mrocklin / dask / 544

94%

master: 94%

push

travis-ci

Index one array with another boolean dask.array

This allows for fancy indexing of the following form:

Example
-------

    In [1]: import numpy as np

    In [2]: import dask.array as da

    In [3]: x = np.array([2, 1, 5, 2, 1])

    In [4]: d = da.from_array(x, chunks=2)

    In [5]: d[d > 1]
    Out[5]: array([2, 5, 2])

Historically we have avoided operations like this for which we don't
know the shape of the output.  We get around this issue by computing
immediately.  I'm considering extending this approach to other
operations going forwards.

11 of 11 new or added lines in 1 file covered. (100.0%)

4806 of 5133 relevant lines covered (93.63%)

0.94 hits per line