The Qt Console for Jupyter¶

Release: 4.5.1
Date: May 31, 2019

To start the Qt console:

$ jupyter qtconsole

Installation¶

The Qt console requires Qt, such as PyQt5, PyQt4, or PySide.

Although pip and conda may be used to install the Qt console, conda is simpler to use since it automatically installs PyQt. Alternatively, qtconsole installation with pip needs additional steps since pip cannot install the Qt requirement.

Install using conda¶

To install:

conda install qtconsole

Note

If the Qt console is installed using conda, it will automatically install the Qt requirement as well.

Install using pip¶

To install:

pip install qtconsole

Important

Make sure that Qt is installed. Unfortunately, Qt cannot be installed using pip. The next section gives instructions on installing Qt.

Installing Qt (if needed)¶

We recommend installing PyQt with conda:

conda install pyqt

or with a system package manager. For Windows, PyQt binary packages may be used.

For example with Linux Debian’s system package manager, use:

sudo apt-get install python3-pyqt5 # PyQt5 on Python 3
sudo apt-get install python3-pyqt4 # PyQt4 on Python 3
sudo apt-get install python-qt4    # PyQt4 on Python 2

Configuration options¶

These options can be set in the configuration file, ~/.jupyter/jupyter_qtconsole_config.py, or at the command line when you start Qt console.

You may enter jupyter qtconsole --help-all to get information about all available configuration options.

Options¶

ConnectionFileMixin.connection_fileUnicode

Default: ''

JSON file in which to store connection info [default: kernel-<pid>.json]

This file will contain the IP, ports, and authentication key needed to connect clients to this kernel. By default, this file will be created in the security dir of the current profile, but can be specified by absolute path.

ConnectionFileMixin.control_portInt

Default: 0

set the control (ROUTER) port [default: random]

ConnectionFileMixin.hb_portInt

Default: 0

set the heartbeat port [default: random]

ConnectionFileMixin.iopub_portInt

Default: 0

set the iopub (PUB) port [default: random]

ConnectionFileMixin.ipUnicode

Default: ''

Set the kernel’s IP address [default localhost]. If the IP address is something other than localhost, then Consoles on other machines will be able to connect to the Kernel, so be careful!

ConnectionFileMixin.shell_portInt

Default: 0

set the shell (ROUTER) port [default: random]

ConnectionFileMixin.stdin_portInt

Default: 0

set the stdin (ROUTER) port [default: random]

ConnectionFileMixin.transport‘tcp’|’ipc’

Default: 'tcp'

No description

JupyterConsoleApp.confirm_exitCBool

Default: True

Set to display confirmation dialog on exit. You can always use ‘exit’ or ‘quit’, to force a direct exit without any confirmation.

JupyterConsoleApp.existingCUnicode

Default: ''

Connect to an already running kernel

JupyterConsoleApp.kernel_nameUnicode

Default: 'python'

The name of the default kernel to start.

JupyterConsoleApp.sshkeyUnicode

Default: ''

Path to the ssh key to use for logging in to the ssh server.

JupyterConsoleApp.sshserverUnicode

Default: ''

The SSH server to use to connect to the kernel.

Application.log_datefmtUnicode

Default: '%Y-%m-%d %H:%M:%S'

The date format used by logging formatters for %(asctime)s

Application.log_formatUnicode

Default: '[%(name)s]%(highlevel)s %(message)s'

The Logging format template

Application.log_level0|10|20|30|40|50|’DEBUG’|’INFO’|’WARN’|’ERROR’|’CRITICAL’

Default: 30

Set the log level by value or name.

JupyterApp.answer_yesBool

Default: False

Answer yes to any prompts.

JupyterApp.config_fileUnicode

Default: ''

Full path of a config file.

JupyterApp.config_file_nameUnicode

Default: ''

Specify a config file to load.

JupyterApp.generate_configBool

Default: False

Generate default config file.

JupyterQtConsoleApp.display_bannerCBool

Default: True

Whether to display a banner upon starting the QtConsole.

JupyterQtConsoleApp.hide_menubarCBool

Default: False

Start the console window with the menu bar hidden.

JupyterQtConsoleApp.maximizeCBool

Default: False

Start the console window maximized.

