mars-project/mars#2488 - Replication/Quicktest code

mars_2488_test.py

import mars
import mars.dataframe as md
import pandas as pd
import numpy as np

# NOTE: pytest mars fails with missing module pyarrow. (non-urgent)
# mars-project/mars/issues/2488

# TYPE: Recreate error -> 7/10/2023
# Original Report - Reproducibility [package requirements]
# https://github.com/mars-project/mars/issues/2488
# Mars - 0.6.11
# pandas - 1.1.3

# Current Proposed - Reproducibility [package requirements]
# scipy              1.7.3
# numpy              1.21.6
# pandas             1.3.5
# pip                23.2.1
# pymars             0+untagged.1296.g   Commit - 0a42ba8

# Note: Mars Version differs.
# Changing between pandas version does not impact error output. Reported error is similar.

# Pandas Component
b = pd.Series([1, 3, 2, np.nan, np.nan])
# print(b)
# 0    1.0
# 1    3.0
# 2    2.0
# 3    NaN
# 4    NaN
# dtype: float64


b.dropna().sort_values(ignore_index=True)
# print(b)
# 0    1.0
# 1    3.0
# 2    2.0
# 3    NaN
# 4    NaN
# dtype: float64

# Mars Component
mars.new_session()
a = md.Series([1,3,2,np.nan,np.nan])
print(a) # Series(op=SeriesDataSource) - Correct.
print("-"*30)
# print(a.shape) # (5,) - Correct.
# Separating the commands works.
# a.dropna()
# a.sort_values(ignore_index=True).execute()

# Combining does not work.
a.dropna().sort_values(ignore_index=True).execute()