pydata / xarray / 5262

96%

push

travis-ci

Bug fixes for Dataset.reduce() and n-dimensional cumsum/cumprod (#2156)

* Bug fixes for Dataset.reduce() and n-dimensional cumsum/cumprod

Fixes GH1470, "Dataset.mean drops coordinates"

Fixes a bug where non-scalar data-variables that did not include the
aggregated dimension were not properly reduced:

    Previously::

        >>> ds = Dataset({'x': ('a', [2, 2]), 'y': 2, 'z': ('b', [2])})
        >>> ds.var('a')
        <xarray.Dataset>
        Dimensions:  (b: 1)
        Dimensions without coordinates: b
        Data variables:
            x        float64 0.0
            y        float64 0.0
            z        (b) int64 2

    Now::

        >>> ds.var('a')
        <xarray.Dataset>
        Dimensions:  (b: 1)
        Dimensions without coordinates: b
        Data variables:
            x        int64 0
            y        int64 0
            z        (b) int64 0

Finally, adds support for n-dimensional cumsum() and cumprod(), reducing
over all dimensions of an array. (This was necessary as part of the above fix.)

* Lint fixup

* remove confusing comments

76 of 76 new or added lines in 5 files covered. (100.0%)

23231 of 23957 relevant lines covered (96.97%)

8.9 hits per line