JupyterQtConsoleApp.plainCBool

Default: False

Use a plaintext widget instead of rich text (plain can’t print/save).

JupyterQtConsoleApp.stylesheetUnicode

Default: ''

path to a custom CSS stylesheet

ConsoleWidget.ansi_codesBool

Default: True

Whether to process ANSI escape codes.

ConsoleWidget.buffer_sizeInt

Default: 500

The maximum number of lines of text before truncation. Specifying a non-positive number disables text truncation (not recommended).

ConsoleWidget.console_heightInt

Default: 25

The height of the console at start time in number of characters (will double with vsplit paging)

ConsoleWidget.console_widthInt

Default: 81

The width of the console at start time in number of characters (will double with hsplit paging)

ConsoleWidget.execute_on_complete_inputBool

Default: True

Whether to automatically execute on syntactically complete input.

If False, Shift-Enter is required to submit each execution. Disabling this is mainly useful for non-Python kernels, where the completion check would be wrong.

ConsoleWidget.font_familyUnicode

Default: ''

The font family to use for the console. On OSX this defaults to Monaco, on Windows the default is Consolas with fallback of Courier, and on other platforms the default is Monospace.

ConsoleWidget.font_sizeInt

Default: 0

The font size. If unconfigured, Qt will be entrusted with the size of the font.

ConsoleWidget.gui_completion‘plain’|’droplist’|’ncurses’

Default: 'ncurses'

The type of completer to use. Valid values are:

‘plain’Show the available completion as a text list: Below the editing area.
‘droplist’: Show the completion in a drop down list navigable: by the arrow keys, and from which you can select completion by pressing Return.
‘ncurses’Show the completion as a text list which is navigable by: tab and arrow keys.

ConsoleWidget.include_other_outputBool

Default: False

Whether to include output from clients other than this one sharing the same kernel.

Outputs are not displayed until enter is pressed.

ConsoleWidget.kind‘plain’|’rich’

Default: 'plain'

The type of underlying text widget to use. Valid values are ‘plain’, which specifies a QPlainTextEdit, and ‘rich’, which specifies a QTextEdit.

ConsoleWidget.other_output_prefixUnicode

Default: '[remote] '

Prefix to add to outputs coming from clients other than this one.

Only relevant if include_other_output is True.

ConsoleWidget.paging‘inside’|’hsplit’|’vsplit’|’custom’|’none’

Default: 'inside'

The type of paging to use. Valid values are:

‘inside’: The widget pages like a traditional terminal.
‘hsplit’: When paging is requested, the widget is split horizontally. The top pane contains the console, and the bottom pane contains the paged text.
‘vsplit’: Similar to ‘hsplit’, except that a vertical splitter is used.
‘custom’: No action is taken by the widget beyond emitting a ‘custom_page_requested(str)’ signal.
‘none’: The text is written directly to the console.

HistoryConsoleWidget.history_lockBool

Default: False

No description

FrontendWidget.bannerUnicode

Default: ''

No description

FrontendWidget.clear_on_kernel_restartBool

Default: True

Whether to clear the console when the kernel is restarted

FrontendWidget.confirm_restartBool

Default: True

Whether to ask for user confirmation when restarting kernel

FrontendWidget.enable_calltipsBool

Default: True

Whether to draw information calltips on open-parentheses.

FrontendWidget.lexer_classDottedObjectName

Default: traitlets.Undefined

The pygments lexer class to use.

JupyterWidget.editorUnicode

Default: ''

A command for invoking a GUI text editor. If the string contains a {filename} format specifier, it will be used. Otherwise, the filename will be appended to the end the command. To use a terminal text editor, the command should launch a new terminal, e.g. "gnome-terminal -- vim".

JupyterWidget.editor_lineUnicode

Default: ''

The editor command to use when a specific line number is requested. The string should contain two format specifiers: {line} and {filename}. If this parameter is not specified, the line number option to the %edit magic will be ignored.

