Clearer error message for pd.concat if empty iterator is passed

[cel4]

I just saw following problem on stackoverflow: http://stackoverflow.com/questions/27665560/concatenating-files-with-python-pandas

It is not possible to concatenate an empty sequence of data frames.

This tiny code fragment

import pandas as pd
dfs = []
pd.concat(dfs)

Throws following error:

/Users/ch/miniconda/envs/sci33/lib/python3.3/site-packages/pandas/tools/merge.py in __init__(self, objs, axis, join, join_axes, keys, levels, names, ignore_index, verify_integrity, copy)
    765 
    766         if len(objs) == 0:
--> 767             raise ValueError('All objects passed were None')
    768 
    769         # consolidate data & figure out what our result ndim is going to be

ValueError: All objects passed were None

I would suggest catching this special case in pd.concat before trying to merge and to raise an error with a clearer error description. For this special case, the current error message is not very helpful.

[jreback]

this is already a special case
what exactly is not clear ? what are u suggesting for a more clear message?

[cel4]

Well, strictly speaking, the error message All objects passed were None is correct, since no object was passed. 😄
I would either check the length two times in merge.py. If it is 0 at the beginning, maybe the message No objects were passed is more accurate.
Even better may be to check in pd.concat if the sequence is empty and raise a ValueError with the message need at least one object to concatenate or something similar.

[jreback]

I would accept a PR to have no objects passed for this case

[cel4]

This is my first attempt: #9172
Waiting for travis.

[springcoil]

Is this one closed or open still?

[cel4]

@springcoil, thanks for the reminder. This should've been closed along with my closed pr. Now my PR is open again and I am finishing it, so this issue will hopefully be merged-closed soon.

[jreback]

no, that's the point of closing on merging of a PR and not before

[jreback]

closed by #9172