Unexpected behaviour of df.to_json(compression="gzip")

[MartinThoma]

Code Sample

import pandas as pd

df = pd.DataFrame({"a": [1, 2, 3, 4], "b": ["A", "B", "C", "D"],})
json_df = df.to_json(lines=True, orient="records", compression="gzip")
print(json_df)

Problem description

The behavior of the "compression" argument is unexpected. It doesn't have any effect. Although it is documented that it is only applied if there is a filename argument, there should at very least be a warning if compression is set, but no filename is given.

Also: Why doesn't it have an effect? Couldn't we output a bytes object?

[R4HMATT]

I can work on adding a warning to this for now. Not sure if the pandas team wants us to output a bytes object though.