JupyterWidget.in_promptUnicode

Default: 'In [<span class="in-prompt-number">%i</span>]: '

No description

JupyterWidget.input_sepUnicode

Default: '\\n'

No description

JupyterWidget.out_promptUnicode

Default: 'Out[<span class="out-prompt-number">%i</span>]: '

No description

JupyterWidget.output_sepUnicode

Default: ''

No description

JupyterWidget.output_sep2Unicode

Default: ''

No description

JupyterWidget.style_sheetUnicode

Default: ''

A CSS stylesheet. The stylesheet can contain classes for:

Qt: QPlainTextEdit, QFrame, QWidget, etc
Pygments: .c, .k, .o, etc. (see PygmentsHighlighter)
QtConsole: .error, .in-prompt, .out-prompt, etc

JupyterWidget.syntax_styleUnicode

Default: ''

If not empty, use this Pygments style for syntax highlighting. Otherwise, the style sheet is queried for Pygments style information.

KernelManager.autorestartBool

Default: True

Should we autorestart the kernel if it dies.

KernelManager.kernel_cmdList

Default: []

DEPRECATED: Use kernel_name instead.

The Popen Command to launch the kernel. Override this if you have a custom kernel. If kernel_cmd is specified in a configuration file, Jupyter does not pass any arguments to the kernel, because it cannot make any assumptions about the arguments that the kernel understands. In particular, this means that the kernel does not receive the option –debug if it given on the Jupyter command line.

KernelManager.shutdown_wait_timeFloat

Default: 5.0

Time to wait for a kernel to terminate before killing it, in seconds.

KernelRestarter.debugBool

Default: False

Whether to include every poll event in debugging output.

Has to be set explicitly, because there will be a lot of output.

KernelRestarter.random_ports_until_aliveBool

Default: True

Whether to choose new random ports when restarting before the kernel is alive.

KernelRestarter.restart_limitInt

Default: 5

The number of consecutive autorestarts before the kernel is presumed dead.

KernelRestarter.time_to_deadFloat

Default: 3.0

Kernel heartbeat interval in seconds.

Session.buffer_thresholdInt

Default: 1024

Threshold (in bytes) beyond which an object’s buffer should be extracted to avoid pickling.

Session.check_pidBool

Default: True

Whether to check PID to protect against calls after fork.

This check can be disabled if fork-safety is handled elsewhere.

Session.copy_thresholdInt

Default: 65536

Threshold (in bytes) beyond which a buffer should be sent without copying.

Session.debugBool

Default: False

Debug output in the Session

Session.digest_history_sizeInt

Default: 65536

The maximum number of digests to remember.

The digest history will be culled when it exceeds this value.

Session.item_thresholdInt

Default: 64

The maximum number of items for a container to be introspected for custom serialization. Containers larger than this are pickled outright.

Session.keyCBytes

Default: b''

execution key, for signing messages.

Session.keyfileUnicode

Default: ''

path to file containing execution key.

Session.metadataDict

Default: {}

Metadata dictionary, which serves as the default top-level metadata dict for each message.

Session.packerDottedObjectName

Default: 'json'

The name of the packer for serializing messages. Should be one of ‘json’, ‘pickle’, or an import name for a custom callable serializer.

Session.sessionCUnicode

Default: ''

The UUID identifying this session.

Session.signature_schemeUnicode

Default: 'hmac-sha256'

The digest scheme used to construct the message signatures. Must have the form ‘hmac-HASH’.

Session.unpackerDottedObjectName

Default: 'json'

The name of the unpacker for unserializing messages. Only used with custom functions for packer.

Session.usernameUnicode

Default: 'username'

Username for the Session. Default is your system username.

Changes in Jupyter Qt console¶

4.5.1 on GitHub

Only use setuptools in setup.py to fix uploading tarballs to PyPI.

4.5 on GitHub

Additions¶

Add Comms to qtconsole.
Add kernel language name as an attribute of JupyterWidget.

Changes¶

Use new traitlets API with decorators.

4.4.4 on GitHub

Prevent cursor from moving to the end of the line while debugging.

4.4.3 on GitHub

Fix complete statements check inside indented block for Python after the IPython 7 release.
Improve auto-scrolling during execution.

4.4.2 on GitHub

Fix incompatibility with PyQt5 5.11.

4.4.1 on GitHub

Fix setting width and height when displaying images with IPython’s Image.
Avoid displaying errors when using Matplotlib to generate pngs from Latex.

