In terms of storing datetime in pytables...there are a couple ways to go
about it floats and strings, with strings seeming to be preferable for
generic cross-platform use:
1) store as double precision floats - Time64 in pytables - mapping to and
from python datetime's like so:
t1 = time.time()
t2 = datetime.datetime.fromtimestamp(t1)
t3 = mktime(t2.timetuple())+1e-6*t2.microsecond
2) store as ISO string, mapping to and from datetime in ugly way (let me
know if theres is a cleaner way) like so:
t4 = t2.__str__()
t4a, t4b = t4.split(".", 1)
t5 = datetime.datetime.strptime(t4a, "%Y-%m-%d %H:%M:%S")
ms = int(t4b.ljust(6,'0')[:6])
t5 = t5.replace(microsecond=ms)
I found a discussion of the disadvantages of storing as float:
>* datetime isn't keen to use floating-point timestamps, because what
*>* 1053879135.83 means varies across platforms (not all boxes start counting
*at
>* 1970, and boxes disagree about whether leap seconds should be counted),
*and
>* because a float doesn't have enough bits of precision to represent all the
*>* date+time combinations datetime can represent. Support for timestamps in
*>* the datetime module is thus minimal -- we're trying to move away from
*them.
...and to add to that there seems to be no time zone information in the
float representation.
And a worthwhile link... http://seehuhn.de/pages/pdate
On Nov 26, 2007 9:23 PM, Bradford Cross <[EMAIL PROTECTED]> wrote:
>
>
> On Nov 26, 2007 4:14 PM, David Worrall <[EMAIL PROTECTED]> wrote:
> >
> > Numpy's necessity for data homogeneity means developing another level
> > of abstraction/conversion which was more fiddle. And for what gain?
> > I decided to use something like your strategy 2: read the relevant
> > data out of pytables into numpy arrays
> >
>
> This is what I am currently doing as well. Write data to tables and
> create an API that allows access to the data as numpy arrays.
>
>
>
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Pytables-users mailing list
Pytables-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/pytables-users
