.loc behavior when using 'slice'

[beobest2]

I test as shown in the documentation below.
I found something strange.

https://koalas.readthedocs.io/en/latest/reference/api/databricks.koalas.Series.loc.html

pandas :
'sidewinder' is at the end of the list, so it prints normally.

>>> pdf = pd.DataFrame([[1, 2], [4, 5], [7, 8]], 
...             index=['cobra', 'viper', 'sidewinder'],  
...             columns=['max_speed', 'shield'])
>>> pdf.loc['cobra':'viper', 'max_speed']
cobra    1
viper    4
Name: max_speed, dtype: int64

but koalas prints 'sidewinder'

>>> kdf = ks.DataFrame([[1, 2], [4, 5], [7, 8]],
...              index=['cobra', 'viper', 'sidewinder'],
...              columns=['max_speed', 'shield'])
>>> kdf.loc['cobra':'viper', 'max_speed']
cobra         1
viper         4
sidewinder    7
Name: max_speed, dtype: int64

>>> kdf.loc['cobra':, 'max_speed']
cobra         1
viper         4
sidewinder    7
Name: max_speed, dtype: int64

>>> kdf.loc['sidewinder':, 'max_speed']
viper         4
sidewinder    7
Name: max_speed, dtype: int64

>>> kdf.loc['viper':, 'max_speed']
viper    4
Name: max_speed, dtype: int64

Input is done in the order ['cobra', 'viper', 'sidewinder'],
but koals appears to be recognized as ['cobra', 'sidewinder', 'viper'].

[itholic]

thanks for reporting, let me take a look at this

[itholic]

fix this at #1159

summary

the existing logic was determining range from start to stop by string order alphabetically, so couldn't keep natural order.

we have to determine the range from start to stop with natural order like pandas.

for more detail

for example, let's assume that we have kdf like below

>>> kdf
            max_speed  shield
cobra               1       2
viper               4       5
sidewinder          7       8

and perform iloc on this with slice from 'cobra' to 'viper' like below,

>>> kdf.loc['cobra':'viper', 'max_speed']

and we expect that the result will not include 'sidewinder' since it is really not between 'cobra' and 'viper' shown kdf above.
(actually, more fundamentally, because pandas is doing like this)

so i fixed it with our new feature NATURAL_ORDER_COLUMN_NAME to keep natural order.