4.4 on GitHub

Additions¶

Control-D enters an EOT character if kernel is executing and input is empty.
Implement block indent on multiline selection with Tab.
Change the syntax highlighting style used in the console at any time. It can be done in the menu View > Syntax Style.

Changes¶

Change Control-Shift-A to select cell contents first.
Change default tab width to 4 spaces.
Enhance handling of input from other clients.
Don’t block the console when the kernel is asked for completions.

Fixes¶

Fix bug that make PySide2 a forbidden binding.
Fix IndexError when copying prompts.
Fix behavior of right arrow key.
Fix behavior of Control-Backspace and Control-Del

4.3.1 on GitHub

Make %clear to delete previous output on Windows.
Fix SVG rendering.

4.3¶

4.3 on GitHub

Rename ConsoleWidget.width/height traits to console_width/console_height to avoid a name clash with the QWidget properties. Note: the name change could be, in rare cases if a name collision exists, a code-breaking change.

Additions¶

Add Shift-Tab shortcut to unindent text
Add Control-R shortcut to rename the current tab
Add Alt-R shortcut to set the main window title
Add Command-Alt-Left and Command-Alt-Right shortcut to switch tabs on macOS
Add support for PySide2
Add support for Python 3.5
Add support for 24 bit ANSI color codes
Add option to create new tab connected to the existing kernel

Changes¶

Change Tab key behavior to always indent to the next increment of 4 spaces
Change Home key behavior to alternate cursor between the beginning of text (ignoring leading spaces) and beginning of the line
Improve documentation of various options and clarified the docs in some places
Move documentation to ReadTheDocs

Fixes¶

Fix automatic indentation of new lines that are inserted in the middle of a cell
Fix regression where prompt would never be shown for --existing consoles
Fix python.exe -m qtconsole on Windows
Fix showing error messages when running a script using %run
Fix invalid cursor position error and subsequent freezing of user input
Fix syntax coloring when attaching to non-IPython kernels
Fix printing when using QT5
Fix Control-K shortcut (delete until end of line) on macOS
Fix history browsing (Up/Down keys) when lines are longer than the terminal width
Fix saving HTML with inline PNG for Python 3
Various internal bugfixes

4.2¶

4.2 on GitHub

various latex display fixes
improvements for embedding in Qt applications (use existing Qt API if one is already loaded)

4.1¶

4.1.1¶

4.1.1 on GitHub

Set AppUserModelID for taskbar icon on Windows 7 and later

4.1.0¶

4.1 on GitHub

fix regressions in copy/paste, completion
fix issues with inprocess IPython kernel
fix jupyter qtconsole --generate-config

4.0¶

4.0.1¶

fix installation issues, including setuptools entrypoints for Windows
Qt5 fixes

4.0.0¶

4.0 on GitHub

First release of the Qt console as a standalone package.

Overview¶

The Qt console is a very lightweight application that largely feels like a terminal, but provides a number of enhancements only possible in a GUI, such as inline figures, proper multi-line editing with syntax highlighting, graphical calltips, and much more. The Qt console can use any Jupyter kernel.

The Qt console with IPython, using inline matplotlib plots.¶

The Qt console frontend has hand-coded emacs-style bindings for text navigation. This is not yet configurable.

Tip

Since the Qt console tries hard to behave like a terminal, by default it immediately executes single lines of input that are complete. If you want to force multi-line input, hit Ctrl-Enter at the end of the first line instead of Enter, and it will open a new line for input. At any point in a multi-line block, you can force its execution (without having to go to the bottom) with Shift-Enter.

Inline graphics¶

One of the most exciting features of the Qt Console is embedded figures. You can plot with matplotlib in IPython, or the plotting library of choice in your kernel.

Saving and Printing¶

The Qt Console has the ability to save your current session, as either HTML or XHTML. Your inline figures will be PNG in HTML, or inlined as SVG in XHTML. PNG images have the option to be either in an external folder, as in many browsers’ “Webpage, Complete” option, or inlined as well, for a larger, but more portable file.

