Just upgraded all of my packages through conda update --all and then got this error when importing xarray:

```python Traceback (most recent call last): File "/var/spool/torque/mom_priv/jobs/63530.rdf-xcat.SC", line 11, in <module> import xarray as xr File "/nerc/n02/n02/amethyst/.conda/envs/my_root/lib/python3.4/site-packages/xarray/init.py", line 10, in <module> from .core.alignment import align, broadcast, broadcast_arrays File "/nerc/n02/n02/amethyst/.conda/envs/my_root/lib/python3.4/site-packages/xarray/core/alignment.py", line 11, in <module> from .indexing import get_indexer_nd File "/nerc/n02/n02/amethyst/.conda/envs/my_root/lib/python3.4/site-packages/xarray/core/indexing.py", line 11, in <module> from . import duck_array_ops, nputils, utils File "/nerc/n02/n02/amethyst/.conda/envs/my_root/lib/python3.4/site-packages/xarray/core/duck_array_ops.py", line 22, in <module> from . import dask_array_compat File "/nerc/n02/n02/amethyst/.conda/envs/my_root/lib/python3.4/site-packages/xarray/core/dask_array_compat.py", line 47, in <module> AxisError = np.AxisError AttributeError: 'module' object has no attribute 'AxisError'

```

May be relevant, when I do conda list, I get two versions of xarray: xarray 0.10.9 <pip> xarray 0.8.2 py34_0

Problem description

My dataset has lots of days, with data for two times for each day: midnight and midday (12am and 12pm). With date[0], I am trying to retrieve only the midnight data from the first day; with date[1], the midday data from the same day; date[3] the midnight data from the next day, etc. Now I'm writing this, I'm not sure I've posted this in the right place so please let me know if I haven't.

Code Sample 1: this gives data for midnight and midday (time: 2)

```python from datetime import datetime, timedelta start = datetime(1979, 1, 1) date = [start.strftime('%Y%m%d')] a = first_dataset.loc[date[0]] print(a)

<xarray.DataArray (time: 2, latitude: 241, longitude: 480)> array([[[ #there are numbers here, cut out for ease of reading ]]]) Coordinates: * longitude (longitude) float32 0.0 0.75 1.5 2.25 3.0 3.75 4.5 5.25 6.0 ... * latitude (latitude) float32 90.0 89.25 88.5 87.75 87.0 86.25 85.5 ... * time (time) datetime64[ns] 1979-01-01 1979-01-01 T12:00:00 #as you can see here, for

midnight the time is just 1979-01-01 instead of 1979-01-01 00:00:00, which I think might be causing

some confusion in my code

```

Code Sample 2: try adding hour to the date

```python from datetime import datetime, timedelta start = datetime(1979, 1, 1) date = [start.strftime('%Y%m%d%H')] #this what I changed a = first_dataset.loc[date[0]] print(a)

Traceback (most recent call last): File "pandas/index.pyx", line 589, in pandas.index.DatetimeEngine.get_loc (pandas/index.c:11661) File "pandas/src/hashtable_class_helper.pxi", line 404, in pandas.hashtable.Int64HashTable.get_item (pandas/hashtable.c:8543) TypeError: an integer is required

During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/nerc/n02/n02/amethyst/.conda/envs/my_root/lib/python3.4/site-packages/pandas/indexes/base.py", line 2134, in get_loc return self._engine.get_loc(key) File "pandas/index.pyx", line 553, in pandas.index.DatetimeEngine.get_loc (pandas/index.c:11829) File "pandas/index.pyx", line 591, in pandas.index.DatetimeEngine.get_loc (pandas/index.c:11735) File "pandas/index.pyx", line 597, in pandas.index.DatetimeEngine._date_check_type (pandas/index.c:11899) KeyError: '1979010100' ```

Expected Output

```python a = first_dataset.loc[date[0]] print(a)

<xarray.DataArray (time: 1, latitude: 241, longitude: 480)> array([[[ #there are numbers here, cut out for ease of reading ]]]) Coordinates: * longitude (longitude) float32 0.0 0.75 1.5 2.25 3.0 3.75 4.5 5.25 6.0 ... * latitude (latitude) float32 90.0 89.25 88.5 87.75 87.0 86.25 85.5 ... * time (time) datetime64[ns] 1979-01-01 ```

I have a variable with time, lon and lat dimensions.

I can specify longitude (uu = u.where(u.longitude >= 30)), similarly with latitude, but I'm struggling to specify a certain date/time.

Printing the time gives the following output: ```

print u.time <xarray.DataArray 'time' (time: 8)> array(['2018-04-01T00:00:00.000000000', '2018-04-01T06:00:00.000000000', '2018-04-01T12:00:00.000000000', '2018-04-01T18:00:00.000000000', '2018-04-02T00:00:00.000000000', '2018-04-02T06:00:00.000000000', '2018-04-02T12:00:00.000000000', '2018-04-02T18:00:00.000000000'], dtype='datetime64[ns]') Coordinates: * time (time) datetime64[ns] 2018-04-01 2018-04-01T06:00:00 ... Attributes: long_name: time ```

I've tried naming the date by a number like uu = u.where(u.time >= 4) but it gives TypeError: invalid type promotion, and it's not very useful when using a dataset with a huge time dimension.

I'd just like some help figuring out the syntax for specifying days/times.