IJulia

IJulia is a Julia-language backend combined with the Jupyter interactive environment (also used by IPython). This combination allows you to interact with the Julia language using Jupyter/IPython's powerful graphical notebook, which combines code, formatted text, math, and multimedia in a single document.

(IJulia notebooks can also be re-used in other Julia code via the NBInclude package.)

Installation

First, download Julia version 0.7 or later and run the installer. Then run the Julia application (double-click on it); a window with a julia> prompt will appear. At the prompt, type:

using Pkg
Pkg.add("IJulia")

to install IJulia.

By default on Mac and Windows, the Pkg.add process will use the Conda.jl package to install a minimal Python+Jupyter distribution (via Miniconda) that is private to Julia (not in your PATH). (You can use using IJulia followed by IJulia.jupyter to find the location jupyter where was installed.) On Linux, it defaults to looking for jupyter in your PATH first, and only installs the Conda Jupyter if that fails; you can force it to use Conda on Linux by setting ENV["JUPYTER"]="" first (see below). (In a Debian or Ubuntu GNU/Linux system, install the package jupyter-client first to use the system jupyter.)

Alternatively, you can install Jupyter (or IPython 3 or later) yourself before adding the IJulia package. To tell IJulia to use your own jupyter installation, you need to set ENV["JUPYTER"] to the path of the jupyter program before running Pkg.add("IJulia") (if jupyter is in your PATH, you can also just use ENV["JUPYTER"]="jupyter"). Alternatively, you can change which jupyter program IJulia is configured with by setting ENV["JUPYTER"] and then running Pkg.build("IJulia").

The simplest way to install Jupyter yourself on Mac and Windows, other than using Julia's Conda distro, is to download the Anaconda package and run its installer. (We recommend that you not use Enthought Canopy/EPD since that can cause problems with the PyCall package.)

On subsequent builds (e.g. when IJulia is updated via Pkg.update), it will use the same jupyter program by default, unless you override it by setting the JUPYTER environment variable, or delete the file joinpath(Pkg.dir("IJulia"), "deps", "JUPYTER"). You can go back to using the Conda jupyter by setting ENV["JUPYTER"]="" and re-running Pkg.build("IJulia").

Running the IJulia Notebook

In Julia, at the julia> prompt, you can type

using IJulia
notebook()

to launch the IJulia notebook in your browser. You can use notebook(detached=true) to launch a notebook server in the background that will persist even when you quit Julia. This is also useful if you want to keep using the current Julia session instead of opening a new one.

julia> using IJulia; notebook(detached=true)
Process(`'C:\Users\JuliaUser\.julia\v0.7\Conda\deps\usr\Scripts\jupyter' notebook`, ProcessRunning)

julia>

By default, the notebook "dashboard" opens in your home directory (homedir()), but you can open the dashboard in a different directory with notebook(dir="/some/path").

Alternatively, you can run

jupyter notebook

from the command line (the Terminal program in MacOS or the Command Prompt in Windows). Note that if you installed jupyter via automated Miniconda installer in Pkg.add, above, then jupyter may not be in your PATH; type import Conda; Conda.SCRIPTDIR in Julia to find out where Conda installed jupyter.

A "dashboard" window like this should open in your web browser. Click on the New button and choose the Julia option to start a new "notebook". A notebook will combine code, computed results, formatted text, and images, just as in IPython. You can enter multiline input cells and execute them with shift-ENTER, and the menu items are mostly self-explanatory. Refer to the IPython documentation for more information, and see also the "Help" menu in the notebook itself.

Given an IJulia notebook file, you can execute its code within any other Julia file (including another notebook) via the NBInclude package.

Updating Julia and IJulia

Julia is improving rapidly, so it won't be long before you want to update to a more recent version. To update the packages only, keeping Julia itself the same, just run:

Pkg.update()

at the Julia prompt (or in IJulia).

If you download and install a new version of Julia from the Julia web site, you will also probably want to update the packages with Pkg.update() (in case newer versions of the packages are required for the most recent Julia). In any case, if you install a new Julia binary (or do anything that changes the location of Julia on your computer), you must update the IJulia installation (to tell Jupyter where to find the new Julia) by running

Pkg.build("IJulia")

at the Julia command line (important: not in IJulia).