Note

Export to SVG+XHTML requires that you are using SVG figures, which is not the default. To switch the inline figure format in IPython to use SVG, do:

In [10]: %config InlineBackend.figure_format = 'svg'

Or, you can add the same line (c.Inline… instead of %config Inline…) to your config files.

This will only affect figures plotted after making this call

The widget also exposes the ability to print directly, via the default print shortcut or context menu.

See these examples of png/html and svg/xhtml output. Note that syntax highlighting does not survive export. This is a known issue, and is being investigated.

Colors and Highlighting¶

Terminal IPython has always had some coloring, but never syntax highlighting. There are a few simple color choices, specified by the colors flag or %colors magic:

LightBG for light backgrounds
Linux for dark backgrounds
NoColor for a simple colorless terminal

The Qt widget, however, has full syntax highlighting as you type, handled by the pygments library. The style argument exposes access to any style by name that can be found by pygments, and there are several already installed.

Screenshot of jupyter qtconsole --style monokai, which uses the ‘monokai’ theme:

Note

Calling jupyter qtconsole -h will show all the style names that pygments can find on your system.

You can also pass the filename of a custom CSS stylesheet, if you want to do your own coloring, via the stylesheet argument. The default LightBG stylesheet:

QPlainTextEdit, QTextEdit { background-color: white;
        color: black ;
        selection-background-color: #ccc}
.error { color: red; }
.in-prompt { color: navy; }
.in-prompt-number { font-weight: bold; }
.out-prompt { color: darkred; }
.out-prompt-number { font-weight: bold; }
/* .inverted is used to highlight selected completion */
.inverted { background-color: black ; color: white; }

Fonts¶

The Qt console is configurable via the ConsoleWidget. To change these, set the font_family or font_size traits of the ConsoleWidget. For instance, to use 9pt Anonymous Pro:

$> jupyter qtconsole --ConsoleWidget.font_family="Anonymous Pro" --ConsoleWidget.font_size=9

Process Management¶

With the two-process ZMQ model, the frontend does not block input during execution. This means that actions can be taken by the frontend while the Kernel is executing, or even after it crashes. The most basic such command is via ‘Ctrl-.’, which restarts the kernel. This can be done in the middle of a blocking execution. The frontend can also know, via a heartbeat mechanism, that the kernel has died. This means that the frontend can safely restart the kernel.

Multiple Consoles¶

Since the Kernel listens on the network, multiple frontends can connect to it. These do not have to all be qt frontends - any Jupyter frontend can connect and run code.

Other frontends can connect to your kernel, and share in the execution. This is great for collaboration. The --existing flag means connect to a kernel that already exists. Starting other consoles with that flag will not try to start their own kernel, but rather connect to yours. kernel-12345.json is a small JSON file with the ip, port, and authentication information necessary to connect to your kernel. By default, this file will be in your Jupyter runtime directory. If it is somewhere else, you will need to use the full path of the connection file, rather than just its filename.

If you need to find the connection info to send, and don’t know where your connection file lives, there are a couple of ways to get it. If you are already running a console connected to an IPython kernel, you can use the %connect_info magic to display the information necessary to connect another frontend to the kernel.

In [2]: %connect_info
{
  "stdin_port":50255,
  "ip":"127.0.0.1",
  "hb_port":50256,
  "key":"70be6f0f-1564-4218-8cda-31be40a4d6aa",
  "shell_port":50253,
  "iopub_port":50254
}

Paste the above JSON into a file, and connect with:
    $> ipython <app> --existing <file>
or, if you are local, you can connect with just:
    $> ipython <app> --existing kernel-12345.json
or even just:
    $> ipython <app> --existing
if this is the most recent kernel you have started.

Otherwise, you can find a connection file by name (and optionally profile) with jupyter_client.find_connection_file():

$> python -c "from jupyter_client import find_connection_file;\
print(find_connection_file('kernel-12345.json'))"
/home/you/Library/Jupyter/runtime/kernel-12345.json

Security¶

Warning

Since the ZMQ code currently has no encryption, listening on an external-facing IP is dangerous. You are giving any computer that can see you on the network the ability to connect to your kernel, and view your traffic. Read the rest of this section before listening on external ports or running a kernel on a shared machine.

