I would avoid drange because
I believe numpy will soon be implementing such a function and I
wouldn't want the possible confusion that comes from that.
what do you mean with "avoiding drange"? You wouldn't use it in your
code? Or do you suggest, matplotlib should not have the function "drange"?
I mean that matplotlib probably shouldn't have drange (although, this
is just IMHO).
I have yet another question: Does someone know, why matplotlib.dates
calculates dates in floats, where 1 = 1day?
I think with Long integers without floating point much of the
rounding-issues could be avoided.
We are most certainly "doing it wrong" as the current code was merely
a hack due to the lack of such features in numpy. The next numpy
release should have this, but we will still need the current hacks for
a little while longer due to support for previous versions of numpy.
Maybe we ought to look into some sort of logic that would utilize
numpy's drange if it is available and fall back to ours otherwise?
On Friday, July 8, 2011, Maximilian Trescher <faucon@...929...> wrote: