DataFrame.to_csv() behaves differently when passing compression parameter

Question

If i use:

df.to_csv(\'dir1/dir2/test.csv.zip\')

or

df.to_csv(\'dir1/dir2/test.csv\', compression=\'zip\')

to