Build wheels with numpy 1.25+ (#4198)

* Build wheels with numpy 1.25+
* Update CI to:
  * Use pip install -e
  * Optionally use build isolation

.github/actions/build-src/action.yaml

@@ -9,6 +9,40 @@ inputs:
     description: 'build MDA docs'
     required: true
     default: false
+  isolation:
+    # Details on build isolation can be found here:
+    # https://pip.pypa.io/en/stable/reference/build-system/pyproject-toml/#build-isolation
+    #
+    # A short description of how this works in pratice is provided here:
+    # * If build isolation is on, pip will create wheels for MDAnalysis in a
+    #   fully isolated temporary environment hence making sure that build
+    #   requirements are handled independently of the runtime environment
+    #   (whatever is already in your Python PATH at the time of install).
+    #   For example, with build isolation on, you could have an old version
+    #   of NumPy available in your current Python environment, but build with
+    #   a much newer one.
+    # * If build isolation is off (through the --no-build-isolation flag), the
+    #   pip build system depends on what is currently avaiable in your Python
+    #   environment. So if NumPy is necessary and you have a version already
+    #   installed, it will just use that directly. There is a massive potential
+    #   for breaking an environment if misused, so this is very much advised
+    #   only in cases where you know exactly what your enviroment does / has
+    #   (such as the tests we run here).
+    #
+    # In practice we use build isolation in MDAnalysis CI when:
+    # * We are trying to test MDAnalysis with a different / older runtime
+    #   version of a key build dependency (e.g. NumPy).
+    # We don't use build isolation when:
+    # * We know we have already installed the required dependencies earlier
+    #   and we want to make sure that we use those dependnecies for the build
+    #   (e.g. we want to check that the packages we pulled for conda-forge
+    #    work for the build, not the ones from PyPi).
+    # * We want to test the build & runtime with a sepecial version of a
+    #   build dependency that won't get picked up by the pip install. For
+    #   example, when we use nightly wheels of NumPy.
+    descriptions: 'Use build isolation for pip installs, if false `--no-build-isolation` is used. '
+    required: true
+    default: false
 
 
 runs:
@@ -19,6 +53,7 @@ runs:
       run: |
         echo ${{ inputs.build-tests }}
         echo ${{ inputs.build-docs }}
+        echo ${{ inputs.isolation }}
 
     - name: check_setup
       shell: bash -l {0}
@@ -34,14 +69,20 @@ runs:
     - name: build_mda_main
       shell: bash -l {0}
       run: |
+        if [ "${{ inputs.isolation }}" == "false" ]; then
+          BUILD_FLAGS="--no-build-isolation"
+        fi
         # install instead of develop would reduce coverage (for .pyx files)
-        cd package/ && python setup.py develop
+        python -m pip install ${BUILD_FLAGS} -v -e ./package
 
     - name: build_mda_tests
       if: ${{ inputs.build-tests == 'true' }}
       shell: bash -l {0}
       run: |
-        cd testsuite/ && python setup.py install
+        if [ "${{ inputs.isolation }}" == "false" ]; then
+          BUILD_FLAGS="--no-build-isolation"
+        fi
+        python -m pip install ${BUILD_FLAGS} -v -e ./testsuite
 
     - name: build_docs
       if: ${{ inputs.build-docs == 'true' }}

.github/workflows/deploy.yaml

@@ -241,18 +241,18 @@ jobs:
           micromamba: true
           full-deps: true
 
-      - name: install_min_deps
-        if: "matrix.type == 'MIN'"
-        run: |
-          pip install pytest pytest-xdist pytest-timeout
-
       - name: pip_install_mda
         run: |
           awk '/__version__ =/ {print $3; exit}' package/MDAnalysis/version.py | tr -d \" > version.dat
           ver=$(python maintainer/norm_version.py --file version.dat)
           pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple MDAnalysis==$ver
           pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple MDAnalysisTests==$ver
 
+      - name: install_min_deps
+        if: "matrix.type == 'MIN'"
+        run: |
+          pip install pytest-xdist pytest-timeout
+
       - name: run_tests
         run: |
           pytest --timeout=200 -n auto --pyargs MDAnalysisTests

.github/workflows/gh-ci-cron.yaml

@@ -69,6 +69,7 @@ jobs:
       with:
         build-tests: true
         build-docs: false
+        isolation: false
 
