Is `line_delimited` being ignored in `dak.from_json`?

[ikrommyd]

Notice the differences between dask-awkward and awkward in the layout:

In [15]: ak.from_json(pathlib.Path("tmp.json"), line_delimited=True)
Out[15]: <Array [[[1.1, 2.2, 3.3], [], [4.4, 5.5]]] type='1 * var * var * float64'>

In [16]: ak.from_json(pathlib.Path("tmp.json"), line_delimited=False)
Out[16]: <Array [[1.1, 2.2, 3.3], [], [4.4, 5.5]] type='3 * var * float64'>

In [17]: dak.from_json("tmp.json", line_delimited=True).compute()
Out[17]: <Array [[[1.1, 2.2, 3.3], [], [4.4, 5.5]]] type='1 * var * var * float64'>

In [18]: dak.from_json("tmp.json", line_delimited=False).compute()
Out[18]: <Array [[[1.1, 2.2, 3.3], [], [4.4, 5.5]]] type='1 * var * var * float64'>

tmp.json was created by:

with open("tmp.json", "w") as file:
    file.write("[[1.1, 2.2, 3.3], [], [4.4, 5.5]]")

[martindurant]

Yes, dak requires line-delimited in order to to partition the data within files. If False, each input file is a one-row partition. So the two versions are not the same generally, only for this specific case.