Improve repr for weekly Periods

[petergx]

It's totally possible I don't understand how to_timestamp should be used with a Period. But, I have a hunch there is a bug here.

In : p = Period('01-Jan-2012', 'W')
In : p.to_timestamp(how='E')
Out: <Timestamp: 2012-01-01 00:00:00>

In : p.to_timestamp('H', how='E')
Out: <Timestamp: 2012-01-01 23:00:00>

(found in 0.8b2)

[wesm]

I should improve the __repr__ for weekly to show the whole timespan.

In [18]: p = Period('01-Jan-2012', 'W')

In [19]: p.asfreq('D', 's')
Out[19]: Period('26-Dec-2011', 'D')

In [20]: p.asfreq('D', 'e')
Out[20]: Period('01-Jan-2012', 'D')

In [21]: p.asfreq('H', 's')
Out[21]: Period('26-Dec-2011 00:00', 'H')

In [22]: p.asfreq('H', 'e')
Out[22]: Period('01-Jan-2012 23:00', 'H')

However, p.to_timestamp(how='e') should possibly raise an exception, have to think about it

[petergx]

Ahh my bad. I think an improved __repr__ would be cool. Tho.. my confusion initially stemmed from the fact:

p.start_time == p.end_time == p.to_timestamp()