ENH: Series from MultiIndex causes segfault

[hayd]

In [3]: m = pd.MultiIndex.from_arrays([[1,2], [3,4]])

In [4]: m
Out[4]:
MultiIndex
[(1, 3), (2, 4)]

In [5]: pd.Series(m)
[1]    786 segmentation fault  ipython

pd.Series(m.values) has same fault (list of tuples)

Probably this should just be disallowed.

[hayd]

Perhaps this should return (not sure):

In [11]: m.to_series()
Out[11]:
1  3    (1, 3)
2  4    (2, 4)
dtype: object

[hayd]

pushing out a fix to NotImplementedError.

Could just as easily return data.to_series(), thoughts?

[jreback]

works for the index case, but not for list of tuples (I would just do NotImplementedError for now)

[hayd]

@jreback what do you mean not for the list of tuples?

In [4]: pd.Series([(1,2), (3,4)])
Out[4]:
0    (1, 2)
1    (3, 4)
dtype: object

[jreback]

hmm...must have been a typo...nvm

[cpcloud]

what's causing the segfault here

[hayd]

@cpcloud that's really weird, what was it doing? Was it just me?

[cpcloud]

i can't repro it

In [1]: paste
In [3]: m = pd.MultiIndex.from_arrays([[1,2], [3,4]])
## -- End pasted text --

In [2]: Series(m)
Out[2]: Series([], dtype: object)

In [3]: m
Out[3]:
MultiIndex
[(1, 3), (2, 4)]

In [4]: Series(m.values)
Out[4]:
0    (1, 3)
1    (2, 4)
dtype: object

In [5]: pd.__version__
Out[5]: '0.12.0.dev-e2638e0'

this is the commit right before this PR

[jtratner]

I was getting a segfault on this too.
On Jul 13, 2013 2:29 PM, "Phillip Cloud" [email protected] wrote:

i can't repro it

In [1]: paste
In [3]: m = pd.MultiIndex.from_arrays([[1,2], [3,4]])

-- End pasted text --

In [2]: Series(m)
Out[2]: Series([], dtype: object)

In [3]: m
Out[3]:
MultiIndex
[(1, 3), (2, 4)]

In [4]: Series(m.values)
Out[4]:
0 (1, 3)
1 (2, 4)
dtype: object

In [5]: pd.version
Out[5]: '0.12.0.dev-e2638e0'

this is the commit right before this PR

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/4187#issuecomment-20924244
.