Negative timedeltas not working correctly

[dstansby]

(sorry for not thinking of a better title...)

Code Sample, a copy-pastable example if possible

import pandas as pd
print(pd.Timedelta(seconds=-0.5)

Problem description

This outputs -1 days +23:59:59.500000

Expected Output

-00:00:00:5000000 or similar

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.0.final.0
python-bits: 64
OS: Linux
OS-release: 4.8.0-39-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_GB.UTF-8
LOCALE: en_GB.UTF-8

pandas: 0.19.2
nose: 1.3.7
pip: 9.0.1
setuptools: 27.2.0
Cython: 0.25.2
numpy: 1.12.0
scipy: 0.18.1
statsmodels: None
xarray: None
IPython: 5.3.0
sphinx: 1.5.3
patsy: None
dateutil: 2.6.0
pytz: 2016.10
blosc: None
bottleneck: None
tables: 3.3.0
numexpr: 2.6.2
matplotlib: 2.0.0.post3537+g0151c05
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: 0.999
httplib2: 0.10.3
apiclient: None
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.9.5
boto: None
pandas_datareader: None

[dstansby]

Just realised that this is because of the weird formatting of timedeltas, sorry.

[jreback]

yeah this is unfortunate but how python does this. I agree its cumbersome (and of course was harder to implement this way).

In [1]: pd.Timedelta('-0.5s')
Out[1]: Timedelta('-1 days +23:59:59.500000')

we however accept negatives in parsing so all is not lost.

what are your thoughts on actually changing the repr though? we could certainly do it.

[jreback]

In [3]: datetime.timedelta(seconds=-0.5)
Out[3]: datetime.timedelta(-1, 86399, 500000)

In [4]: datetime.timedelta(seconds=0.5)
Out[4]: datetime.timedelta(0, 0, 500000)

[scls19fr]

Changing repr would be a great idea!
see #17232