location of ma in maskedarray branch
I noticed that the new masked array module resides in numpy/ma in the maskedarray branch as opposed to numpy/core/ma like it does in the current trunk. Was this intentional? Code that explicitly imports ma from the core subfolder will break from this change (like the __init__.py script for the ma subfolder in matplotlib for example). - Matt
Matt Knox wrote:
I noticed that the new masked array module resides in numpy/ma in the maskedarray branch as opposed to numpy/core/ma like it does in the current trunk. Was this intentional? Code that explicitly imports ma from the core subfolder will break from this change (like the __init__.py script for the ma subfolder in matplotlib for example).
I modified mpl to handle this new location, assuming it was intentional and likely to remain. The change is included in release 0.91.2 as well as in the svn trunk (formerly the transforms branch) and the svn maintenance branch derived from 0.91.2). The previous location is still supported also. Eric
Hi Matt On Tue, Jan 22, 2008 at 10:37:29PM -0500, Matt Knox wrote:
I noticed that the new masked array module resides in numpy/ma in the maskedarray branch as opposed to numpy/core/ma like it does in the current trunk. Was this intentional? Code that explicitly imports ma from the core subfolder will break from this change (like the __init__.py script for the ma subfolder in matplotlib for example).
Yes, the move was intentional and was discussed beforehand. I may be mistaken, but as far as I know numpy.core is not a public API, so code should rather do from numpy import ma (that works with both the current trunk and the maskedarray branch). Regards Stéfan
participants (3)
-
Eric Firing -
Matt Knox -
Stefan van der Walt