"Cannot convert tz-naive timestamps"

[vfilimonov]

I'm not sure if it is an issue of DOCs or code.

Docs say that tz_convert "Convert the axis to target time zone. If it is time zone naive, it will be localized to the passed time zone."

But the code in contrast raises an exception:

tmp = pd.Series([0], index=[pd.Timestamp('2008-12-16 13:10:00')])
tmp.tz_convert('UTC')

Results in TypeError: Cannot convert tz-naive timestamps, use tz_localize to localize

[jorisvandenbossche]

I think it is a docs thing. Also the docs on DatetimeIndex.tz_convert are correct in that way: http://pandas.pydata.org/pandas-docs/stable/generated/pandas.DatetimeIndex.tz_convert.html#pandas.DatetimeIndex.tz_convert

@jreback correct?

[jreback]

yeh, this is just an old/not updated doc-string. This will only convert if it is already non-naive, while tz_localize only will work if its naive. This is the original API, and was separate on purpose to prevent non-explict conversions.

want to do a PR to update the doc-strings (maybe add a Raises section to both of these functions in tseries/index.py as well as the functions in core/generic.py

[vfilimonov]

Sure, will do docs.

@jreback , what do you mean by Raises section?

[jorisvandenbossche]

A section in the docstring, see https://github.com/numpy/numpy/blob/master/doc/HOWTO_DOCUMENT.rst.txt#sections

[vfilimonov]

OK, will do

[jorisvandenbossche]

Closed by #9906