IPython requires Python 2.7 or ≥ 3.3.
If you need to use Python 2.6 or 3.2, you can find IPython 1.0 here.
If you have setuptools, the quickest way to get up and running with IPython is:
$ pip install "ipython[all]"
This will download and install IPython and its main optional dependencies:
To run IPython’s test suite, use the iptest command:
This document describes in detail the steps required to install IPython, and its various optional dependencies. For a few quick ways to get started with package managers or full Python distributions, see the install page of the IPython website.
IPython is organized into a number of subpackages, each of which has its own dependencies. All of the subpackages come with IPython, so you don’t need to download and install them separately. However, to use a given subpackage, you will need to install all of its dependencies.
Please let us know if you have problems installing IPython or any of its dependencies.
IPython and most dependencies can be installed via pip. In many scenarios, this is the simplest method of installing Python packages. More information about pip can be found on its PyPI page.
On Windows, IPython requires setuptools. We hope to change this in the future, but for now on Windows, you must install setuptools to use IPython.
More general information about installing Python packages can be found in Python’s documentation.
Given a properly built Python, the basic interactive IPython shell will work with no external dependencies. However, some Python distributions (particularly on Windows and OS X), don’t come with a working readline module. The IPython shell will work without readline, but will lack many features that users depend on, such as tab completion and command line editing. If you install IPython with setuptools, (e.g. with pip), then the appropriate readline for your platform will be installed. See below for details of how to make sure you have a working readline.
If you have setuptools or pip, the easiest way of getting IPython is to simply use pip:
$ pip install ipython
If you don’t want to use pip, or don’t have it installed, just grab the latest stable build of IPython from here. Then do the following:
$ tar -xzf ipython.tar.gz $ cd ipython $ python setup.py install
If you are installing to a location (like /usr/local) that requires higher permissions, you may need to run the last command with sudo.
As mentioned above, on Windows, IPython requires setuptools, and it also requires the PyReadline library to properly support coloring and keyboard management (features that the default windows console doesn’t have). So on Windows, the installation procedure is:
IPython by default runs in a terminal window, but the normal terminal application supplied by Microsoft Windows is very primitive. You may want to download the excellent and free Console application instead, which is a far superior tool. You can even configure Console to give you by default an IPython tab, which is very convenient to create new IPython sessions directly from the working terminal.
It is also possible to install the development version of IPython from our Git source code repository. To do this you will need to have Git installed on your system. Then just do:
$ git clone --recursive https://github.com/ipython/ipython.git $ cd ipython $ python setup.py install
Some users want to be able to follow the development branch as it changes. If you have setuptools installed, this is easy. Simply replace the last step by:
$ python setupegg.py develop
This creates links in the right places and installs the command line script to the appropriate places.
Then, if you want to update your IPython at any time, just do:
$ git pull
$ git submodule update
$ python setup.py submodule
Another option is to copy git hooks to your ./git/hooks/ directory to ensure that your submodules are up to date on each pull.
There are a number of basic optional dependencies that most users will want to get. These are:
If you are comfortable installing these things yourself, have at it, otherwise read on for more details.
IPython uses several other modules, such as pexpect and path.py, if they are installed on your system, but it can also use bundled versions from IPython.external, so there’s no need to install them separately.
As indicated above, on Windows, to get full functionality in the console version of IPython, PyReadline is needed. PyReadline is a separate, Windows only implementation of readline that uses native Windows calls through ctypes. The easiest way of installing PyReadline is you use the binary installer available here.
On OSX, if you are using the built-in Python shipped by Apple, you will be missing a proper readline implementation as Apple ships instead a library called libedit that provides only some of readline’s functionality. While you may find libedit sufficient, we have occasional reports of bugs with it and several developers who use OS X as their main environment consider libedit unacceptable for productive, regular use with IPython.
Therefore, IPython on OS X depends on the gnureadline module. We will not consider completion/history problems to be bugs for IPython if you are using libedit.
To get a working readline module on OS X, just do (with pip installed):
$ pip install gnureadline
Other Python distributions on OS X (such as Anaconda, fink, MacPorts) already have proper readline so you likely don’t have to do this step.
When IPython is installed with setuptools, (e.g. using the pip command), the correct readline should be installed if you specify the terminal optional dependencies:
$ pip install "ipython[terminal]"
To run the IPython test suite you will need the nose package. Nose provides a great way of sniffing out and running all of the IPython tests. The simplest way of getting nose is to use pip:
$ pip install nose
Another way of getting this is to do:
$ pip install "ipython[test]"
For more installation options, see the nose website.
Once you have nose installed, you can run IPython’s test suite using the iptest command:
IPython.parallel provides a nice architecture for parallel computing, with a focus on fluid interactive workflows. These features require just one package: PyZMQ. See the next section for PyZMQ details.
On a Unix style platform (including OS X), if you want to use setuptools, you can just do:
$ pip install "ipython[zmq]" # will include pyzmq
Security in IPython.parallel is provided by SSH tunnels. By default, Linux and OSX clients will use the shell ssh command, but on Windows, we also support tunneling with paramiko.
IPython 0.11 introduced some new functionality, including a two-process execution model using ZeroMQ for communication. The Python bindings to ZeroMQ are found in the PyZMQ project, which is pip install-able. If you are on Python 2.7, 3.3, or 3.4 on OSX or Windows, pyzmq has eggs and wheels that include ZeroMQ itself.
IPython.kernel.zmq depends on pyzmq >= 2.2.
The IPython notebook is a notebook-style web interface to IPython and can be started with the command ipython notebook.
Like the IPython.parallel and IPython.frontend.qt.console packages, the HTML notebook requires ZeroMQ and PyZMQ.
The IPython notebook uses the Tornado project for its HTTP server. Tornado 2.1 is required, in order to support current versions of browsers, due to an update to the websocket protocol.
A quick and easy method is to install it from a python session:
from IPython.external.mathjax import install_mathjax install_mathjax()
If you need tighter configuration control, you can download your own copy of MathJax from http://www.mathjax.org/download/ - use the MathJax-2.0 link. When you have the file stored locally, install it with:
python -m IPython.external.mathjax /path/to/source/mathjax-MathJax-v2.0-20-g07669ac.zip
For unusual needs, IPython can tell you what directory it wants to find MathJax in:
python -m IPython.external.mathjax -d /some/other/mathjax
By default Mathjax will be installed in your ipython profile directory, but you can make system wide install, please refer to the documentation and helper function of IPython.external.mathjax
The IPython notebook is officially supported on the following browsers:
The is mainly due to the notebook’s usage of WebSockets and the flexible box model.
The following browsers are unsupported:
The following specific combinations are known NOT to work:
There are some early reports that the Notebook works on Internet Explorer 10, but we expect there will be some CSS issues related to the flexible box model.
The most important dependency of nbconvert is Pandoc 1.10 or later, a document format translation program. This is not a Python package, so it cannot be expressed as a regular IPython dependency with setuptools.
To install pandoc on Linux, you can generally use your package manager:
sudo apt-get install pandoc
On other platforms, you can get pandoc from their website.