There should have been more innovation, making Python 3 compelling enough for Numpy use to deprecate Python 2 in 2010 for obvious, noncontroversial technical re

There should have been more innovation, making Python 3 compelling enough for Numpy use to deprecate Python 2 in 2010 for obvious, noncontroversial technical reasons. Instead there has been a thoroughly unpleasant and glacially slow transition between "we have to support Python 3 for bleeding edge users" to "we have to support Python 2 for retro users".