drop support for python 3.6 (#4720)

* stop to testing on python 3.6

* update the trove classifiers

* update the required minimum python version

* update the minimum python version in the install docs

* update the pipelines ci matrix

* update the minimum versions check ci

* update whats-new.rst

* update the PR number

* upgrade setuptools to 40.4

* update the required setuptools version in the install instructions

* update the policy override for setuptools

* update whats-new.rst

* update the minimum required version of setuptools in setup.cfg

* update the remaining mentions of the setuptools version

azure-pipelines.yml

@@ -8,14 +8,12 @@ jobs:
 - job: Linux
   strategy:
     matrix:
-      py36-bare-minimum:
-        conda_env: py36-bare-minimum
-      py36-min-all-deps:
-        conda_env: py36-min-all-deps
-      py36-min-nep18:
-        conda_env: py36-min-nep18
-      py36:
-        conda_env: py36
+      py37-bare-minimum:
+        conda_env: py37-bare-minimum
+      py37-min-all-deps:
+        conda_env: py37-min-all-deps
+      py37-min-nep18:
+        conda_env: py37-min-nep18
       py37:
         conda_env: py37
       py38:
@@ -121,6 +119,6 @@ jobs:
   - template: ci/azure/add-conda-to-path.yml
   - bash: |
       conda install -y pyyaml
-      python ci/min_deps_check.py ci/requirements/py36-bare-minimum.yml
-      python ci/min_deps_check.py ci/requirements/py36-min-all-deps.yml
+      python ci/min_deps_check.py ci/requirements/py37-bare-minimum.yml
+      python ci/min_deps_check.py ci/requirements/py37-min-all-deps.yml
     displayName: minimum versions policy

ci/min_deps_check.py

@@ -35,9 +35,11 @@
     # setuptools-scm doesn't work with setuptools < 36.7 (Nov 2017).
     # The conda metadata is malformed for setuptools < 38.4 (Jan 2018)
     # (it's missing a timestamp which prevents this tool from working).
+    # setuptools < 40.4 (Sep 2018) from conda-forge cannot be installed into a py37
+    # environment
     # TODO remove this special case and the matching note in installing.rst
-    #      after July 2021.
-    "setuptools": (38, 4),
+    #      after March 2022.
+    "setuptools": (40, 4),
 }
 has_errors = False
 

ci/requirements/py36-bare-minimum.yml → ci/requirements/py37-bare-minimum.yml

@@ -3,7 +3,7 @@ channels:
   - conda-forge
   - nodefaults
 dependencies:
-  - python=3.6
+  - python=3.7
   - coveralls
   - pip
   - pytest
@@ -12,4 +12,4 @@ dependencies:
   - pytest-xdist
   - numpy=1.15
   - pandas=0.25
-  - setuptools=38.4
+  - setuptools=40.4

ci/requirements/py36-min-all-deps.yml → ci/requirements/py37-min-all-deps.yml

@@ -7,7 +7,7 @@ dependencies:
   # Run ci/min_deps_check.py to verify that this file respects the policy.
   # When upgrading python, numpy, or pandas, must also change
   # doc/installing.rst and setup.py.
-  - python=3.6
+  - python=3.7
   - black
   - boto3=1.9
   - bottleneck=1.2
@@ -33,7 +33,7 @@ dependencies:
   - numba=0.46
   - numpy=1.15
   - pandas=0.25
-  # - pint  # See py36-min-nep18.yml
+  # - pint  # See py37-min-nep18.yml
   - pip
   - pseudonetcdf=3.0
   - pydap=3.2
@@ -45,8 +45,8 @@ dependencies:
   - rasterio=1.0
   - scipy=1.3
   - seaborn=0.9
-  - setuptools=38.4
-  # - sparse  # See py36-min-nep18.yml
+  - setuptools=40.4
+  # - sparse  # See py37-min-nep18.yml
   - toolz=0.10
   - zarr=2.3
   - pip:

ci/requirements/py36-min-nep18.yml → ci/requirements/py37-min-nep18.yml

@@ -5,7 +5,7 @@ channels:
 dependencies:
   # Optional dependencies that require NEP18, such as sparse and pint,
   # require drastically newer packages than everything else
-  - python=3.6
+  - python=3.7
   - coveralls
   - dask=2.9
   - distributed=2.9
@@ -18,5 +18,5 @@ dependencies:
   - pytest-env
   - pytest-xdist
   - scipy=1.3
-  - setuptools=38.4
+  - setuptools=40.4
   - sparse=0.8

doc/installing.rst

@@ -6,8 +6,8 @@ Installation
 Required dependencies
 ---------------------
 
-- Python (3.6 or later)
-- setuptools (38.4 or later)
+- Python (3.7 or later)
+- setuptools (40.4 or later)
 - `numpy <http://www.numpy.org/>`__ (1.15 or later)
 - `pandas <http://pandas.pydata.org/>`__ (0.25 or later)
 
@@ -100,7 +100,7 @@ dependencies:
 
 - **Python:** 42 months
   (`NEP-29 <https://numpy.org/neps/nep-0029-deprecation_policy.html>`_)
-- **setuptools:** 42 months (but no older than 38.4)
+- **setuptools:** 42 months (but no older than 40.4)
 - **numpy:** 24 months
   (`NEP-29 <https://numpy.org/neps/nep-0029-deprecation_policy.html>`_)
 - **dask and dask.distributed:** 12 months (but no older than 2.9)

doc/whats-new.rst

@@ -22,6 +22,18 @@ v0.16.3 (unreleased)
 
 Breaking changes
 ~~~~~~~~~~~~~~~~
+- xarray no longer supports python 3.6
+
+  The minimum versions of some other dependencies were changed:
+  ============ ====== ====
+  Package      Old    New
+  ============ ====== ====
+  Python       3.6    3.7
+  setuptools   38.4   40.4
+  ============ ====== ====
+
+  (:issue:`4688`, :pull:`4720`)
+  By `Justus Magin <https://github.com/keewis>`_.
 - As a result of :pull:`4684` the default units encoding for
   datetime-like values (``np.datetime64[ns]`` or ``cftime.datetime``) will now
   always be set such that ``int64`` values can be used.  In the past, no units

requirements.txt

@@ -4,4 +4,4 @@
 
 numpy >= 1.15
 pandas >= 0.25
-setuptools >= 38.4
+setuptools >= 40.4

setup.cfg

@@ -64,21 +64,21 @@ classifiers =
     Intended Audience :: Science/Research
     Programming Language :: Python
     Programming Language :: Python :: 3
-    Programming Language :: Python :: 3.6
     Programming Language :: Python :: 3.7
+    Programming Language :: Python :: 3.8
     Topic :: Scientific/Engineering
 
 [options]
 packages = find:
 zip_safe = False  # https://mypy.readthedocs.io/en/latest/installed_packages.html
 include_package_data = True
-python_requires = >=3.6
+python_requires = >=3.7
 install_requires =
     numpy >= 1.15
     pandas >= 0.25
-    setuptools >= 38.4  # For pkg_resources
+    setuptools >= 40.4  # For pkg_resources
 setup_requires =
-    setuptools >= 38.4
+    setuptools >= 40.4
     setuptools_scm