pd.Timestamp("now") returns oddly

[cancan101]

I would expect these two return the same:

In [10]: pd.Timestamp.now()

Out[10]:
Timestamp('2014-02-24 17:54:18.370128', tz=None)

In [11]: pd.Timestamp("now")

Out[11]:
Timestamp('2014-02-24 22:54:18', tz=None)

[jreback]

these differ by the time zone
try utcnow

[cancan101]

I consider this to be a bug. Since there is no timezone associated with the return from pd.Timestamp("now"), I would argue that it should be consistent with pd.Timestamp.now()

[jreback]

nope
it gets passed to dateutil which is somewhat unpredictable - user beware

try passing pm - it works

[cancan101]

pm?

[cancan101]

Also the string is interpreted by Pandas: https://github.com/pydata/pandas/blob/master/pandas/src/datetime/np_datetime_strings.c#L453

[jreback]

ok that's fair; prob should allow utcnow as well (and fix now)

can u post what numpy does with now?

is it consistent? (that's where this code came from)

[cancan101]

Numpy supports my cases (albeit they handle timezones):

In [7]:
numpy.datetime64("now")

Out[7]:
numpy.datetime64('2014-02-24T18:44:10-0500')

[jreback]

ok

[jreback]

This was fixed by #9000

In [18]: pd.Timestamp('now')
Out[18]: Timestamp('2015-03-02 20:31:02.162480')

In [19]: pd.Timestamp.now()
Out[19]: Timestamp('2015-03-02 20:31:04.322196')