API: only support python 3.5+

Matplotlib 3.0 will not support python2.7

.appveyor.yml

@@ -19,9 +19,6 @@ environment:
  # theoretically the CONDA_INSTALL_LOCN could be only two: one for 32bit,
  # one for 64bit because we construct envs anyway. But using one for the
  # right python version is hopefully making it fast due to package caching.
- - PYTHON_VERSION: "2.7"
- CONDA_INSTALL_LOCN: "C:\\Miniconda-x64"
- TEST_ALL: "no"
  - PYTHON_VERSION: "3.5"
  CONDA_INSTALL_LOCN: "C:\\Miniconda35-x64"
  TEST_ALL: "no"

.circleci/config.yml

@@ -82,9 +82,9 @@ doc-bundle-run: &doc-bundle
 #
 
 jobs:
- docs-python35:
+ docs-python36:
  docker:
- - image: circleci/python:3.5
+ - image: circleci/python:3.6
  steps:
  - checkout
 
@@ -115,9 +115,9 @@ jobs:
  name: "Deploy new docs"
  command: ./.circleci/deploy-docs.sh
 
- docs-python27:
+ docs-python35:
  docker:
- - image: circleci/python:2.7
+ - image: circleci/python:3.5
  steps:
  - checkout
 
@@ -129,22 +129,13 @@ jobs:
  <<: *deps-install
  environment:
  NUMPY_VERSION: "==1.7.1"
- # Linkchecker only works with python 2.7 for the time being.
- # Linkchecker is currently broken with requests 2.10.0 so force an earlier version.
- - run: pip install --user $PRE requests==2.9.2 linkchecker
  - run: *mpl-install
 
  - run: *doc-build
 
  # We don't build the LaTeX docs here, so linkchecker will complain
  - run: touch doc/build/html/Matplotlib.pdf
 
- # Linkchecker only works with python 2.7 for the time being
- - run:
- name: linkchecker
- command: ~/.local/bin/linkchecker build/html/index.html
- working_directory: doc
-
  - run: *doc-bundle
  - store_artifacts:
  path: doc/build/sphinx-gallery-files.tar.gz
@@ -166,4 +157,4 @@ workflows:
  build:
  jobs:
  - docs-python35
- - docs-python27
+ - docs-python36

.travis.yml

@@ -65,7 +65,7 @@ env:
 
 matrix:
  include:
- - python: 2.7
+ - python: 3.5
  # pytest-cov>=2.3.1 due to https://github.com/pytest-dev/pytest-cov/issues/124.
  env:
  - CYCLER=cycler==0.10
@@ -78,7 +78,7 @@ matrix:
  - PYTEST=pytest==3.1.0
  - PYTEST_COV=pytest-cov==2.3.1
  - SPHINX=sphinx==1.3
- - python: 3.4
+ - python: 3.5
  env: PYTHON_ARGS=-OO
  - python: 3.6
  env: DELETE_FONT_CACHE=1 PANDAS='pandas<0.21.0' PYTEST_PEP8=pytest-pep8 RUN_PEP8=--pep8
@@ -111,7 +111,9 @@ before_install:
  else
  brew update
  brew tap homebrew/gui
- brew install python libpng ffmpeg imagemagick mplayer ccache
+ brew install python3 libpng ffmpeg imagemagick mplayer ccache
+ # make 'python' mean 'python3'
+ ln -s /usr/local/bin/python3 /usr/local/bin/python
  # We could install ghostscript and inkscape here to test svg and pdf
  # but this makes the test time really long.
  # brew install ghostscript inkscape

INSTALL.rst

@@ -32,43 +32,14 @@ Although not required, we suggest also installing ``IPython`` for
 interactive use. To easily install a complete Scientific Python
 stack, see :ref:`install_scipy_dists` below.
 
-.. _installing_windows:
-
-Windows
--------
-
-In case Python 2.7 or 3.4 are not installed for all users,
-the Microsoft Visual C++ 2008
-(`64 bit <https://www.microsoft.com/en-us/download/details.aspx?id=15336>`__
-or
-`32 bit <https://www.microsoft.com/en-us/download/details.aspx?id=29>`__
-for Python 2.7) or Microsoft Visual C++ 2010
-(`64 bit <https://www.microsoft.com/en-us/download/details.aspx?id=14632>`__
-or
-`32 bit <https://www.microsoft.com/en-us/download/details.aspx?id=5555>`__
-for Python 3.4) redistributable packages need to be installed.
 
 macOS
 -----
 
-If you are using Python 2.7 on a Mac you may need to do::
-
- xcode-select --install
-
-so that *subprocess32*, a dependency, may be compiled.
-
 To use the native OSX backend you will need :ref:`a framework build
 <osxframework-faq>` build of Python.
 
 
-Linux
------
-
-On extremely old versions of Linux and Python 2.7 you may need to
-install the master version of *subprocess32* (`see comments
-<https://github.com/google/python-subprocess32/issues/12#issuecomment-304724113>`__).
-
-
 Test Data
 ---------
 
@@ -167,7 +138,7 @@ Dependencies
 
 Matplotlib requires a large number of dependencies:
 
- * `Python <https://www.python.org/downloads/>`_ (>= 2.7 or >= 3.4)
+ * `Python <https://www.python.org/downloads/>`_ (>= 3.5)
  * `NumPy <http://www.numpy.org>`_ (>= |minimum_numpy_version|)
  * `setuptools <https://setuptools.readthedocs.io/en/latest/>`__
  * `dateutil <https://pypi.python.org/pypi/python-dateutil>`_ (>= 2.1)
@@ -177,10 +148,6 @@ Matplotlib requires a large number of dependencies:
  * FreeType (>= 2.3)
  * `cycler <http://matplotlib.org/cycler/>`__ (>= 0.10.0)
  * `six <https://pypi.python.org/pypi/six>`_
- * `backports.functools_lru_cache <https://pypi.python.org/pypi/backports.functools_lru_cache>`_
- (for Python 2.7 only)
- * `subprocess32 <https://pypi.python.org/pypi/subprocess32/>`_ (for Python
- 2.7 only, on Linux and macOS only)
  * `kiwisolver <https://github.com/nucleic/kiwi>`__ (>= 1.0.0)
 
 Optionally, you can also install a number of packages to enable better user
@@ -325,8 +292,6 @@ without fiddling with environment variables::
  conda install pyqt
  # this package is only available in the conda-forge channel
  conda install -c conda-forge msinttypes
- # for Python 2.7
- conda install -c conda-forge backports.functools_lru_cache
 
  # copy the libs which have "wrong" names
  set LIBRARY_LIB=%CONDA_DEFAULT_ENV%\Library\lib

doc/faq/installing_faq.rst

@@ -216,11 +216,6 @@ the disk image installer only works for Python.org Python, and will not get
 picked up by other Pythons. If all these fail, please :ref:`let us know
 <reporting-problems>`.
 
-Windows Notes
-=============
-
-See :ref:`installing_windows`.
-
 .. _install-from-git:
 
 Install from source