html_url,issue_url,id,node_id,user,created_at,updated_at,author_association,body,reactions,performed_via_github_app,issue
https://github.com/pydata/xarray/issues/1044#issuecomment-253477840,https://api.github.com/repos/pydata/xarray/issues/1044,253477840,MDEyOklzc3VlQ29tbWVudDI1MzQ3Nzg0MA==,4160723,2016-10-13T10:37:24Z,2016-10-14T13:07:41Z,MEMBER,"After seeing the discussion in #680, I'm wondering if showing the firsts values of the flattened array wouldn't be enough here, e.g., something like this:

```
>>> d
<xarray.DataArray (a: 2, b: 5, c: 2, d: 10)>
  array          int64 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 ...
Coordinates:
  * a        (a) <U1 'A' 'B'
  * b        (b) <U5 'Cat 1' 'Cat 2' 'Cat 3' 'Cat 4' 'Cat 5'
  * c        (c) <U1 'J' 'K'
  * d        (d) int64 0 1 2 3 4 5 6 7 8 9
```

This example is more consistent with the repr of `Dataset` data variables, and similarly we could customize the repr of dask arrays and lazy arrays (loaded from netcdf files) like this: 

```
>>> d.chunk((10, 5, 5, 10))
<xarray.DataArray (a: 2, b: 5, c: 2, d: 10)>
  dask.array     int64 chunksize=(10, 5, 5, 10)
Coordinates:
  * a        (a) <U1 'A' 'B'
  * b        (b) <U5 'Cat 1' 'Cat 2' 'Cat 3' 'Cat 4' 'Cat 5'
  * c        (c) <U1 'J' 'K'
  * d        (d) int64 0 1 2 3 4 5 6 7 8 9
```

```
>>> d.name = 'myvar'
>>> d.to_netcdf('data.nc')
>>> xr.open_dataset('data.nc').myvar
<xarray.DataArray 'myvar' (a: 2, b: 5, c: 2, d: 10)>
  lazy-array     int64
Coordinates:
  * a        (a) <U1 'A' 'B'
  * b        (b) <U5 'Cat 1' 'Cat 2' 'Cat 3' 'Cat 4' 'Cat 5'
  * c        (c) <U1 'J' 'K'
  * d        (d) int64 0 1 2 3 4 5 6 7 8 9
```
","{""total_count"": 1, ""+1"": 1, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,182638499
https://github.com/pydata/xarray/issues/1044#issuecomment-253649762,https://api.github.com/repos/pydata/xarray/issues/1044,253649762,MDEyOklzc3VlQ29tbWVudDI1MzY0OTc2Mg==,4160723,2016-10-13T21:52:46Z,2016-10-13T21:57:24Z,MEMBER,"In most cases I found the `DataArray` repr useful for quickly checking the dimensions (both names and sizes), the attributes and the types/values of both data and labels (I mean just checking here if the values are consistent regarding their units, acceptable ranges, etc.), but rarely for in-depth checking of the data values along each dimension, hence my suggestion of a flat (subset) array.

To inspect the data of high dimensional datarrays, I've mainly used the indexing logic of xarray to extract slices of <3 dimensions. However, I admit that for quick inspection purposes I actually like your suggestion of having a specific repr method that would allow showing small data slices as labeled tables, especially if we choose to always use a flat array for the repr of `Dataarray` (i.e., even when the number of dimensions <3). Why not something like:

``` python
>>> d.slice_repr(a=0, b=0)
d   0   1   2   3   4   5   6   7   8   9
c                                        
J   0   1   2   3   4   5   6   7   8   9
K  10  11  12  13  14  15  16  17  18  19
```

This is equivalent to

``` python
>>> dslice = d.isel(a=0, b=0)
>>> pd.DataFrame(data=dslice.data, index=dslice.c, columns=dslice.d)
```

Except that `slice_repr()` would return a string instead of a data object (or an array or a dataframe).
Not sure about the name and/or signature of `slice_repr()`, though.
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,182638499