This guide assumes a default Linux install and a default Maya installation (I will use Centos 7 and Maya 2016.5 x64 for this guide).
The most basic setup you need is a system install of Python 2.7.x. The exact x
number is not very important; the important part is that it is a release of
2.7. Python 2.7 is included by default on nearly all systems, so this is likely
already set up for you. Accompanying this install is a small tool called
virtualenv, a vital ingredient for our plans. You can check for virtualenv
by calling it in the terminal with virtualenv --help. Install it if you do
not have it.
Create a virtual environment by calling virtualenv ~/.virtualenvs/mayapy in
any directory. The terminal will give some feedback about installing setuptools
and pip, then it finishes. The virtual environment has been set up with the
system install of Python 2.7 and is ready to be activated.
Navigate to your virtual environment root: ~/.virtualenvs/mayapy and source
the activation script using source bin/activate.
Congratulations! You now have an isolated environment to test packages for Maya. Using Numpy as an example, I will show how you can utilize it.
Numpy is a combination Python/C library for computation-heavy math. Because it is partly written in C, it is more difficult to distribute and use in non-standard locations. This walkthrough describes how I use Numpy in Maya.
Installing Numpy is surprisingly simple with pip, Python’s package manager.
Simply run pip install numpy and it will download and compile the package
based on the Python interpreter you set up in the virtual environment. Enter
into a Python interpreter by calling python in your activated environment.
Call import numpy to import the Numpy library. If it gives no feedback, it
imported without error. Great!
Say you want to use Numpy in Maya for writing a tool. If you copy the numpy
library over to your Maya scripts directory, you would expect Maya to find
the library and import it just as in your virtual environment. However, Maya
will spit out an error saying a library could not be loaded. Without moving
Numpy, we can fix this with one line:
import os, site; site.addsitedir(os.path.expanduser("~/.virtualenvs/mayapy/lib/python2.7/site-packages"))
Importing Numpy should result in no output. You now have a fast math library
for use in Maya and your scripts. Any tool that requires Numpy will have
access to Numpy through this path, so there is no need to have multiple copies
of Numpy floating around.
Follow the same instructions as with Numpy.
The next logical step is to add this path to your Maya configuration using
userSetup.py. Copy the same code line to the top of userSetup.py and it
is now available on Maya startup.
It is important to note an added benefit to using site-packages from the virtual environment: you can develop in the virtual environment as you would normally and access all the same libraries within Maya.
To use with a single tool, make a ‘deps’ folder in the tool root.
tool_root/
deps/
numpy/
scipy/
...
__init__.py
...
In __init__.py add the line:
import os, site; site.addsitedir(os.path.join(os.path.dirname(__file__), 'deps'))
virtualenv ~/.virtualenvs/mayapy to create a new virtual environment.cd ~/.virtualenvs/mayapy && source bin/activate to activate the
environment.pip install numpy to install Numpy.import os, site;
site.addsitedir(os.path.expanduser("~/.virtualenvs/mayapy/lib/python2.7/site-packages"))
at the top of userSetup.pyHave questions/comments/revision? Leave a report or pull request on Github.