[Numpy-discussion] timeseries, the sandbox (was timeseries - dates prior to 1970)

Pierre GM pgmdevlist@gmail....
Mon Jun 21 06:19:02 CDT 2010


On Jun 20, 2010, at 7:43 PM, Pierre GM wrote:
> On Jun 20, 2010, at 7:28 PM, Bevan Jenkins wrote:
>> 
>> I downloaded the git branch,
<FYI> http://github.com/pierregm/scikits.timeseries-sandbox </FYI>

>> I had to use the Download 
>> pierregm/scikits.timeseries-sandbox at master via a zip file as I have not 
>> managed to get Git to work at Uni yet.
>> After deleting the old scikit I built and installed the new version but I am 
>> still getting the same error.
> 
> Note that it's a work in progress. I'll let you know when the problem you experienced is fixed.

I just pushed the weekend's changes, including a fix for your problem. 
Now it should be possible to define an array of dates w/ a timestep different from 1:
>>> date_array(start_date=Date("T", "2010-06-21 00:00"), length=4, timestep=15)
DateArray([21-Jun-2010 00:00, 21-Jun-2010 00:15, 21-Jun-2010 00:30,
       21-Jun-2010 00:45],
          freq='T')
Feedback is essential at this point...

>> I notoiced you had some new tests to dealing with high- freq dates before the 
>> unix epoch.  When i run
>> ts.test() i get
>> Ran 0 tests in 0.000s
>> 
>> OK
>> 
>> So I imagine I am doing something incorrectly. 
> 
> Not necessarily. It works on my side (but I install using python setup.py develop...), I'll try and see where it goes wrong.

In the meantime, you can still run nosetests on the directory where you downloaded the sources...



More information about the NumPy-Discussion mailing list