Extremely slow repr() on small data frame

[njsmith]

I have a small data frame here (obtained by slicing a much larger data frame with a MultiIndex):

In [6]: x.shape
Out[6]: (4, 32)

Its repr() looks like this:

In [7]: print repr(x)
<class 'pandas.core.frame.DataFrame'>
MultiIndex: 4 entries, ('arquant9.crw', 10, 336640) to ('arquant9.crw', 10, 336652)
Data columns:
lle     4  non-null values
lhz     4  non-null values
MiPf    4  non-null values
LLPf    4  non-null values
RLPf    4  non-null values
LMPf    4  non-null values
RMPf    4  non-null values
LDFr    4  non-null values
RDFr    4  non-null values
LLFr    4  non-null values
RLFr    4  non-null values
LMFr    4  non-null values
RMFr    4  non-null values
LMCe    4  non-null values
RMCe    4  non-null values
MiCe    4  non-null values
MiPa    4  non-null values
LDCe    4  non-null values
RDCe    4  non-null values
LDPa    4  non-null values
RDPa    4  non-null values
LMOc    4  non-null values
RMOc    4  non-null values
LLTe    4  non-null values
RLTe    4  non-null values
LLOc    4  non-null values
RLOc    4  non-null values
MiOc    4  non-null values
A2      4  non-null values
HEOG    4  non-null values
rle     4  non-null values
rhz     4  non-null values
dtypes: float64(32)

Producing this repr takes a bizarrely long time, it's very noticeable in interactive use:

In [8]: time repr(x)
CPU times: user 1.39 s, sys: 0.06 s, total: 1.45 s
Wall time: 1.36 s

The object is available here: http://vorpus.org/~njs/tmp/slow-repr-x.pickle

[lodagro]

Seems related to the index

In [68]: time repr(x)
CPU times: user 3.91 s, sys: 0.10 s, total: 4.01 s
Wall time: 3.99 s
...

In [69]: time repr(x.reset_index())
CPU times: user 0.02 s, sys: 0.01 s, total: 0.03 s
Wall time: 0.03 s

[lodagro]

Some profiling and a little code change ...

In [11]: time repr(x)
CPU times: user 0.03 s, sys: 0.00 s, total: 0.03 s
Wall time: 0.04 s

[njsmith]

Fabulous, thanks!