[Numpy-discussion] compress() signature

[Christoph T. Weidemann]

The signature for numpy.compress() seems somewhat inconsistent with
other functions, because the first argument it expects is not the
array but the condition. This tends to throw me off when I use this
function, because the common order for other functions seems to be:
1) input array
2) function specific argument(s) if applicable
3) axis
4) out (if applicable)
5) optional function specific argument(s) if applicable

I think it would be nice to change the signature of compress() to
bring it in line with the other functions, but obviously such a change
would break some code without added functionality. What do people here
think?