BUG: tail raises on empty DataFrame

[MichaelWS]

closes #5846

calling tail on an empty frame threw an exception. In addition, df.tail(0) returns self without indexing.

[jreback]

had and tail should have very similar checking code for the edge cases

[MichaelWS]

added a check for empty frame for head as well.

[jreback]

@MichaelWS fyi....I edited the title / top of the PR.

putting the closes issue number sets up a link the original issue, then automatically closes it when its merged.

[MichaelWS]

Thanks.

[jreback]

you need similar logic on head, no?

[MichaelWS]

I added similar logic on head now. The question is this changes how df.head(0) works. It now returns self. before it returned the same as df.head(1)

[jreback]

don't change the test

can u show me what is returned kn both current and your case for head(0)?

[MichaelWS]

Currently it returns:
In [3]: pd.DataFrame([1,2,3]).head(0)
Out[3]:
Empty DataFrame
Columns: [0]
Index: []

[0 rows x 1 columns]

I could easily keep that functionality. I am not sure why someone would call df.head(0) for a non-empty frame.

[MichaelWS]

My case would be:

In [3]: pd.DataFrame([1,2,3]).head(0)
Out[3]:
0
0 1
1 2
2 3

[3 rows x 1 columns]

[jreback]

yours is my right bug neither is the original

I think it should return self

iloc[0] on an empty frame is wrong too I think - can I make an issue for that one?

[MichaelWS]

yes, please make an issue on that.
I agree with you. I think self makes more sense.

[jreback]

can you rebase this?

[MichaelWS]

sure. this is done.

[MichaelWS]

The only real functionality change is df.head(0) returns self

[jreback]

hmm...ok why don't you put an additional note in the API section (leave your existing one their). I don't think its a big deal but should document it (reference this issue or original again)

[MichaelWS]

I put a note in the docs.

[jreback]

put this in the API section (the last one)

[jreback]

thanks for the PR!