scikit-learn · adrinjalali · Jan 23, 2024 · Dec 29, 2023 · Dec 30, 2023 · Dec 30, 2023
diff --git a/Makefile b/Makefile
@@ -23,6 +23,14 @@ in: inplace # just a shortcut
 inplace:
 	$(PYTHON) setup.py build_ext -i
 
+dev-meson:
+	# Temporary script to try the experimental meson build. Once meson is
+	# accepted as the default build tool, this will go away.
+	python build_tools/build-meson-editable-install.py
+
+clean-meson:
+	pip uninstall -y scikit-learn
+
 test-code: in
 	$(PYTEST) --showlocals -v sklearn --durations=20
 test-sphinxext:

diff --git a/azure-pipelines.yml b/azure-pipelines.yml
@@ -171,6 +171,7 @@ jobs:
         DISTRIB: 'conda'
         LOCK_FILE: './build_tools/azure/pylatest_conda_forge_mkl_linux-64_conda.lock'
         COVERAGE: 'true'
+        BUILD_WITH_MESON: 'true'
         SKLEARN_TESTS_GLOBAL_RANDOM_SEED: '42'  # default global random seed
 
 # Check compilation with Ubuntu 22.04 LTS (Jammy Jellyfish) and scipy from conda-forge

diff --git a/build_tools/azure/install.sh b/build_tools/azure/install.sh
@@ -118,8 +118,10 @@ scikit_learn_install() {
         export LDFLAGS="$LDFLAGS -Wl,--sysroot=/"
     fi
 
+    if [[ "$BUILD_WITH_MESON" == "true" ]]; then
+        make dev-meson
     # TODO use a specific variable for this rather than using a particular build ...
-    if [[ "$DISTRIB" == "conda-pip-latest" ]]; then
+    elif [[ "$DISTRIB" == "conda-pip-latest" ]]; then
         # Check that pip can automatically build scikit-learn with the build
         # dependencies specified in pyproject.toml using an isolated build
         # environment:

diff --git a/build_tools/azure/pylatest_conda_forge_mkl_linux-64_conda.lock b/build_tools/azure/pylatest_conda_forge_mkl_linux-64_conda.lock
@@ -1,6 +1,6 @@
 # Generated by conda-lock.
 # platform: linux-64
-# input_hash: 0e751f4212c4e51710aad471314a8b385a5e12fe3536c2a766f949da61eabb88
+# input_hash: 0cee038efd0cc93a79f66b1cdbbc359ac52521c98df956b3e4042575e89711f5
 @EXPLICIT
 https://conda.anaconda.org/conda-forge/linux-64/_libgcc_mutex-0.1-conda_forge.tar.bz2#d7c89558ba9fa0495403155b64376d81
 https://conda.anaconda.org/conda-forge/linux-64/ca-certificates-2023.11.17-hbcca054_0.conda#01ffc8d36f9eba0ce0b3c1955fa780ee
@@ -20,7 +20,7 @@ https://conda.anaconda.org/conda-forge/linux-64/alsa-lib-1.2.10-hd590300_0.conda
 https://conda.anaconda.org/conda-forge/linux-64/attr-2.5.1-h166bdaf_1.tar.bz2#d9c69a24ad678ffce24c6543a0176b00
 https://conda.anaconda.org/conda-forge/linux-64/aws-c-common-0.9.0-hd590300_0.conda#71b89db63b5b504e7afc8ad901172e1e
 https://conda.anaconda.org/conda-forge/linux-64/bzip2-1.0.8-hd590300_5.conda#69b8b6202a07720f448be700e300ccf4
-https://conda.anaconda.org/conda-forge/linux-64/c-ares-1.24.0-hd590300_0.conda#f5842b88e9cbfa177abfaeacd457a45d
+https://conda.anaconda.org/conda-forge/linux-64/c-ares-1.25.0-hd590300_0.conda#89e40af02dd3a0846c0c1131c5126706
 https://conda.anaconda.org/conda-forge/linux-64/gettext-0.21.1-h27087fc_0.tar.bz2#14947d8770185e5153fdd04d4673ed37
 https://conda.anaconda.org/conda-forge/linux-64/gflags-2.2.2-he1b5a44_1004.tar.bz2#cddaf2c63ea4a5901cf09524c490ecdc
 https://conda.anaconda.org/conda-forge/linux-64/graphite2-1.3.13-h58526e2_1001.tar.bz2#8c54672728e8ec6aa6db90cf2806d220
@@ -50,6 +50,7 @@ https://conda.anaconda.org/conda-forge/linux-64/libzlib-1.2.13-hd590300_5.conda#
 https://conda.anaconda.org/conda-forge/linux-64/lz4-c-1.9.4-hcb278e6_0.conda#318b08df404f9c9be5712aaa5a6f0bb0
 https://conda.anaconda.org/conda-forge/linux-64/mpg123-1.32.3-h59595ed_0.conda#bdadff838d5437aea83607ced8b37f75
 https://conda.anaconda.org/conda-forge/linux-64/ncurses-6.4-h59595ed_2.conda#7dbaa197d7ba6032caf7ae7f32c1efa0
+https://conda.anaconda.org/conda-forge/linux-64/ninja-1.11.1-h924138e_0.conda#73a4953a2d9c115bdc10ff30a52f675f
 https://conda.anaconda.org/conda-forge/linux-64/nspr-4.35-h27087fc_0.conda#da0ec11a6454ae19bff5b02ed881a2b1
 https://conda.anaconda.org/conda-forge/linux-64/openssl-3.2.0-hd590300_1.conda#603827b39ea2b835268adb8c821b8570
 https://conda.anaconda.org/conda-forge/linux-64/pixman-0.43.0-h59595ed_0.conda#6b4b43013628634b6cfdee6b74fd696b
@@ -159,6 +160,7 @@ https://conda.anaconda.org/conda-forge/noarch/toml-0.10.2-pyhd8ed1ab_0.tar.bz2#f
 https://conda.anaconda.org/conda-forge/noarch/tomli-2.0.1-pyhd8ed1ab_0.tar.bz2#5844808ffab9ebdb694585b50ba02a96
 https://conda.anaconda.org/conda-forge/linux-64/tornado-6.3.3-py311h459d7ec_1.conda#a700fcb5cedd3e72d0c75d095c7a6eda
 https://conda.anaconda.org/conda-forge/noarch/typing_extensions-4.9.0-pyha770c72_0.conda#a92a6440c3fe7052d63244f3aba2a4a7
+https://conda.anaconda.org/conda-forge/noarch/wheel-0.42.0-pyhd8ed1ab_0.conda#1cdea58981c5cbc17b51973bcaddcea7
 https://conda.anaconda.org/conda-forge/linux-64/xcb-util-image-0.4.0-h8ee46fc_1.conda#9d7bcddf49cbf727730af10e71022c73
 https://conda.anaconda.org/conda-forge/linux-64/xkeyboard-config-2.40-hd590300_0.conda#07c15d846a2e4d673da22cbd85fdb6d2
 https://conda.anaconda.org/conda-forge/linux-64/xorg-libxext-1.3.4-h0b41bf4_2.conda#82b6df12252e6f32402b96dacc656fec
@@ -173,9 +175,12 @@ https://conda.anaconda.org/conda-forge/noarch/joblib-1.3.2-pyhd8ed1ab_0.conda#4d
 https://conda.anaconda.org/conda-forge/linux-64/libclang-15.0.7-default_hb11cfb5_4.conda#c90f4cbb57839c98fef8f830e4b9972f
 https://conda.anaconda.org/conda-forge/linux-64/libgoogle-cloud-2.12.0-hac9eb74_1.conda#0dee716254497604762957076ac76540
 https://conda.anaconda.org/conda-forge/linux-64/libxkbcommon-1.6.0-h5d7e998_0.conda#d8edd0e29db6fb6b6988e1a28d35d994
+https://conda.anaconda.org/conda-forge/noarch/meson-1.3.1-pyhd8ed1ab_0.conda#54744574be599bff37ee4c3624ed02d2
 https://conda.anaconda.org/conda-forge/linux-64/mkl-2022.2.1-h84fe81f_16997.conda#a7ce56d5757f5b57e7daabe703ade5bb
 https://conda.anaconda.org/conda-forge/linux-64/pillow-10.2.0-py311ha6c5da5_0.conda#a5ccd7f2271f28b7d2de0b02b64e3796
+https://conda.anaconda.org/conda-forge/noarch/pip-23.3.2-pyhd8ed1ab_0.conda#8591c748f98dcc02253003533bc2e4b1
 https://conda.anaconda.org/conda-forge/linux-64/pulseaudio-client-16.1-hb77b528_5.conda#ac902ff3c1c6d750dd0dfc93a974ab74
+https://conda.anaconda.org/conda-forge/noarch/pyproject-metadata-0.7.1-pyhd8ed1ab_0.conda#dcb27826ffc94d5f04e241322239983b
 https://conda.anaconda.org/conda-forge/noarch/pytest-7.4.4-pyhd8ed1ab_0.conda#a9d145de8c5f064b5fa68fb34725d9f4
 https://conda.anaconda.org/conda-forge/noarch/python-dateutil-2.8.2-pyhd8ed1ab_0.tar.bz2#dd999d1cc9f79e67dbb855c8924c7984
 https://conda.anaconda.org/conda-forge/linux-64/sip-6.7.12-py311hb755f60_0.conda#02336abab4cb5dd794010ef53c54bd09
@@ -184,6 +189,7 @@ https://conda.anaconda.org/conda-forge/linux-64/blas-1.0-mkl.tar.bz2#349aef876b1
 https://conda.anaconda.org/conda-forge/linux-64/gstreamer-1.22.8-h98fc4e7_1.conda#1b52a89485ab573a5bb83a5225ff706e
 https://conda.anaconda.org/conda-forge/linux-64/harfbuzz-8.3.0-h3d44ed6_0.conda#5a6f6c00ef982a9bc83558d9ac8f64a0
 https://conda.anaconda.org/conda-forge/linux-64/libblas-3.9.0-16_linux64_mkl.tar.bz2#85f61af03fd291dae33150ffe89dc09a
+https://conda.anaconda.org/conda-forge/noarch/meson-python-0.15.0-pyh0c530f3_0.conda#3bc64565ca78ce3bb80248d09926d8f9
 https://conda.anaconda.org/conda-forge/linux-64/pyqt5-sip-12.12.2-py311hb755f60_5.conda#e4d262cc3600e70b505a6761d29f6207
 https://conda.anaconda.org/conda-forge/noarch/pytest-cov-4.1.0-pyhd8ed1ab_0.conda#06eb685a3a0b146347a58dda979485da
 https://conda.anaconda.org/conda-forge/noarch/pytest-xdist-3.5.0-pyhd8ed1ab_0.conda#d5f595da2daead898ca958ac62f0307b

