Re: [Numpy-discussion] Towards using scons in numpy: 1st step