##// END OF EJS Templates
Merge pull request #12328 from Carreau/global-config
Merge pull request #12328 from Carreau/global-config

File last commit:

r25743:fe834bf4
r25744:ae5d1094 merge
Show More
intro.rst
232 lines | 7.0 KiB | text/x-rst | RstLexer

Introduction to IPython configuration

Setting configurable options

Many of IPython's classes have configurable attributes (see :doc:`options/index` for the list). These can be configured in several ways.

Python config files

To create the blank config files, run:

ipython profile create [profilename]

If you leave out the profile name, the files will be created for the default profile (see :ref:`profiles`). These will typically be located in :file:`~/.ipython/profile_default/`, and will be named :file:`ipython_config.py`, for hitorical reasons you may also find fiels named with IPytho nprefix instead of Jupyter: :file:`ipython_notebook_config.py`, etc. The settings in :file:`ipython_config.py` apply to all IPython commands.

by default, configuration files are fully featured Python scripts that can execute arbitrary code, the main usage is to set value on the config object c which exist in your configuration file.

You can then configure class attributes like this:

c.InteractiveShell.automagic = False

Be careful with spelling--incorrect names will simply be ignored, with no error.

To add to a collection which may have already been defined elsewhere or have default values, you can use methods like those found on lists, dicts and sets: append, extend, :meth:`~traitlets.config.LazyConfigValue.prepend` (like extend, but at the front), add and update (which works both for dicts and sets):

c.InteractiveShellApp.extensions.append('Cython')

Example config file

# sample ipython_config.py

c.TerminalIPythonApp.display_banner = True
c.InteractiveShellApp.log_level = 20
c.InteractiveShellApp.extensions = [
    'myextension'
]
c.InteractiveShellApp.exec_lines = [
    'import numpy',
    'import scipy'
]
c.InteractiveShellApp.exec_files = [
    'mycode.py',
    'fancy.ipy'
]
c.InteractiveShell.colors = 'LightBG'
c.InteractiveShell.confirm_exit = False
c.InteractiveShell.editor = 'nano'
c.InteractiveShell.xmode = 'Context'

c.PrefilterManager.multi_line_specials = True

c.AliasManager.user_aliases = [
 ('la', 'ls -al')
]

Json Configuration files

in case where executability of configuration can be problematic, or configurations need to be modified programatically, IPython also support a limited set of functionalities via .json config files.

You can defined most of the configuration options via a json object which hierarchy represent the value you woudl normally set on the c object of .py configuration files. The following ipython_config.json file:

{
    "InteractiveShell": {
        "colors": "LightBG",
        "editor": "nano"
    },
    "InteractiveShellApp": {
        "extensions": [
            "myextension"
        ]
    }
}

Is equivalent to the following ipython_config.py:

c.InteractiveShellApp.extensions = [
    'myextension'
]

c.InteractiveShell.colors = 'LightBG'
c.InteractiveShell.editor = 'nano'

Command line arguments

Every configurable value can be set from the command line, using this syntax:

ipython --ClassName.attribute=value

Many frequently used options have short aliases and flags, such as --matplotlib (to integrate with a matplotlib GUI event loop) or --pdb (automatic post-mortem debugging of exceptions).

To see all of these abbreviated options, run:

ipython --help
jupyter notebook --help
# etc.

Options specified at the command line, in either format, override options set in a configuration file.

The config magic

You can also modify config from inside IPython, using a magic command:

%config IPCompleter.greedy = True

At present, this only affects the current session - changes you make to config are not saved anywhere. Also, some options are only read when IPython starts, so they can't be changed like this.

Running IPython from Python

If you are using :ref:`embedding` to start IPython from a normal python file, you can set configuration options the same way as in a config file by creating a traitlets config object and passing it to start_ipython like in the example below.

Profiles

IPython can use multiple profiles, with separate configuration and history. By default, if you don't specify a profile, IPython always runs in the default profile. To use a new profile:

ipython profile create foo   # create the profile foo
ipython --profile=foo        # start IPython using the new profile

Profiles are typically stored in :ref:`ipythondir`, but you can also keep a profile in the current working directory, for example to distribute it with a project. To find a profile directory on the filesystem:

ipython locate profile foo

The IPython directory

IPython stores its files---config, command history and extensions---in the directory :file:`~/.ipython/` by default.

To see where IPython is looking for the IPython directory, use the command ipython locate, or the Python function :func:`IPython.paths.get_ipython_dir`.

Systemwide configuration

It can be usefull to deploy systemwide ipython or ipykernel configuration when managing environment for many users. At startup time IPython and IPykernel will search for configuration file in multiple systemwide locations, mainly:

  • /etc/ipython/
  • /usr/local/etc/ipython/

When the global install is a standalone python distribution it may also search in distribution specific location, for example:

  • $ANACONDA_LOCATION/etc/ipython/

In those locations, Terminal IPython will look for a file called ipython_config.py and ipython_config.json, ipykernel will look for ipython_kernel_config.py and ipython_kernel.json.

Configuration files are loaded in order and merged with configuration on later locatoin taking precedence on earlier locations (that is to say a user can overwrite a systemwide configuration option).

You can see all locations in which IPython is looking for configuration files by starting ipython in debug mode:

$ ipython --debug -c 'exit()'

Identically with ipykernel though the command is currently blocking until this process is killed with Ctrl-\:

$ python -m ipykernel --debug