Install Django

Problem

You want to install the Django framework to develop an application or web site.

Solution

There are various ways to get the Django framework. You can download Django from its main site https://www.djangoproject.com/download/ and install it like a regular Python application (e.g.python setup.py install). You can also download and install Django via an operating system(OS) package administration tool such as apt-get -- available on Linux distributions like Debian and Ubuntu. Or you can also download and install Django via a Python package administration tool like pip.

How it works

The list of Django installation options including their pros and cons is presented in Table 1.

Table 1 - Django installation options - Pros and Cons
ApproachProsCons
Recommended option
Download/install from Python package manager (pip)
Allows install on virtual Python environment
Dependencies are taken care of automatically
Latest version may not be available
Download from main site (tar.gz)Easiest access to latest stable releaseRequires manual download and install
Requires getting other dependencies
Download from Git
Access to the latest featuresCan contain bugs
Requires getting other dependencies
Download/install from OS package manager (apt-get)Easy to install
Dependencies are taken care of automatically
Latest version may not be available
Installed on global Python environment

As illustrated in table 1, the recommended option to install Django is to use a Python package manager because it provides the most flexibility. Next, I'll describe each of the steps to install Django with this approach. Once I finish these steps, I'll also describe the steps to install Django from Git, which can be helpful if you want to try out the latest Django features.

Install Python (Pre-requisite)

Since Django runs on Python, you first need to install Python. Django 1.11 requires at a minimum Python 2.7 or higher. In addition, Django 1.11 can also run on Python 3.5 or higher, but see the sidenote if you will be using the Python 3 branch.

Note Django and Python 3

Django uses Six (http://pythonhosted.org/six/) to run Python 2 and Python 3 compatible logic. Six is a set of utilities that wraps over differences between Python 2 and Python 3. Even if you're experienced with Python 3 (e.g. know the main differences between writing code for Python 3.x vs. Python 2.x) I recommend you read the following Django/Python 3 specific document which recommends a series of practices which Django itself uses, that help keep your Python application compatiable between Python 2 and Python 3 https://docs.djangoproject.com/en/1.11/topics/python3/ .

If you use a Unix/Linux OS, Python may already be installed on your system. If you type which python on a Unix/Linux terminal and it returns a response (e.g. /usr/bin/python) it indicates the location of the Python executable, if there is no response it indicates the Python executable is not available on the system. If you don't have Python on your system and you're using a Debian or Ubuntu Linux distribution, you can use the OS package manager apt-get to install Python by typing: apt-get install python. If you have a Unix/Linux distribution that is not Debian or Ubuntu and you need to install Python, consult your Unix/Linux documentation for available Python packages or download the Python sources from http://python.org/download/ to do the installation.

If you have a system that runs on a Windows OS or Mac OS, Python installers are available for download from http://python.org/download/.

Irrespective of your system's OS, once you've finished the Python installation, ensure Python is installed correctly and accessible from anywhere on your system. Open a terminal and type python, you should enter a Python interactive session like the one illustrated in listing 1.

Listing 1 - Python interactive session

[user@~]$ python
Python 2.7.12 (default, Nov 19 2016, 06:48:10) 
[GCC 5.4.0 20160609] on linux2
Type "help", "copyright", "credits" or "license" for more information.

If you aren't able to enter a Python interactive session, review the Python installation process because you will not be able to continue with the following sections.

Update [or install] pip package manager (Pre-requisite)

To make Python package installation and management easier, Python uses a package manager called pip. If you're using Python 2.7.9 (or greater 2.x branch) or Python 3.4 (or a greater 3.x branch), pip comes insalled by default. Next, let's upgrade pip on your system as illustrated in listing 2.

Listing 2 - Update pip package manager

[user@~]$ pip install --upgrade pip
Collecting pip
  Downloading pip-9.0.1-py2.py3-none-any.whl (1.3MB)
Installing collected packages: pip
  Found existing installation: pip 8.1.1
Successfully installed pip-9.0.1

As you can see in listing 2, to update pip you invoke the pip executable with the arguments install --upgrade pip. Upon execution, pip searches for a package by the provided name -- in this case pip itself -- downloads it and performs an upgrade in case it's already installed. If the installation output on your system is similar to the one in listing 2 -- without any errors -- you have successfully updated pip.

If you see an error like The program 'pip' is currently not installed or pip not found, it means your Python installation is not equipped with pip. In this case, you'll need to install the pip executable by downloading https://bootstrap.pypa.io/get-pip.py and then executing the downloaded file with the command: python get-pip.py. Once the pip executable is installed, run the pip update procedure from listing 2.

With pip on your system, you're ready to move on to the next step.

Install virtualenv (Optional pre-requisite)

Virtualenv is not essential to develop Django applications, but I highly recommend you use it because it allows you to create virtual Python environments on a single system. By using virtual Python environments, applications can run in their own 'sandbox' in isolation of other Python applications. Initially virtualenv can appear to be of little benefit, but it can be of tremendous help for tasks like replicating a development environment to a production environment and avoiding version conflicts that can arise between different applications.

