When using dask.dataframe.DataFrame.to_csv, I noticed that lines end with \r\r\n, as opposed to simply \r\n. I'm guessing this is a Windows issue, although I haven't had the opportunity to test on another system.

The following example shows the bytes produced when using pandas and dask to write the same data to csv. The pandas output is what I was expecting for both. Hopefully this makes it clear what's going on.

Code

import sys
import dask, dask.dataframe as dd
import pandas as pd

df = pd.DataFrame({'a': [0]})
ddf = dd.from_pandas(df, npartitions=1)

df.to_csv('temp-pandas.csv')
pandas_bytes = open('temp-pandas.csv', 'rb').read()

ddf.to_csv('temp-dask_*.csv')
dask_bytes = open('temp-dask_0.csv', 'rb').read()

print('pandas result:', pandas_bytes)
print('dask result  :', dask_bytes)
print('---')
print('dask version:', dask.__version__)
print('Python version:', sys.version)

Output

pandas result: b',a\r\n0,0\r\n'
dask result  : b',a\r\r\n0,0\r\r\n'
---
dask version: 1.2.2
Python version: 3.7.3 (default, Mar 27 2019, 17:13:21) [MSC v.1915 64 bit (AMD64)]

Thanks for the work on an awesome package!