Add method to return the output as an xarray?

[nbren12]

Hi @spencerkclark. I've found myself using xpartition for scoring workloads. Usually there is a big data reduction involved so the output does fit in memory. I've been using xpartition like this

    import xpartition
    block_list = []
    n = 20
    for i in range(n):
        print(i)
        index = output.z500.partition.indexers(n, i, dims=['initial_time'])
        block = output.isel(index).compute()
        block_list.append(block)
    output = xr.concat(block_list, dim='initial_time')

Maybe it's worth adding some helper that does this.

[spencerkclark]

Neat, yeah, I'd be open to adding something like this if we can implement a version that handles partitions along n-dimensions. I've found myself in this situation before too. Keeping with accessors for the time being maybe we could spell it da.partition.compute(steps, dims)?