BUG: asfreq silently drops rows when index is not sorted

[Gerenuk]

pd.DataFrame.asfreq seems to silently drop rows when the data is not sorted.

d=pd.DataFrame(range(3), index=pd.to_datetime(["2021-01-01", "2021-03-01", "2021-02-01"])).asfreq("MS")
d
# output:
# 2021-01-01  0
# 2021-02-01  2

Silent data loss is very dangerous in data analysis as it's hard to detect.

Expected behavior:

• an exception, because dropping rows is not useful
• or automatic sorting (with possibly a parameter in asfreq to be explicit)

INSTALLED VERSIONS
------------------
commit           : 9d598a5e1eee26df95b3910e3f2934890d062caa
python           : 3.8.5.final.0
pandas           : 1.2.1
numpy            : 1.19.2

[nmay231]

The source of the issue seems to be on this line.

I'm willing to take on this issue. I just need to which behavior is preferred.

I would lean towards handling this implicitly. I don't see a need to raise an error or warning.

[MarcoGorelli]

Thanks @Gerenuk for the excellent report - I don't know what the correct solution would be, but can confirm it reproduces

[nmay231]

take