Create Series from (remote) partitions

Is there an equivalent to from_partitions which creates a Series (rather than a DF)?

Best practices here would be to create a dataframe then use df.squeeze() to create the Series. It will just convert from a dataframe view to the Series.

Would that be sufficient for your purposes?

Of course this makes it possible. Performance is going to be terrible though as I suspect it will always involve multiple unnecessary copies instead of just re-using unmaterialized objects. Is that true?

Just as a motivating example, here is a common np+pd schema I have in mind:

a,b,c,.. = partitioned_1d_arrays()
df = DataFrame({'a': a, 'b': b, 'c': c, ...})

Ideally, modin would take concrete or unmaterialized partition objects. If unmaterialized, there should be no copy if possible.

Series and DataFrame are just views on the underlying structure. There is no real cost associated with going between them and nothing is copied.