[Numpy-discussion] Maskedarray implementations

[Matt Knox]

I think it's reasonably safe to say at this point that most people are in 
favor of the new maskedarray implementation becoming the default numpy.ma at 
some point in the future. So the question is, when/how will the migration 
process be done?

 - just swap the whole thing as is and hope for the best?
 - start a numpy 1.1 branch and put it in there as a replacement for numpy.ma?
 - put it in numpy as a separate module from numpy.ma initially? 
(eg. "numpy.ma_new" ?)
 - some other approach?

The first option would be perfectly fine by me since I don't use the standard 
numpy.ma anyway, but I suspect some people might have a problem with that. So 
what is the best way to do this?

- Matt