Well, it’s a bit late for posting this, considering I have been playing with Python for few months as I posted before, but, I recently reinstalled my Windows, so I need to setup everything once again. This time, I am trying to do it the right way, and my previous Python setup isn’t really completely setup anyway, since I did most of my python related works on my Linux Mint install. But recently, I have been spending more time on Windows because of various things, so, well, here goes. The distribution system in Python is a little but complicated, but you can read more about it here and here.

Basically, the standard Python distribution utilities is the distutils, with it, you can install from the package source with python setup.py install. But there some problems with it, one of them is that it has no way to track dependencies between packages, so there are other tools that try to solve that problem, one of which is the setuptools. When you install the setuptools, it allows you to package your scripts and specify any dependencies with it, which can be installed via its distribution system, easy_install which allows you to install packages with easy_install package_name. But, the setuptools introduces another set of problem, where it causes weird behaviour and some packages unusable unless it is installed via setuptools. So, the recommended way to package and distribute python package is to use plain vanilla distutils to package the scripts, and use pip for distribution. Instead of specifying dependencies in the setup.py file, it allows you to create a separate file to specify the dependencies, and process that instead. There are also distribute which is a fork of setuptools that tries to solve some of its shortcomings.

Ok, now that we got clear, lets actually start setting up python, the first one is installing Python itself, simple enough, download the MSI installer, run it, and Python is now installed. The next step is adding the python directory to the PATH environment variables. This is so that you can run the various python commands without typing the full path. The first one to add is the Python root directory (C:\Python27 in my case), and also the Scripts directory (C:\Python27\Scripts in my case). I didn’t add the second one to the PATH before, I wondered why can’t I run commands such as pip and virtualenv directly before. and another thing, I recently watched a video from pycon 2012 talking about the state of Python on Windows, and I think the speaker mentioned something about the new releases of Python installers for Windows includes the option to add the environment variables and also have some of the directory structure changed, but as I only watched the video briefly, I am not sure whether it only applies to the new Python 3.x release or to the 2.x release as well. I think I will just update it later.

By now, you can say that Python is already completely setup on Windows, but there are several other packages that should be installed.

The first one is to get a python package manager, which can be used to install and manage the installed packages on your system. There are several options you can choose, I am using distribute together with pip. To install distribute, the easiest way is download the distribute_setup.py, and run it. To install pip, you can download the zip files from http://pypi.python.org/pypi/pip, extract it, navigate to the folder, and run python setup.py install.

The next thing you should do is to install virtualenv. virtualenv is basically a little python utility to setup a virtual isolated python installation apart from your main python installation. With it, you can install different version of the same libraries at the same time without having them conflicts with each other. To install it, the easiest would be to use pip that you installed before, so, run pip install virtualenv.

After you install it, you can run the command virtualenv ENV to create a new virtualenv named ENV. Activate it by going into ENV\scripts\activate which should change your shell prompt accordingly. Now any new packages changes will be made locally to the active virtualenv without affecting the global installation. After you are done, you can deactivate it and switch back to the main python installation by running ENV\scripts\deactivate.bat.

Next thing to do is to install django itself. Make sure your virtualenv is active, unless you want django to be installed to the main django installation. The easiest way to install django is to use pip: pip install django.

And, you’re done! easy, isn’t it? You can test it by running the python shell, and type import django. If there is no error, then you’re good to go. Now, if you installed your django into your virtualenv, trying to run django-admin.py might gives you an ImportError saying the it can’t find the django.core module. This is because the .py file extensions are associated with the main Python installation in your computer, and running django-admin.py directly from the command line will cause it to run the script with that Python installation which doesn’t have django installed, so, what you can do is either run the script with python manually by specify the path of the django-admin.py, or you can modify your virtualenv activate script to change the file association, either way, its up to you.