I need to get the first Δt from the start of my dataset, i.e. ds.sel(start_time, start_time + timedelta). However, due to pandas using M8[ns] but datetime.datetime not supporting this, the index gets converted to an int and indexing fails. Inspection tells me that by the time the index reaches pandas it is already an int. This is ultimately due to the numpy problem that timedelta64(0, 'ns').item() is an int, but it would be very nice if xarray had a workaround so that we can use indexing such as shown below.

``` In [282]: time = pd.date_range('2000-01-01', freq='H', periods=365 * 24)

In [283]: ds = xarray.Dataset({'foo': ('time', np.arange(365 * 24)), 'time': time})

In [284]: ds.sel(time=slice(ds["time"][0], ds["time"][10]))

TypeError Traceback (most recent call last) <ipython-input-284-a101e126e1b0> in <module>() ----> 1 ds.sel(time=slice(ds["time"][0], ds["time"][10]))

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/xarray/core/dataset.py in sel(self, method, tolerance, drop, indexers) 1180 """ 1181 pos_indexers, new_indexes = indexing.remap_label_indexers( -> 1182 self, indexers, method=method, tolerance=tolerance 1183 ) 1184 result = self.isel(drop=drop, pos_indexers)

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/xarray/core/indexing.py in remap_label_indexers(data_obj, indexers, method, tolerance) 286 else: 287 idxr, new_idx = convert_label_indexer(index, label, --> 288 dim, method, tolerance) 289 pos_indexers[dim] = idxr 290 if new_idx is not None:

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/xarray/core/indexing.py in convert_label_indexer(index, label, index_name, method, tolerance) 183 indexer = index.slice_indexer(_try_get_item(label.start), 184 _try_get_item(label.stop), --> 185 _try_get_item(label.step)) 186 if not isinstance(indexer, slice): 187 # unlike pandas, in xarray we never want to silently convert a slice

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/pandas/tseries/index.py in slice_indexer(self, start, end, step, kind) 1496 1497 try: -> 1498 return Index.slice_indexer(self, start, end, step, kind=kind) 1499 except KeyError: 1500 # For historical reasons DatetimeIndex by default supports

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/pandas/indexes/base.py in slice_indexer(self, start, end, step, kind) 2995 """ 2996 start_slice, end_slice = self.slice_locs(start, end, step=step, -> 2997 kind=kind) 2998 2999 # return a slice

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/pandas/indexes/base.py in slice_locs(self, start, end, step, kind) 3174 start_slice = None 3175 if start is not None: -> 3176 start_slice = self.get_slice_bound(start, 'left', kind) 3177 if start_slice is None: 3178 start_slice = 0

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/pandas/indexes/base.py in get_slice_bound(self, label, side, kind) 3113 # For datetime indices label may be a string that has to be converted 3114 # to datetime boundary according to its resolution. -> 3115 label = self._maybe_cast_slice_bound(label, side, kind) 3116 3117 # we need to look up the label

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/pandas/tseries/index.py in _maybe_cast_slice_bound(self, label, side, kind) 1444 1445 if is_float(label) or isinstance(label, time) or is_integer(label): -> 1446 self._invalid_indexer('slice', label) 1447 1448 if isinstance(label, compat.string_types):

/dev/shm/gerrit/venv/stable-3.5/lib/python3.5/site-packages/pandas/indexes/base.py in _invalid_indexer(self, form, key) 1282 "indexers [{key}] of {kind}".format( 1283 form=form, klass=type(self), key=key, -> 1284 kind=type(key))) 1285 1286 def get_duplicates(self):

TypeError: cannot do slice indexing on <class 'pandas.tseries.index.DatetimeIndex'> with these indexers [946684800000000000] of <class 'int'> ```

A great beauty of numpys masked arrays is that it works with any dtype, since it does not use nan. Unfortunately, when I try to put my data into an xarray.Dataset, it converts ints to float, as shown below:

``` In [137]: x = arange(30, dtype="i1").reshape(3, 10)

In [138]: xr.Dataset({"count": (["x", "y"], ma.masked_where(x%5>3, x))}, coords={"x": range(3), "y": ...: range(10)}) Out[138]: <xarray.Dataset> Dimensions: (x: 3, y: 10) Coordinates: * y (y) int64 0 1 2 3 4 5 6 7 8 9 * x (x) int64 0 1 2 Data variables: count (x, y) float64 0.0 1.0 2.0 3.0 nan 5.0 6.0 7.0 8.0 nan 10.0 ... ```

This happens in the function _maybe_promote.

Such type “promotion” is unaffordable for me; the memory consumption of my multi-gigabyte arrays would explode by a factor 4. Secondly, many of my integer-dtype fields are bit arrays, for which floating point representation is not desirable.

It would greatly benefit xarray if it could use masking while preserving the dtype of input data.

(See also: Stackoverflow question)

da.plot.pcolormesh, where da is a DataArray, fails with TypeError: invalid type promotion when one of the coordinates is a datetime array:

``` $ cat mwe.py

!/usr/bin/env python3.6

import xarray import numpy

da = xarray.DataArray( numpy.arange(3*4).reshape(3,4), dims=("x", "y"), coords={"x": [1,2,3], "y": [numpy.datetime64(f"2000-01-{x:02d}") for x in range(1, 5)]})

da.plot.pcolormesh() $ ./mwe.py Traceback (most recent call last): File "./mwe.py", line 13, in <module> da.plot.pcolormesh() File "/dev/shm/gerrit/venv/stable-3.6/lib/python3.6/site-packages/xarray/plot/plot.py", line 547, in plotmethod return newplotfunc(allargs) File "/dev/shm/gerrit/venv/stable-3.6/lib/python3.6/site-packages/xarray/plot/plot.py", line 500, in newplotfunc kwargs) File "/dev/shm/gerrit/venv/stable-3.6/lib/python3.6/site-packages/xarray/plot/plot.py", line 667, in pcolormesh primitive = ax.pcolormesh(x, y, z, kwargs) File "/dev/shm/gerrit/venv/stable-3.6/lib/python3.6/site-packages/matplotlib/init.py", line 1710, in inner return func(ax, *args, kwargs) File "/dev/shm/gerrit/venv/stable-3.6/lib/python3.6/site-packages/matplotlib/axes/_axes.py", line 5636, in pcolormesh coords = np.column_stack((X, Y)).astype(float, copy=False) File "/dev/shm/gerrit/venv/stable-3.6/lib/python3.6/site-packages/numpy/lib/shape_base.py", line 353, in column_stack return _nx.concatenate(arrays, 1) TypeError: invalid type promotion ```