DatetimeIndex.values gives bad output

[ijmcf]

For example:

dr = pandas.date_range('03/06/2012 00:00', periods=25, freq='W-FRI', tz='US/Eastern')
dr.values
array([1970-01-16 67:00:00, 1970-01-16 234:00:00, 1970-01-16 145:00:00,
1970-01-16 63:00:00, 1970-01-16 229:00:00, 1970-01-16 142:00:00,
1970-01-16 57:00:00, 1970-01-16 217:00:00, 1970-01-16 136:00:00,
1970-01-16 43:00:00, 1970-01-16 217:00:00, 1970-01-16 122:00:00,
1970-01-16 33:00:00, 1970-01-16 203:00:00, 1970-01-16 123:00:00,
1970-01-16 32:00:00, 1970-01-16 198:00:00, 1970-01-16 109:00:00,
1970-01-16 21:00:00, 1970-01-16 187:00:00, 1970-01-16 104:00:00,
1970-01-16 12:00:00, 1970-01-16 177:00:00, 1970-01-16 91:00:00,
1970-01-16 03:00:00], dtype=datetime64[ns])

This happens whether the DatetimeIndex was timezone aware or naive. On the other hand, converting to a list works fine:

list(dr)

[wesm]

This can't be fixed. It's a bug in NumPy 1.6.x