The original mlabwrap is a Python-Matlab bridge that enables one to execute Matlab code from Python, by automatically starting up a Matlab process in the background and talking to it.

However, mlabwrap requires a compilation step against Matlab, which can be quite an inconvenience. So Dani Valevski rewrote the low-level parts of mlabwrap in Python, and made these available as open source on Google Code.

The code you are looking at now, is simply the pure python mlabwrap code that has been packaged together and slightly improved by Charl Botha.

• More intelligent determination of Matlab binary path and versioning.
• Many bug fixes for scipy 0.12 + matlab 2011b.
• mlab.name will return the value of matlab variable name. This syntax only used to work for invoking functions.
• Error reporting.
• Documentation.
• setuptools configuration, so you can install mlabwrap with a single pip invocation! See below for details.

Install mlabwrap-purepy:

sudo easy_install pip
sudo pip install numpy
sudo pip install scipy
sudo pip install git+https://github.com/cpbotha/mlabwrap-purepy.git

These instructions will install pip, numpy, scipy and mlabwrap-purepy system-wide. If you're hip enough to use virtualenv, you're hip enough to figure out how to do this in a virtualenv. :)

Also see the docstring in mlabwrap.py. It goes something like this:

import mlabwrap
mlab = mlabwrap.init()

If the main matlab executable (or a symlink to it) is not on your system PATH, pass its full path as the first parameter to init. Now you can do:

sorted = mlab.sort([1 2 3])

Or even:

mlab._do("a = [3 2 1]; b = mean(a);", nout=0)
print mlab.b