display_width doesn't apply to dask-backed arrays

[rabernat]

The representation of dask-backed arrays in xarray's __repr__ methods results in very long lines which often overflow the desired line width. Unfortunately, this can't be controlled or overridden with xr.set_options(display_width=...).

Code Sample, a copy-pastable example if possible

import xarray as xr
xr.set_options(display_width=20)
ds = (xr.DataArray(range(100))
      .chunk({'dim_0': 10})
      .to_dataset(name='really_long_long_name'))
ds

<xarray.Dataset>
Dimensions:                (dim_0: 100)
Dimensions without coordinates: dim_0
Data variables:
    really_long_long_name  (dim_0) int64 dask.array<shape=(100,), chunksize=(10,)>

Problem description

[this should explain why the current behavior is a problem and why the expected output is a better solution.]

Expected Output

We need to decide how to abbreviate dask arrays with something more concise. I'm not sure the best way to do this. Maybe

   really_long_long_name  (dim_0) int64 dask chunks=(10,)

[shoyer]

short_dask_repr controls this in xarray:

xarray/xarray/core/formatting.py

Lines 401 to 411 in 6d55f99

	def short_dask_repr(array, show_dtype=True):
	"""Similar to dask.array.DataArray.__repr__, but without
	redundant information that's already printed by the repr
	function of the xarray wrapper.
	"""
	chunksize = tuple(c[0] for c in array.chunks)
	if show_dtype:
	return 'dask.array<shape=%s, dtype=%s, chunksize=%s>' % (
	array.shape, array.dtype, chunksize)
	else:
	return 'dask.array<shape=%s, chunksize=%s>' % (array.shape, chunksize)

A good place to start might be dropping shape -- that information is largely redundant with dimension sizes.

[stale]

In order to maintain a list of currently relevant issues, we mark issues as stale after a period of inactivity

If this issue remains relevant, please comment here or remove the stale label; otherwise it will be marked as closed automatically

[shoyer]

As of #3211, we no longer include the redundant shape in short reprs for dask arrays.

So I'm going to close this, though of course further improvements to the short repr would be welcome!