[Numpy-discussion] RFC: A proposal for implementing some date/time types in NumPy

[Francesc Alted]

A Friday 11 July 2008, Pierre GM escrigué:
> On Fri, Jul 11, 2008 at 12:47 PM, Francesc Alted
> <falted@pytables.org>
>
> wrote:
> > A Friday 11 July 2008, Pierre GM escrigué:
> > > Our approach for dealing with dates was to translate them into
> > > integers through a particular class (Date).
> >
> > That's very interesting.  We will have a look at your
> > implementation and see if we can reuse code/ideas.  I suppose this
> > code in your TimeSeries module, right?
>
> Yes, in the src directory (c_dates.c).
> In addition, we have a DateArray class, which implements an array of
> Dates (you didn't see that coming...) and whose dtype is int64.
>
> The reason why we wanted to stick to ints was to permit a direct
> correspondence index<->date in an array. If you know the first date
> and the timestep, you can get the date corresponding to any element
> of your series, and vice-versa.

Ah!  Very smart!  I wonder if we could use this to implement a special 
array with a fixed timestep that could be indexed by time instead than 
by index.  Something like:

t1 = datetime.datetime(1,2,3)
t2 = datetime.datetime(3,4,5)

and then:

arr[numpy.datetime(t1):numpy.datetime(t2)]

would select the events between t1 and t2 timestamps.

Powerful!  But that would introduce more complexity and besides this is 
not directly related with our goal.  Interesting anyway.

-- 
Francesc Alted