So, as discussed @wang @cseed, the problem: I have a bunch of files where the rows are all in the same order, and I’d like to paste (in the unix sense) them all together. However, they don’t have a unique identifier on them.
import_matrix currently requires a key column. I’d love to be able to do None and have it automatically run ds.indexed_rows(), both creating an index column and keying on it. I don’t mind that this does a 2nd (or 3rd if imputation is needed) pass through the file to get the bounds of each chunk - it’s a relative edge case so speed is not suuuper crucial, but would certainly be useful down the line.