Without virtualenv you can still proceed to install Django and any other Python package using pip, but the issue is all packages are installed under the global Python installation. Initially this can seem convenient, because you only need to install packages once in the global Python installation. But it's not that convenient if you think about some of the following questions.

What happens if a new Django version is released after your first project and you want to start a second project ? Do you upgrade the first project to run on the new Django version or start the second project as if the new Django version doesn't exist ? The first option requires additional work, while the second option requires you to develop on an outdated Django version. By using virtual Python environments you avoid this problem, because each project can run its own Django version in isolation.

If you consider this potential version conflict for any Python package, you'll realize why I recommend you use virtualenv. Many Python packages have specific version dependencies (e.g. Package A depends on package B version 2.3 and package C version 1.5). If you update a new package with specific cross-dependency versions, it can be very easy to break a Python installation if you're using a global Python installation. With virtualenv you can have multiple Python installations without them interfering with one another.

Now that I've explained the benefits of virtualenv, lets install the virtualenv executable with pip, as show in listing 3.

Listing 3 - Install virtualenv with pip

[user@~]$  pip install virtualenv
Downloading/unpacking virtualenv
  Downloading virtualenv-15.1.0.tar.gz (1.8Mb): 1.8Mb downloaded
  Running setup.py egg_info for package virtualenv   
  Installing collected packages: virtualenv
  Running setup.py install for virtualenv
  Installing virtualenv script to /usr/local/bin
  Installing virtualenv-2.7 script to /usr/local/bin
Successfully installed virtualenv
Cleaning up...

As illustrated in listing 3, pip automatically downloads and installs the requested package. Similar to the pip package executable, a virtualenv executable is also installed that should be accessible from anywhere on your system. The virtualenv executable allows you to create virtual Python environments. Listing 4 illustrates how to create a virtual Python environment with virtualenv.

Listing 4 - Create virtual Python environment with virtualenv

[user@~]$ virtualenv --no-site-packages mydjangosandbox
Running virtualenv with interpreter /usr/bin/python2
New python executable in mydjangosandbox/bin/python2
Also creating executable in mydjangosandbox/bin/python
Installing setuptools, pkg_resources, pip, wheel...done.

The virtualenv executable accepts several parameters. The task in listing 4 makes use of the parameter --no-site-packages, which tells virtualenv to create a virtual Python environment without any of the packages already installed in the global Python installation. This creates a virtual Python environment like the one you had when you made the initial Python global installation. Following the virtualenv parameters, you only need to specify an argument for the name of the virtual Python environment, which in the case of listing 4 is mydjangosandbox. Upon execution, virtualenv creates a directory with the virtual Python environment whose contents are illustrated in listing 5.

Listing 5 - Virtual Python environment directory structure

+<virtual_environment_name>
|
|
+---+-<bin>
|   |
|   +-activate
|   +-easy_install
|   +-pip
|   +-python
|   +-wheel
|
+---+-<include>
|
+---+-<lib>
|
+---+-<local>
|
+---+-<share>

As illustrated in listing 5, the virtual Python environment has a similar directory structure to the global Python installation. The bin directory contains executables for the virtual environment, the include directory is linked to the global Python installation header files, the lib directory is a copy of the global Python installation libraries and where packages for the virtual environment are installed, the local directory is linked to top level directories of the virtual directory to simulate a Python installation and the share directory is used to place shared Python packages.

The most important part of the virtual environment are the executables under the bin directory. If you use any of these executables, such as pip, easy_install, python or wheel they execute under the context of the virtual Python environment. For example, the pip under bin installs packages for the virtual environment. Similarly, an application that runs on the python executable under bin is only able to load packages installed on the virtual Python environment. This is the 'sandbox' behavior I mentioned previously.

Even though access to different virtual Python environments and executables is a powerful feature, having different pip and python executables for multiple virtual Python environments and the global Python installation can become confusing due to long access paths and relative paths. For this reason, virtualenv has a mechanism to load virtual environments so that if you execute pip, python or any other executable from anywhere on your system, the executables from a selected virtual environments are used (instead of the default global Python installation executables). This is achieved with the activate executable inside the bin directory, a process illustrated in listing 6.

Listing 6 - Activate virtual Python environment

[user@~]$ source ./bin/activate
[(mydjangosandbox)user@~] $ 
# NOTE: source is a Unix/Linux specific command, for other OS just execute activate

Notice in listing 6 how after invoking the activate executable, the command prompt adds the virtual environment name between parenthesis. This means the executables under the bin directory of the virtual Python environment mydjangosandbox are used over those in the global Python installation. To exit a virtual Python environment just type deactivate and you fallback to using the global Python installation executables.

