• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1.. highlight:: none
2
3.. _installing-index:
4
5*************************
6Installing Python Modules
7*************************
8
9:Email: distutils-sig@python.org
10
11As a popular open source development project, Python has an active
12supporting community of contributors and users that also make their software
13available for other Python developers to use under open source license terms.
14
15This allows Python users to share and collaborate effectively, benefiting
16from the solutions others have already created to common (and sometimes
17even rare!) problems, as well as potentially contributing their own
18solutions to the common pool.
19
20This guide covers the installation part of the process. For a guide to
21creating and sharing your own Python projects, refer to the
22`Python packaging user guide`_.
23
24.. _Python Packaging User Guide: https://packaging.python.org/en/latest/tutorials/packaging-projects/
25
26.. note::
27
28   For corporate and other institutional users, be aware that many
29   organisations have their own policies around using and contributing to
30   open source software. Please take such policies into account when making
31   use of the distribution and installation tools provided with Python.
32
33
34Key terms
35=========
36
37* ``pip`` is the preferred installer program. Starting with Python 3.4, it
38  is included by default with the Python binary installers.
39* A *virtual environment* is a semi-isolated Python environment that allows
40  packages to be installed for use by a particular application, rather than
41  being installed system wide.
42* ``venv`` is the standard tool for creating virtual environments, and has
43  been part of Python since Python 3.3. Starting with Python 3.4, it
44  defaults to installing ``pip`` into all created virtual environments.
45* ``virtualenv`` is a third party alternative (and predecessor) to
46  ``venv``. It allows virtual environments to be used on versions of
47  Python prior to 3.4, which either don't provide ``venv`` at all, or
48  aren't able to automatically install ``pip`` into created environments.
49* The `Python Package Index <https://pypi.org>`__ is a public
50  repository of open source licensed packages made available for use by
51  other Python users.
52* the `Python Packaging Authority
53  <https://www.pypa.io/>`__ is the group of
54  developers and documentation authors responsible for the maintenance and
55  evolution of the standard packaging tools and the associated metadata and
56  file format standards. They maintain a variety of tools, documentation,
57  and issue trackers on `GitHub <https://github.com/pypa>`__.
58* ``distutils`` is the original build and distribution system first added to
59  the Python standard library in 1998. While direct use of ``distutils`` is
60  being phased out, it still laid the foundation for the current packaging
61  and distribution infrastructure, and it not only remains part of the
62  standard library, but its name lives on in other ways (such as the name
63  of the mailing list used to coordinate Python packaging standards
64  development).
65
66.. versionchanged:: 3.5
67   The use of ``venv`` is now recommended for creating virtual environments.
68
69.. seealso::
70
71   `Python Packaging User Guide: Creating and using virtual environments
72   <https://packaging.python.org/installing/#creating-virtual-environments>`__
73
74
75Basic usage
76===========
77
78The standard packaging tools are all designed to be used from the command
79line.
80
81The following command will install the latest version of a module and its
82dependencies from the Python Package Index::
83
84    python -m pip install SomePackage
85
86.. note::
87
88   For POSIX users (including macOS and Linux users), the examples in
89   this guide assume the use of a :term:`virtual environment`.
90
91   For Windows users, the examples in this guide assume that the option to
92   adjust the system PATH environment variable was selected when installing
93   Python.
94
95It's also possible to specify an exact or minimum version directly on the
96command line. When using comparator operators such as ``>``, ``<`` or some other
97special character which get interpreted by shell, the package name and the
98version should be enclosed within double quotes::
99
100    python -m pip install SomePackage==1.0.4    # specific version
101    python -m pip install "SomePackage>=1.0.4"  # minimum version
102
103Normally, if a suitable module is already installed, attempting to install
104it again will have no effect. Upgrading existing modules must be requested
105explicitly::
106
107    python -m pip install --upgrade SomePackage
108
109More information and resources regarding ``pip`` and its capabilities can be
110found in the `Python Packaging User Guide <https://packaging.python.org>`__.
111
112Creation of virtual environments is done through the :mod:`venv` module.
113Installing packages into an active virtual environment uses the commands shown
114above.
115
116.. seealso::
117
118    `Python Packaging User Guide: Installing Python Distribution Packages
119    <https://packaging.python.org/installing/>`__
120
121
122How do I ...?
123=============
124
125These are quick answers or links for some common tasks.
126
127... install ``pip`` in versions of Python prior to Python 3.4?
128--------------------------------------------------------------
129
130Python only started bundling ``pip`` with Python 3.4. For earlier versions,
131``pip`` needs to be "bootstrapped" as described in the Python Packaging
132User Guide.
133
134.. seealso::
135
136   `Python Packaging User Guide: Requirements for Installing Packages
137   <https://packaging.python.org/installing/#requirements-for-installing-packages>`__
138
139
140.. installing-per-user-installation:
141
142... install packages just for the current user?
143-----------------------------------------------
144
145Passing the ``--user`` option to ``python -m pip install`` will install a
146package just for the current user, rather than for all users of the system.
147
148
149... install scientific Python packages?
150---------------------------------------
151
152A number of scientific Python packages have complex binary dependencies, and
153aren't currently easy to install using ``pip`` directly. At this point in
154time, it will often be easier for users to install these packages by
155`other means <https://packaging.python.org/science/>`__
156rather than attempting to install them with ``pip``.
157
158.. seealso::
159
160   `Python Packaging User Guide: Installing Scientific Packages
161   <https://packaging.python.org/science/>`__
162
163
164... work with multiple versions of Python installed in parallel?
165----------------------------------------------------------------
166
167On Linux, macOS, and other POSIX systems, use the versioned Python commands
168in combination with the ``-m`` switch to run the appropriate copy of
169``pip``::
170
171   python2   -m pip install SomePackage  # default Python 2
172   python2.7 -m pip install SomePackage  # specifically Python 2.7
173   python3   -m pip install SomePackage  # default Python 3
174   python3.4 -m pip install SomePackage  # specifically Python 3.4
175
176Appropriately versioned ``pip`` commands may also be available.
177
178On Windows, use the ``py`` Python launcher in combination with the ``-m``
179switch::
180
181   py -2   -m pip install SomePackage  # default Python 2
182   py -2.7 -m pip install SomePackage  # specifically Python 2.7
183   py -3   -m pip install SomePackage  # default Python 3
184   py -3.4 -m pip install SomePackage  # specifically Python 3.4
185
186.. other questions:
187
188   Once the Development & Deployment part of PPUG is fleshed out, some of
189   those sections should be linked from new questions here (most notably,
190   we should have a question about avoiding depending on PyPI that links to
191   https://packaging.python.org/en/latest/mirrors/)
192
193
194Common installation issues
195==========================
196
197Installing into the system Python on Linux
198------------------------------------------
199
200On Linux systems, a Python installation will typically be included as part
201of the distribution. Installing into this Python installation requires
202root access to the system, and may interfere with the operation of the
203system package manager and other components of the system if a component
204is unexpectedly upgraded using ``pip``.
205
206On such systems, it is often better to use a virtual environment or a
207per-user installation when installing packages with ``pip``.
208
209
210Pip not installed
211-----------------
212
213It is possible that ``pip`` does not get installed by default. One potential fix is::
214
215    python -m ensurepip --default-pip
216
217There are also additional resources for `installing pip.
218<https://packaging.python.org/en/latest/tutorials/installing-packages/#ensure-pip-setuptools-and-wheel-are-up-to-date>`__
219
220
221Installing binary extensions
222----------------------------
223
224Python has typically relied heavily on source based distribution, with end
225users being expected to compile extension modules from source as part of
226the installation process.
227
228With the introduction of support for the binary ``wheel`` format, and the
229ability to publish wheels for at least Windows and macOS through the
230Python Package Index, this problem is expected to diminish over time,
231as users are more regularly able to install pre-built extensions rather
232than needing to build them themselves.
233
234Some of the solutions for installing `scientific software
235<https://packaging.python.org/science/>`__
236that are not yet available as pre-built ``wheel`` files may also help with
237obtaining other binary extensions without needing to build them locally.
238
239.. seealso::
240
241   `Python Packaging User Guide: Binary Extensions
242   <https://packaging.python.org/extensions/>`__
243