By default (for security reasons), the kernel only listens on localhost, so you can only connect multiple frontends to the kernel from your local machine. You can specify to listen on an external interface by specifying the ip argument:

$> jupyter qtconsole --ip=192.168.1.123

If you specify the ip as 0.0.0.0 or ‘*’, that means all interfaces, so any computer that can see yours on the network can connect to the kernel.

Messages are not encrypted, so users with access to the ports your kernel is using will be able to see any output of the kernel. They will NOT be able to issue shell commands as you due to message signatures.

Warning

If you disable message signatures, then any user with access to the ports your kernel is listening on can issue arbitrary code as you. DO NOT disable message signatures unless you have a lot of trust in your environment.

The one security feature Jupyter does provide is protection from unauthorized execution. Jupyter’s messaging system will sign messages with HMAC digests using a shared-key. The key is never sent over the network, it is only used to generate a unique hash for each message, based on its content. When the kernel receives a message, it will check that the digest matches, and discard the message. You can use any file that only you have access to to generate this key, but the default is just to generate a new UUID.

SSH Tunnels¶

Sometimes you want to connect to machines across the internet, or just across a LAN that either doesn’t permit open ports or you don’t trust the other machines on the network. To do this, you can use SSH tunnels. SSH tunnels are a way to securely forward ports on your local machine to ports on another machine, to which you have SSH access.

In simple cases, Jupyter’s tools can forward ports over ssh by simply adding the --ssh=remote argument to the usual --existing... set of flags for connecting to a running kernel, after copying the JSON connection file (or its contents) to the second computer.

Warning

Using SSH tunnels does not increase localhost security. In fact, when tunneling from one machine to another both machines have open ports on localhost available for connections to the kernel.

There are two primary models for using SSH tunnels with Jupyter. The first is to have the Kernel listen only on localhost, and connect to it from another machine on the same LAN.

First, let’s start a kernel on machine worker, listening only on loopback:

user@worker $> ipython kernel
[IPKernelApp] To connect another client to this kernel, use:
[IPKernelApp] --existing kernel-12345.json

In this case, the IP that you would connect to would still be 127.0.0.1, but you want to specify the additional --ssh argument with the hostname of the kernel (in this example, it’s ‘worker’):

user@client $> jupyter qtconsole  --ssh=worker --existing /path/to/kernel-12345.json

Which will write a new connection file with the forwarded ports, so you can reuse them:

[JupyterQtConsoleApp] To connect another client via this tunnel, use:
[JupyterQtConsoleApp] --existing kernel-12345-ssh.json

Note again that this opens ports on the client machine that point to your kernel.

Note

the ssh argument is simply passed to openssh, so it can be fully specified user@host:port but it will also respect your aliases, etc. in .ssh/config if you have any.

The second pattern is for connecting to a machine behind a firewall across the internet (or otherwise wide network). This time, we have a machine login that you have ssh access to, which can see kernel, but client is on another network. The important difference now is that client can see login, but not worker. So we need to forward ports from client to worker via login. This means that the kernel must be started listening on external interfaces, so that its ports are visible to login:

user@worker $> ipython kernel --ip=0.0.0.0
[IPKernelApp] To connect another client to this kernel, use:
[IPKernelApp] --existing kernel-12345.json

Which we can connect to from the client with:

user@client $> jupyter qtconsole --ssh=login --ip=192.168.1.123 --existing /path/to/kernel-12345.json

Note

The IP here is the address of worker as seen from login, and need only be specified if the kernel used the ambiguous 0.0.0.0 (all interfaces) address. If it had used 192.168.1.123 to start with, it would not be needed.

Manual SSH tunnels¶

It’s possible that Jupyter’s ssh helper functions won’t work for you, for various reasons. You can still connect to remote machines, as long as you set up the tunnels yourself. The basic format of forwarding a local port to a remote one is:

[client] $> ssh <server> <localport>:<remoteip>:<remoteport> -f -N

This will forward local connections to localport on client to remoteip:remoteport via server. Note that remoteip is interpreted relative to server, not the client. So if you have direct ssh access to the machine to which you want to forward connections, then the server is the remote machine, and remoteip should be server’s IP as seen from the server itself, i.e. 127.0.0.1. Thus, to forward local port 12345 to remote port 54321 on a machine you can see, do:

[client] $> ssh machine 12345:127.0.0.1:54321 -f -N

But if your target is actually on a LAN at 192.168.1.123, behind another machine called login, then you would do:

[client] $> ssh login 12345:192.168.1.16:54321 -f -N

The -f -N on the end are flags that tell ssh to run in the background, and don’t actually run any commands beyond creating the tunnel.

