Example:
In [40]: array([1, 2], dtype('>l')).dtype
Out[40]: dtype('>i8')
In [41]: array([1, 2], dtype('<l')).dtype
Out[41]: dtype('int64')
In [42]: array([1, 2], dtype('>q')).dtype
Out[42]: dtype('>i8')
In [43]: array([1, 2], dtype('<q')).dtype
Out[43]: dtype('int64')
In [44]: array([1, 2], dtype('>M8[D]')).dtype
Out[44]: dtype('>M8[D]')
In [45]: array([1, 2], dtype('<M8[D]')).dtype
Out[45]: dtype('<M8[D]')
I rather like the '>i8' form with '<' if appropriate, although it's
probably a bit late to do anything about it. The datetime requires a patch
to generate the swapped form. Long term, it might be good to strive for
consistency. Thoughts?
Chuck.
_______________________________________________
NumPy-Discussion mailing list
[email protected]
http://mail.scipy.org/mailman/listinfo/numpy-discussion