     - name: run_tests
       run: |
@@ -112,11 +113,13 @@ jobs:
       run: |
         sed -i "s/#extra_cflags =/extra_cflags = -march=native -mtune=native/g" package/setup.cfg
         cat package/setup.cfg
+
     - name: build_srcs
       uses: ./.github/actions/build-src
       with:
         build-tests: true
         build-docs: false
+        isolation: false
 
     - name: run_tests
       run: |
@@ -164,6 +167,7 @@ jobs:
       with:
         build-tests: true
         build-docs: false
+        isolation: false
 
     - name: run_tests
       run: |

.github/workflows/gh-ci.yaml

@@ -88,6 +88,15 @@ jobs:
       with:
         build-tests: true
         build-docs: false
+        # If we specifically define an old NumPy version then we isolate the
+        # the build, i.e. make sure that we build with NumPy 1.25+ but we keep
+        # the old version of NumPy as what we use for runtime testing
+        isolation: ${{ matrix.numpy != '' }}
+
+    - name: check_deps
+      run: |
+        micromamba list
+        pip list
 
     - name: run_tests
       if: contains(matrix.name, 'asv_check') != true
@@ -152,6 +161,7 @@ jobs:
       with:
         build-tests: true
         build-docs: true
+        isolation: false
 
     - name: doctests
       if: github.event_name == 'pull_request'

azure-pipelines.yml

@@ -120,14 +120,12 @@ jobs:
       pip list
     displayName: 'List of installed dependencies'
   - powershell: |
-      cd package
-      python setup.py install
-      cd ..
-      cd testsuite
-      python setup.py install
-      cd ..
+      python -m pip install ./package
+      python -m pip install ./testsuite
     displayName: 'Build MDAnalysis'
     condition: and(succeeded(), eq(variables['BUILD_TYPE'], 'normal'))
+  - powershell: pip list
+    displayName: 'Check installed packages'
   - powershell: |
       cd testsuite
       pytest MDAnalysisTests --disable-pytest-warnings -n auto --timeout=200 -rsx --cov=MDAnalysis

package/CHANGELOG

@@ -35,6 +35,9 @@ Enhancements
     (Issue #3546)
 
 Changes
+  * Package builds are now made using NumPy 1.25 or higher. This provides
+    backwards runtime compatibility with older NumPy builds within the limit
+    of NEP29 (PR #4108).
   * Removed `-ffast-math` compiler flag to avoid potentially incorrect
     floating point results in MDAnalysis *and* other codes when
     MDAnalysis shared libraries are loaded --- see 

package/pyproject.toml

@@ -3,20 +3,11 @@
 requires = [
     "Cython>=0.28",
     "packaging",
-    # lowest NumPy we can use for a given Python,
-    # In part adapted from: https://github.com/scipy/oldest-supported-numpy/blob/main/setup.cfg
-    # except for more exotic platform (Mac Arm flavors)
-    # aarch64, AIX, s390x, and arm64 all support 1.21 so we can safely pin to this
-    # Note: MDA does not build with PyPy so we do not support it in the build system
-    # Scipy: On windows avoid 1.21.6, 1.22.0, and 1.22.1 because they were built on vc142
-    # Let's set everything to this version to make things clean, also avoids other issues
-    # on other archs
-    "numpy==1.22.3; python_version=='3.9' and platform_python_implementation != 'PyPy'",
-    "numpy==1.22.3; python_version=='3.10' and platform_python_implementation != 'PyPy'",
-    "numpy==1.23.2; python_version=='3.11' and platform_python_implementation != 'PyPy'",
-    # For unreleased versions of Python there is currently no known supported
-    # NumPy version. In that case we just let it be a bare NumPy install
-    "numpy<2.0; python_version>='3.12'",
+    # NumPy 1.25+ offers backwards compatibility that tracks with greater
+    # than NEP29 (see: https://numpy.org/doc/stable/dev/depending_on_numpy.html#adding-a-dependency-on-numpy)
+    # We pin our wheel installation to that and a maximum of numpy 2.0 since
+    # this will likely involve several breaking changes
+    "numpy>=1.25,<2.0; python_version>='3.9'",
     "setuptools",
     "wheel",
 ]