easier installation of Neurokernel dependencies using conda
Creation of an environment in which to install and run Neurokernel has been
somewhat tedious to date because of the number of both Python and non-Python
packages required by the software, the need for several manual steps to
compensate for setuptools' less-than-robust dependency handling, and the need
for certain packages containing binary extensions (e.g., pycuda, mpi4py) to be
properly compiled and linked.
To make Neurokernel installation easier, you can now download/install all of
Neurokernel's dependencies (apart from CUDA) using conda , a nice open-source
package/environment manager developed by Continuum Analytics that can handle
both Python and non-Python software; instructions are available in the updated
Neurokernel README file. Some of the packages will only work on 64-bit Ubuntu
14.04 - if you need packages for other distributions, let us know!
As always, feel free to post questions here on the mailing list and bug
reports on the GitHub issue tracker.