of freedom <= 0 for slice.
    var = nanvar(a, axis=axis, dtype=dtype, out=out, ddof=ddof,

tests/test_core.py: 342 warnings
  /home/runner/micromamba-root/envs/flox-tests/lib/python3.8/site-packages/dask/array/reductions.py:640: RuntimeWarning: All-NaN slice encountered
    return np.nanmax(x_chunk, axis=axis, keepdims=keepdims)

tests/test_core.py: 36 warnings
  /home/runner/work/flox/flox/tests/test_core.py:209: RuntimeWarning: All-NaN slice encountered
    expected = getattr(np, func)(array[..., ~nanmask], axis=-1, **kwargs)

tests/test_core.py: 342 warnings
  /home/runner/micromamba-root/envs/flox-tests/lib/python3.8/site-packages/dask/array/reductions.py:611: RuntimeWarning: All-NaN slice encountered
    return np.nanmin(x_chunk, axis=axis, keepdims=keepdims)

tests/test_xarray.py: 15 warnings
  /home/runner/work/flox/flox/flox/xarray.py:571: DeprecationWarning: flox.xarray.resample_reduce is now deprecated. Please use Xarray's resample method directly.
    warnings.warn(

tests/test_xarray.py::test_multi_index_groupby_sum[flox]
tests/test_xarray.py::test_multi_index_groupby_sum[flox]
tests/test_xarray.py::test_multi_index_groupby_sum[numpy]
tests/test_xarray.py::test_multi_index_groupby_sum[numpy]
tests/test_xarray.py::test_multi_index_groupby_sum[numba]
tests/test_xarray.py::test_multi_index_groupby_sum[numba]
  /home/runner/work/flox/flox/flox/xarray.py:252: DeprecationWarning: Deleting a single level of a MultiIndex is deprecated. Previously, this deleted all levels of a MultiIndex. Please also drop the following variables: {'y', 'x'} to avoid an error in the future.
    ds = ds.drop_vars([var for var in maybe_drop if var in ds.variables])

tests/test_xarray.py::test_multi_index_groupby_sum[flox]
tests/test_xarray.py::test_multi_index_groupby_sum[numpy]
tests/test_xarray.py::test_multi_index_groupby_sum[numba]
  /home/runner/work/flox/flox/flox/xarray.py:252: DeprecationWarning: Deleting a single level of a MultiIndex is deprecated. Previously, this deleted all levels of a MultiIndex. Please also drop the following variables: {'b', 'c'} to avoid an error in the future.
    ds = ds.drop_vars([var for var in maybe_drop if var in ds.variables])

tests/test_xarray.py::test_datetime_array_reduce[flox-count-False]
tests/test_xarray.py::test_datetime_array_reduce[flox-mean-False]
tests/test_xarray.py::test_datetime_array_reduce[numpy-count-False]
tests/test_xarray.py::test_datetime_array_reduce[numpy-mean-False]
tests/test_xarray.py::test_datetime_array_reduce[numba-count-False]
tests/test_xarray.py::test_datetime_array_reduce[numba-mean-False]
  /home/runner/micromamba-root/envs/flox-tests/lib/python3.8/site-packages/xarray/coding/cftime_offsets.py:1130: FutureWarning: Argument `closed` is deprecated in favor of `inclusive`.
    return pd.date_range(

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html

[tlogan2000]

I will comment here as I think my issue is related. Please let me know if you prefer a separate issue.

I have noticed that a recent upgrade to use flox in connection with xarray and dask has caused an inability for me to silence associated numpy warnings even via a blanket approach warnings.simplefilter("ignore").

The Xarray and dask combination have always been noisy wrt to warnings but this used to work to avoid messy jupyter notebook outputs. Since moving to flox I can't seem to achieve the same result

[dcherian]

@tlogan2000 I don't know how flox would affect that.

However I have now silenced every numpy warning from inside flox. So maybe that helps.

[tlogan2000]

@tlogan2000 I don't know how flox would affect that.

However I have now silenced every numpy warning from inside flox. So maybe that helps.

Excellent, thanks it definitely seems to help but I am still getting a recurring :

aggregations.py:259: RuntimeWarning: invalid value encountered in divide
  chunk=("sum", "nanlen"),

[dcherian]

Are you using the flox engine?

[tlogan2000]

I can't really say TBH, my experience with flox to date has been to simply install and let xarray defaults handle how and when it is employed. I am one of the xclim library https://xclim.readthedocs.io/en/stable/ developers and am basically using xarray either directly or via associated xclim calculations in my data processing. In general, performance for resampling (a core procedure for a large number of xclim climate indicator calculations) and use of dask.distributed clients has seen a large boost with flox installed ... my only difficulty has been with the warnings

[dcherian]

Hmm ok in that case I'll need a reproducible example to track it down.