ReadCSV: parameter nrows cause fallback to Pandas

[akharche]

Adding 'dtype' parameter to read_csv() falls back to Pandas implementation.

dtype = np.int16
df = pd.read_csv(filename, nrows=100000, dtype=dtype, header=None)

UserWarning: Parameters provided defaulting to pandas implementation.

[devin-petersohn]

Hi @akharche, thanks for the report!

I believe it is the nrows=100000 argument that is causing the default to pandas. dtype parameter is supported.

[akharche]

Hi @devin-petersohn, thanks for the clarification!

Should I reopen the issue with 'nrows' parameter?

[devin-petersohn]

No we can keep the issue. @aregm has changed the title.