Stopping Kernels and Consoles¶

Since there can be many consoles per kernel, the shutdown mechanism and dialog are probably more complicated than you are used to. Since you don’t always want to shutdown a kernel when you close a window, you are given the option to just close the console window or also close the Kernel and all other windows. Note that this only refers to all other local windows, as remote Consoles are not allowed to shutdown the kernel, and shutdowns do not close Remote consoles (to allow for saving, etc.).

Rules:

Restarting the kernel automatically clears all local Consoles, and prompts remote Consoles about the reset.

Shutdown closes all local Consoles, and notifies remotes that the Kernel has been shutdown.

Remote Consoles may not restart or shutdown the kernel.

Qt and the REPL¶

Note

This section is relevant regardless of the frontend you use to write Qt Code. This section is mostly there as it is easy to get confused and assume that writing Qt code in the QtConsole should change from usual Qt code. It should not. If you get confused, take a step back, and try writing your code using the pure terminal based jupyter console that does not involve Qt.

An important part of working with the REPL – QtConsole, Jupyter notebook, IPython terminal – when you are writing your own Qt code is to remember that user code (in the kernel) is not in the same process as the frontend. This means that there is not necessarily any Qt code running in the kernel, and under most normal circumstances there isn’t. This is true even if you are running the QtConsole.

Warning

When executing code from the qtconsole prompt, it is not possible to access the QtApplication instance of the QtConsole itself.

A common problem listed in the PyQt4 Gotchas is the fact that Python’s garbage collection will destroy Qt objects (Windows, etc.) once there is no longer a Python reference to them, so you have to hold on to them. For instance, in:

from PyQt4 import QtGui

def make_window():
    win = QtGui.QMainWindow()

def make_and_return_window():
    win = QtGui.QMainWindow()
    return win

make_window() will never draw a window, because garbage collection will destroy it before it is drawn, whereas make_and_return_window() lets the caller decide when the window object should be destroyed. If, as a developer, you know that you always want your objects to last as long as the process, you can attach them to the QApplication instance itself:

from PyQt4 import QtGui, QtCore

# do this just once:
app = QtCore.QCoreApplication.instance()
if not app:
    # we are in the kernel in most of the case there is NO qt code running.
    # we need to create a Gui APP.
    app = QtGui.QApplication([])
app.references = set()
# then when you create Windows, add them to the set
def make_window():
    win = QtGui.QMainWindow()
    app.references.add(win)

Now the QApplication itself holds a reference to win, so it will never be garbage collected until the application itself is destroyed.

Embedding the QtConsole in a Qt application¶

There are a few options to integrate the Jupyter Qt console with your own application:

Use qtconsole.rich_jupyter_widget.RichJupyterWidget in your Qt application. This will embed the console widget in your GUI and start the kernel in a separate process, so code typed into the console cannot access objects in your application. See examples/embed_qtconsole.py for an example.
Start an IPython kernel inside a PyQt application ( ipkernel_qtapp.py in the ipykernel repository shows how to do this). Then launch the Qt console in a separate process to connect to it. This means that the console will be in a separate window from your application’s UI, but the code entered by the user runs in your application.
Start a special IPython kernel, the ipykernel.inprocess.ipkernel.InProcessKernel, which allows a QtConsole in the same process. See examples/inprocess_qtconsole.py for an example. This allows both the kernel and the console interface to be part of your application, but it is not well supported. We encourage you to use one of the above options instead if you can.

Regressions¶

There are some features, where the qt console lags behind the Terminal frontend:

!cmd input: Due to our use of pexpect, we cannot pass input to subprocesses launched using the ‘!’ escape, so you should never call a command that requires interactive input. For such cases, use the terminal IPython. This will not be fixed, as abandoning pexpect would significantly degrade the console experience.