Inconsistent DataFrame.query results

[mattytrentini]

Having seen some inconsistent results from processing DataFrames - literally, running scripts over static data would generate different outcomes - we were able to boil down the following example:

def compare(df):
    df1 = df.query('x < 2000')
    df2 = df.query('x < 2000')
    return df1.equals(df2)

If I understand correctly, calling compare should never return false and yet it does, randomly:

>>> df = pd.DataFrame({'x': range(100000), 'y': range(100000)})
>>> for i in xrange(1000):
...     if not compare(df):
...         print ('failed at iteration', i)
...
failed at iteration 10
failed at iteration 66
failed at iteration 80
failed at iteration 141
failed at iteration 149
            ...
failed at iteration 878
failed at iteration 999

This seems like a defect in query?

Relevant versions:

>python --version
Python 2.7.11 :: Anaconda 2.4.0 (64-bit)
>systeminfo | findstr /B /C:"OS Name" /C:"OS Version" /C:"System Type"
OS Name:                   Microsoft Windows 8.1 Pro
OS Version:                6.3.9600 N/A Build 9600
System Type:               x64-based PC

[jreback]

duplicate of this

install numexpr=>2.4.6 and all will be well

usually conda update numexpr will suffice

[mattytrentini]

Thanks!

Sorry to waste your time, I did search but didn't stumble on the other duplicates.

Always nice to see an issue resolved and closed in less time than it took to write! :)

[jreback]

hah, yeah have seen this one a number of times. Not much we can do until Anaconda updates the version. To be more insidious, this is ONLY on windows.