Installing additional Julia kernels

You can also install additional Julia kernels, for example, to pass alternative command-line arguments to the julia executable, by using the IJulia.installkernel function. See the help for this function (? IJulia.installkernel in Julia) for complete details.

For example, if you want to run Julia with all deprecation warnings disabled, you can do:

using IJulia
IJulia.installkernel("Julia nodeps", "--depwarn=no")

and a kernel called Julia nodeps 0.7 (if you are using Julia 0.7) will be installed (will show up in your main Jupyter kernel menu) that lets you open notebooks with this flag.

Troubleshooting:

• If you ran into a problem with the above steps, after fixing the problem you can type Pkg.build() to try to rerun the install scripts.
• If you tried it a while ago, try running Pkg.update() and try again: this will fetch the latest versions of the Julia packages in case the problem you saw was fixed. Run Pkg.build("IJulia") if your Julia version may have changed. If this doesn't work, you could try just deleting the whole .julia directory in your home directory (on Windows, it is called Users\USERNAME\.julia in your home directory) via rm(Pkg.dir(),recursive=true) in Julia and re-adding the packages.
• On MacOS, you currently need MacOS 10.7 or later; MacOS 10.6 doesn't work (unless you compile Julia yourself, from source code).
• Internet Explorer 8 (the default in Windows 7) or 9 don't work with the notebook; use Firefox (6 or later) or Chrome (13 or later). Internet Explorer 10 in Windows 8 works (albeit with a few rendering glitches), but Chrome or Firefox is better.
• If the notebook opens up, but doesn't respond (the input label is In[*] indefinitely), try creating a new Python notebook (not Julia) from the New button in the Jupyter dashboard, to see if 1+1 works in Python. If it is the same problem, then probably you have a firewall running on your machine (this is common on Windows) and you need to disable the firewall or at least to allow the IP address 127.0.0.1. (For the Sophos endpoint security software, go to "Configure Anti-Virus and HIPS", select "Authorization" and then "Websites", and add 127.0.0.1 to "Authorized websites"; finally, restart your computer.)
• Try running jupyter --version and make sure that it prints 3.0.0 or larger; earlier versions of IPython are no longer supported by IJulia.
• You can try setting ENV["JUPYTER"]=""; Pkg.build("IJulia") to force IJulia to go back to its own Conda-based Jupyter version (if you previously tried a different jupyter).

IJulia features

There are various features of IJulia that allow you to interact with a running IJulia kernel.

Detecting that code is running under IJulia

If your code needs to detect whether it is running in an IJulia notebook (or other Jupyter client), it can check isdefined(Main, :IJulia) && Main.IJulia.inited.

Customizing your IJulia environment

If you want to run code every time you start IJulia---but only when in IJulia---add a startup_ijulia.jl file to your Julia config directory, e.g., ~/.julia/config/startup_ijulia.jl.

Julia and IPython Magics

One difference from IPython is that the IJulia kernel does not use "magics", which are special commands prefixed with % or %% to execute code in a different language. Instead, other syntaxes to accomplish the same goals are more natural in Julia, work in environments outside of IJulia code cells, and are often more powerful.

However, if you enter an IPython magic command in an IJulia code cell, it will print help explaining how to achieve a similar effect in Julia if possible. For example, the analogue of IPython's %load filename in IJulia is IJulia.load("filename").

Prompting for user input

When you are running in a notebook, ordinary I/O functions on stdin do not function. However, you can prompt for the user to enter a string in one of two ways:

• readline() and readline(stdin) both open a stdin> prompt widget where the user can enter a string, which is returned by readline.

• IJulia.readprompt(prompt) displays the prompt string prompt and returns a string entered by the user. IJulia.readprompt(prompt, password=true) does the same thing but hides the text the user types.

