kwgoodman

Oh, wait, sorry. We'd of course just code the label support in larry.rollaxis and then just call np.rollaxis on the data part of the larry. We could copy the docstring...

Here's a quick hack that borrows heavily from np.rollaxis. Does it behave the way you want? ``` import numpy as np def rollaxis(lar, axis, start=0): lar.x = np.rollaxis(lar.x, axis, start)...

I guess my hack doesn't behave the same way as np.rollaxis: ``` >> a = np.ones((3,4,5,6)) >> b = np.rollaxis(a, 1, 3) >> a.shape (3, 4, 5, 6) >> b.shape...

Sorry for the rapid-fire comments and corresponding bugs and mistakes. I'm trying to crank something out quickly. Second attempt: ``` import numpy as np import la def rollaxis(lar, axis, start=0):...