As you've now learned, virtualenv works transparently allowing you to maintain different Python installations each with its own set of executables like the main python interpreter and the pip package manager. You only need to take care of switching between virtual environments so you install and run Python applications in the appropriate virtual environment.

In future sections I won't make many references to virtualenv since it's isn't directly related to Django. Though I recommend you use it, I'll leave it up to you if you want to keep using the global Python installation python and pip executables for everything or if you prefer to keep virtual Python environments with their own executables to make Python application management easier.

Install Django

Once you have all the previous tools working on your system, the actual Django installation is very simple. Listing 7 illustrates how to install Django using pip

Listing 7 - Install Django with pip

[user@~]$ pip install Django==1.11
Downloading/unpacking Django==1.11
  Downloading Django-1.11.tar.gz (6.9MB): 6.9MB downloaded
  Running setup.py egg_info for package Django
    
Installing collected packages: Django
  Running setup.py install for Django
    changing mode of build/scripts-2.7/django-admin.py from 664 to 775
    changing mode of /python/mydjangosandbox/bin/django-admin.py to 775
Successfully installed Django
Cleaning up...

The pip install task in listing 7 uses a syntax to install a specific package version. The syntax Django==1.11 tells pip to download and install the Django 1.11 version. With this same syntax you can install an older Django version. If you don't specify a package version, pip downloads and installs the most recent available version for the specified package.

Sometimes a Django release may take a few days to become available through pip, in which case you'll receive an error. In such cases you can download the release directly from the Django main site at https://www.djangoproject.com/download/. Once you download the release file in tar.gz format, you can use pip to make the installation as illustrated in listing 8.

Listing 8 - Install Django from local tar.gz file with pip

[user@~]$ pip install /home/Downloads/Django-1.11.tar.gz 
Unpacking /home/Downloads/Django-1.11.tar.gz
  Running setup.py egg_info for package from file:///home/Downloads/Django-1.11.tar.gz
    
Installing collected packages: Django
  Running setup.py install for Django
    changing mode of build/scripts-2.7/django-admin.py from 664 to 775
    
    changing mode of /python/mydjangosandbox/bin/django-admin.py to 775
Successfully installed Django
Cleaning up...

Notice in listing 8 how pip is capable of installing Python packages directly from a compressed file on the local file system.

Install Django from Git

If you want to use the most recent functionality in Django, then you'll need to install Django from its Git repository. The Git repository contains the latest changes made to Django. Even though the Django Git version can be unstable, it's the only way to develop with the newest Django features or get bug-fixes for problems that aren't yet available in public releases.

Note You need to install Git

You need to install Git to execute the following tasks. You can download Git for several OS at http://git-scm.com/

Just like the prior pip installation examples, pip is sufficiently flexible to make a Django installation from Git. There are two alternatives to use pip with Git. You can provide the remote Django Git repository, in which case pip clones the repository locally and discards it after the installation, as illustrated listing 9. Or you can clone the Django Git repository locally -- where you'll be able to make modifications at a later time -- and then run pip to do the installation, as illustrated in listing 10.

Listing 9 - Install Django from remote Git with pip

[user@~]$ pip install git+https://github.com/django/django.git
Downloading/unpacking git+https://github.com/django/django.git
  Cloning https://github.com/django/django.git to /tmp/pip-AgZIgk-build
  Running setup.py egg_info for package from git+https://github.com/django/django.git
  Installing collected packages: Django
  Running setup.py install for Django
   changing mode of build/scripts-2.7/django-admin.py from 664 to 775
   changing mode of /python/mydjangosandbox/bin/django-admin.py to 775
Successfully installed Django
Cleaning up...
Listing 10 - Download Django from Git and install locally with pip

[user@~]$ git clone https://github.com/django/django.git 
Cloning into django...
remote: Reusing existing pack: 265136, done.
remote: Counting objects: 42, done.
remote: Compressing objects: 100% (42/42), done.
remote: Total 265178 (delta 15), reused 2 (delta 0)
Receiving objects: 100% (265178/265178), 98.02 MiB | 1.30 MiB/s, done.
Resolving deltas: 100% (190025/190025), done.
# Assuming Django Git download made to /home/Downloads/django/
[user@~]$ pip install /home/Downloads/django/
Unpacking /home/Downloads/django
  Running setup.py egg_info for package from file:///tmp/django
  Installing collected packages: Django
  Running setup.py install for Django
    changing mode of build/scripts-2.7/django-admin.py from 664 to 775
    changing mode of /python/mydjangosandbox/bin/django-admin.py to 775
Successfully installed Django
Cleaning up...

Notice in listing 9 the syntax to download a remote Git repository is git+ followed by the remote Git location. In this case https://github.com/django/django.git represents the Django Git repository. In listing 10 the Django Git repository is cloned locally first and then pip is executed with the argument of the local Git repository directory.