Clearing output

Analogous to the IPython.display.clear_output() function in IPython, IJulia provides a function:

IJulia.clear_output(wait=false)

to clear the output from the current input cell. If the optional wait argument is true, then the front-end waits to clear the output until a new output is available to replace it (to minimize flickering). This is useful to make simple animations, via repeated calls to IJulia.clear_output(true) followed by calls to display(...) to display a new animation frame.

Default display size

When Julia displays a large data structure such as a matrix, by default it truncates the display to a given number of lines and columns. In IJulia, this truncation is to 30 lines and 80 columns by default. You can change this default by the LINES and COLUMNS environment variables, respectively, which can also be changed within IJulia via ENV (e.g. ENV["LINES"] = 60). (Like in the REPL, you can also display non-truncated data structures via print(x).)

Preventing truncation of output

The new default behavior of IJulia is to truncate stdout (via show or println) after 512kb. This to prevent browsers from getting bogged down when displaying the results. This limit can be increased to a custom value, like 1MB, as follows

IJulia.set_max_stdio(1 << 20)

Setting the current module

The module that code in an input cell is evaluated in can be set using Main.IJulia.set_current_module(::Module). It defaults to Main.

Opting out of soft scope

By default, IJulia evaluates user code using "soft" global scope, via the SoftGlobalScope.jl package: this means that you don't need explicit global declarations to modify global variables in for loops and similar, which is convenient for interactive use.

To opt out of this behavior, making notebooks behave similarly to global code in Julia .jl files, you can set IJulia.SOFTSCOPE[] = false at runtime, or include the environment variable IJULIA_SOFTSCOPE=no environment of the IJulia kernel when it is launched.

Low-level Information

Using older IPython versions

While we strongly recommend using IPython version 3 or later (note that this has nothing to do with whether you use Python version 2 or 3), we recognize that in the short term some users may need to continue using IPython 2.x. You can do this by checkout out the ipython2 branch of the IJulia package:

Pkg.checkout("IJulia", "ipython2")
Pkg.build("IJulia")

Manual installation of IPython

First, you will need to install a few prerequisites:

• You need version 3.0 or later of IPython, or version 4 or later of Jupyter. Note that IPython 3.0 was released in February 2015, so if you have an older operating system you may have to install IPython manually. On Mac and Windows systems, it is currently easiest to use the Anaconda Python installer.

• To use the IPython notebook interface, which runs in your web browser and provides a rich multimedia environment, you will need to install the jsonschema, Jinja2, Tornado, and pyzmq (requires apt-get install libzmq-dev and possibly pip install --upgrade --force-reinstall pyzmq on Ubuntu if you are using pip) Python packages. (Given the pip installer, pip install jsonschema jinja2 tornado pyzmq should normally be sufficient.) These should have been automatically installed if you installed IPython itself via easy_install or pip.

• To use the IPython qtconsole interface, you will need to install PyQt4 or PySide.

• You need Julia version 0.7 or later.

Once IPython 3.0+ and Julia 0.7+ are installed, you can install IJulia from a Julia console by typing:

Pkg.add("IJulia")

This will download IJulia and a few other prerequisites, and will set up a Julia kernel for IPython.

If the command above returns an error, you may need to run Pkg.update(), then retry it, or possibly run Pkg.build("IJulia") to force a rebuild.

Other IPython interfaces

Most people will use the notebook (browser-based) interface, but you can also use the IPython qtconsole or IPython terminal interfaces by running ipython qtconsole --kernel julia-0.7 or ipython console --kernel julia-0.7, respectively. (Replace 0.7 with whatever major Julia version you are using.)

Debugging IJulia problems

If IJulia is crashing (e.g. it gives you a "kernel appears to have died" message), you can modify it to print more descriptive error messages to the terminal by doing:

ENV["IJULIA_DEBUG"]=true
Pkg.build("IJulia")

Restart the notebook and look for the error message when IJulia dies. (This changes IJulia to default to verbose = true mode, and sets capture_stderr = false, hopefully sending a bunch of debugging to the terminal where you launched jupyter).

When you are done, set ENV["IJULIA_DEBUG"]=false and re-run Pkg.build("IJulia") to turn off the debugging output.