Install pip, virtualenv, and virtualenvwrapper on Mac OS X 10.6.4

I decide to remove the python pre-installed on Mac, and install python with the version I want, so I remove everything under /Library/Frameworks/Python.framework/

Now, install python26, tcl and pip. Tcl is one of Pip's required packages. The port tool is supposed to download and install tcl when I install pip. However, in my case, port took forever to download tcl. So I had to intall tcl separately before installing pip, just to make the dependents resolved.

sudo port install python26
sudo port install tcl
sudo port install py26-pip

To see where these ports are installed, run port location , e.g.:

port location python26

MacPorts are installed as images. See the following quote from MacPorts.org (http://guide.macports.org/chunked/internals.images.html)

"""
MacPorts port by default is not installed into its final or "activated" location, but rather to an intermediate location that is only made available to other ports and end-users after an activation phase that makes hard links of all its files in ${prefix} that point to the port's files in the image repository. Therefore deactivating a port image to install a different version only removes the hard links in ${prefix} pointing to the previous port version's image -- the deactivated port's image is not disturbed.
"""

(If you are curious about where ${prefix} is defined, open /opt/local/etc/macports/macports.conf, and check "prefix".)

To clean work files, distribution files, and archives, run port clean --all . For example:

sudo port clean --all python26

Create a symbolic link to pip-2.6 so that we can simply type pip instead of pip-2.6:

sudo ln -s /opt/local/bin/pip-2.6 /opt/local/bin/pip

If you check pip-2.6, it's a symbolic link that points to /opt/local/Library/Frameworks/Python.framework/Versions/2.6/bin/pip

/opt/local/Library/Frameworks is defined in macports.conf as "frameworks_dir".

Install virtualenv and virtualenvwrapper

sudo pip install virtualenv
sudo pip install virtualenvwrapper

Create a symbolic link to virtualenv

sudo ln -s /opt/local/Library/Frameworks/Python.framework/Versions/2.6/bin/virtualenv /opt/local/bin/virtualenv

Create a directory to hold the virtual environments.

mkdir ~/.virtualenvs

Add the following two lines in ~/.profile

export WORKON_HOME=$HOME/.virtualenvs
source /opt/local/Library/Frameworks/Python.framework/Versions/2.6/bin/virtualenvwrapper.sh

After editing it, reload the startup file (e.g., run: source ~/.profile).

Create a virtual environment, e.g. "testenv".

mkvirtualenv testenv

A directory "testenv" will be created under ~/.virtualenvs, which hosts the isolated virtual environment. It creates a testenv/lib/python2.6/site-packages directory, where any libraries you install will go. In a minute, I will talk how to install libraries in a virtual environment.

The above command will install a virtual environment that inherits any libraries from your existing python installation's directory. If you want to create a environment without inherits, just add --no-site-packages before the virtual environment name. You may need that when you can't change packages under the python installation's site-packages directory. For more information, see --no-site-packages on virtualenv.openplans.org.

It will also install Setuptools for you. If you use the --distribute option, it will install distribute instead.

mkvirtualenv --no-site-packages --distribute testenv

To see all virtual environments, type workon without any option.

workon

Switch to testenv

workon testenv

If you want to delete testenv, you need to deactivate and then remove it.

deactivate
rmvirtualenv testenv

To install a package to a virtual environment, use pip and the -E option.

pip install -E yourvirtualenv yourpackage

Yolk is a convenient tool to list packages used in your virtual environment. Install yolk:

pip install -E testenv yolk

To list packages used in a virtual environment, you need to activate the environment if it's inactive, and type yolk -l

workon testenv
yolk -l

References

• SaltyCrane Blog - Notes on using pip and virtualenv with Django