Not sure if feature or bug, but definitely made me look twice.

```

ds = xray.Dataset({'time': (['time'], [np.datetime64('2014-04-26T05:00:00-0700')])}) ds.time.values[0] numpy.datetime64('2014-04-26T05:00:00.000000000-0700')

ds.to_dataframe().index[0] Timestamp('2014-04-26 12:00:00') ```

I'd expected to see the DataFrame index maintaining its local timezone info:

Timestamp('2014-04-26 05:00:00-0700', tz='America/Los_Angeles')

That said, the UTC that I actually get back is the same time so math should still work. I'm opening this issue mostly as a matter for discussion -- feel free to close if you think xray should be pushing users towards UTC.

This is special cased for np.ndarrays: applying to DataArrays is not only inefficient but would also be wrong if the applied function wanted to change metadata.

Fixes #326

Need to apply a transformation function across one dimension of a DataArray, where that non-xray function speaks in ndarrays. Currently the only ways to do this involve wrapping the function. An example:

``` import numpy as np import xray from scipy.ndimage.morphology import binary_opening

da = xray.DataArray(np.random.random_integers(0, 1, (10, 10, 3)), dims=['row', 'col', 'time'])

I want to apply an operation the 2D image at each point in time

da.groupby('time').apply(binary_opening)

AttributeError: 'numpy.ndarray' object has no attribute 'dims'

def wrap_binary_opening(da, kwargs): return xray.DataArray(binary_opening(da.values, kwargs), da.coords)

da.groupby('time').apply(wrap_binary_opening) da.groupby('time').apply(wrap_binary_opening, iterations=2) # func may take custom args ```

My proposed solution is that apply would automatically coerce func's return value to a DataArray.

xray.Dataset( {'password': ([], 'abcd')} ).to_netcdf('/tmp/bar.nc')

results in

``` ----> 1 xray.Dataset( {'password': ([], 'abcd')} ).to_netcdf('/tmp/bar.nc')

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/core/dataset.pyc in to_netcdf(self, filepath, kwdargs) 801 """ 802 with backends.NetCDF4DataStore(filepath, mode='w', kwdargs) as store: --> 803 self.dump_to_store(store) 804 805 dump = to_netcdf

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/core/dataset.pyc in dump_to_store(self, store, encoder) 793 if encoder: 794 variables, attributes = encoder(variables, attributes) --> 795 store.store(variables, attributes) 796 store.sync() 797

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/backends/netCDF4_.pyc in store(self, variables, attributes) 100 # to write times, for example, would fail. 101 cf_variables, cf_attrs = cf_encoder(variables, attributes) --> 102 AbstractWritableDataStore.store(self, cf_variables, cf_attrs) 103 104 def open_store_variable(self, var):

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/backends/common.pyc in store(self, variables, attributes) 153 variables = dict((k, v) for k, v in iteritems(variables) 154 if not (k in neccesary_dims and is_trivial_index(v))) --> 155 self.set_variables(variables) 156 157 def set_dimensions(self, dimensions):

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/backends/common.pyc in set_variables(self, variables) 165 def set_variables(self, variables): 166 for vn, v in iteritems(variables): --> 167 self.set_variable(_encode_variable_name(vn), v) 168 self.set_necessary_dimensions(v) 169

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/backends/netCDF4_.pyc in set_variable(self, name, variable) 151 attrs = variable.attrs.copy() 152 if self.format == 'NETCDF4': --> 153 variable, datatype = _nc4_values_and_dtype(variable) 154 else: 155 variable = encode_nc3_variable(variable)

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/backends/netCDF4_.pyc in _nc4_values_and_dtype(var) 47 # use character arrays instead of unicode, because unicode suppot in 48 # netCDF4 is still rather buggy ---> 49 data, dims = maybe_convert_to_char_array(var.values, var.dims) 50 var = Variable(dims, data, var.attrs, var.encoding) 51 dtype = var.dtype

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/backends/netcdf3.pyc in maybe_convert_to_char_array(data, dims) 55 def maybe_convert_to_char_array(data, dims): 56 if data.dtype.kind == 'S' and data.dtype.itemsize > 1: ---> 57 data = conventions.string_to_char(data) 58 dims = dims + ('string%s' % data.shape[-1],) 59 return data, dims

/export/data/envs/popcorn/lib/python2.7/site-packages/xray/conventions.pyc in string_to_char(arr) 344 if kind not in ['U', 'S']: 345 raise ValueError('argument must be a string') --> 346 return arr.view(kind + '1').reshape(*[arr.shape + (-1,)]) 347 348

ValueError: new type not compatible with array. ```

Note this does not fail when the 0d value is a number. This succeeds:

xray.Dataset( {'password': ([], 1)} ).to_netcdf('/tmp/bar.nc')

When concatenating Datasets, a variable's string length is limited to the length in the first of the Datasets being concatenated.

```

import xray first = xray.Dataset({'animal': ('animal', ['horse'])}) second = xray.Dataset( {'animal': ('animal', ['aardvark_0'])}) xray.Dataset.concat([first, second], dimension='animal')['animal'] <xray.DataArray 'animal' (animal: 2)> array(['horse', 'aardv'], dtype='|S5') Coordinates: animal: Index([u'horse', u'aardv'], dtype='object') Attributes: Empty ```

(Note the |S5 dtype and the truncated aardv)

I think this is the offending line: https://github.com/xray/xray/blob/master/xray/core/variable.py#L623 May want to use dtype=object for strings to avoid this issue.