ffill, bfill are broken

[VelizarVESSELINOV]

"""Pandas bug."""
from io import StringIO

from pandas import read_csv

dtf = read_csv(StringIO('''G,M
G1253,
G1253,0
G1253,0
G1253,
G1253,
G1253,21.3
G1253,32.3
G1253,98.5
G1253,
G1253,121.9
G1253,
G1253,125.6
G1253,136.6
G1253,
G1253,144.2
G1253,
G1253,161.8
G1253,
G1253,180.1
G1253,180.1'''))

grp = dtf.groupby('G')

dtf['M1'] = grp.M.ffill()
dtf['M2'] = grp.M.fillna(method='ffill')

print(dtf[dtf.M1 != dtf.M2][['M1', 'M2']])

assert(dtf.M1.equals(dtf.M2))

Problem description

Expect ffill to be working and equal to fillna(ffill)

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.5.final.0
python-bits: 64
OS: Darwin
OS-release: 17.5.0
machine: x86_64
processor: i386
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8

pandas: 0.23.0
pytest: None
pip: 10.0.1
setuptools: 39.0.1
Cython: None
numpy: 1.14.2
scipy: 1.1.0
pyarrow: 0.9.0.post1
xarray: None
IPython: 6.2.1
sphinx: None
patsy: None
dateutil: 2.7.2
pytz: 2018.3
blosc: None
bottleneck: None
tables: 3.4.3
numexpr: 2.6.4
feather: None
matplotlib: 2.2.0
openpyxl: 2.5.0
xlrd: 1.1.0
xlwt: 1.3.0
xlsxwriter: None
lxml: 4.2.1
bs4: None
html5lib: 0.9999999
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[WillAyd]

I couldn't reproduce this on master - what difference are you actually seeing?

[VelizarVESSELINOV]

       M1     M2
0     NaN    NaN
3    21.3    0.0
4    21.3    0.0
8   121.9   98.5
10  125.6  121.9
13  144.2  136.6
15  161.8  144.2
17    NaN  161.8
Traceback (most recent call last):  File "bug1.py", line 35, in <module>
    assert(dtf.M1.equals(dtf.M2))
AssertionError

[WillAyd]

Hmm OK. I was able to reproduce on v0.23 but not on master - can you try on the latter and confirm it is resolved there?

[uds5501]

@VelizarVESSELINOV Use current master build. Their is no such error here:

>>> print(dtf[dtf.M1 != dtf.M2][['M1', 'M2']])
   M1  M2
0 NaN NaN

although the error still persists in v.0.23.0 (triaged succesfully there and can confirm with your output)

[VelizarVESSELINOV]

I'm surprised that v0.23 contains so basic bug that was not caught by your automatic testing and it is taking so much time to deliver a fix version. When is it expected v0.23.1 version?

Can I assume that fillna(method='f/bfill') is working in v0.23?

Didn't manage to compile the pandas master after several tentatives, so I trust you that it is fixed in the master.

[WillAyd]

@VelizarVESSELINOV v0.23.1 is slated for release today. Given you just opened this 5 hours ago, I'd say that's a fairly quick turnaround :-)

[WillAyd]

Resolved via #21212