diff --git a/build_tools/azure/pylatest_conda_forge_mkl_linux-64_environment.yml b/build_tools/azure/pylatest_conda_forge_mkl_linux-64_environment.yml
@@ -21,6 +21,8 @@ dependencies:
   - pytest-cov
   - coverage
   - ccache
+  - meson-python
+  - pip
   - pytorch=1.13
   - pytorch-cpu
   - polars

diff --git a/build_tools/build-meson-editable-install.py b/build_tools/build-meson-editable-install.py
@@ -0,0 +1,45 @@
+import re
+import shlex
+import subprocess
+from pathlib import Path
+
+
+def main():
+    pyproject_path = Path("pyproject.toml")
+
+    if not pyproject_path.exists():
+        raise SystemExit(
+            "Can not find pyproject.toml. You should run this script from the"
+            " scikit-learn root folder."
+        )
+
+    old_pyproject_content = pyproject_path.read_text(encoding="utf-8")
+    if 'build-backend = "mesonpy"' not in old_pyproject_content:
+        new_pyproject_content = re.sub(
+            r"\[build-system\]",
+            r'[build-system]\nbuild-backend = "mesonpy"',
+            old_pyproject_content,
+        )
+        pyproject_path.write_text(new_pyproject_content, encoding="utf-8")
+
+    command = shlex.split(
+        "pip install --editable .  --verbose --no-build-isolation "
+        "--config-settings editable-verbose=true"
+    )
+
+    exception = None
+    try:
+        subprocess.check_call(command)
+    except Exception as e:
+        exception = e
+    finally:
+        pyproject_path.write_text(old_pyproject_content, encoding="utf-8")
+
+    if exception is not None:
+        raise RuntimeError(
+            "There was some error when running the script"
+        ) from exception
+
+
+if __name__ == "__main__":
+    main()
diff --git a/build_tools/update_environments_and_lock_files.py b/build_tools/update_environments_and_lock_files.py
@@ -98,6 +98,8 @@ def remove_from(alist, to_remove):
         "channel": "conda-forge",
         "conda_dependencies": common_dependencies + [
             "ccache",
+            "meson-python",
+            "pip",
             "pytorch",
             "pytorch-cpu",
             "polars",

diff --git a/doc/developers/advanced_installation.rst b/doc/developers/advanced_installation.rst
@@ -201,6 +201,88 @@ It is however preferred to use pip.
 On Unix-like systems, you can equivalently type ``make in`` from the top-level
 folder. Have a look at the ``Makefile`` for additional utilities.
 
+.. _building_with_meson:
+
+Building with Meson
+-------------------
+
+Support for Meson is experimental, in scikit-learn 1.5.0.dev0.
+`Open an issue <https://github.com/scikit-learn/scikit-learn/issues/new>`__ if
+you encounter any problems!
+
+Make sure you have `meson-python` and `ninja` installed, either with `conda`:
+
+.. code-block:: bash
+
+    conda install -c conda-forge meson-python ninja -y
+
+or with pip:
+
+.. code-block:: bash
+
+    pip install meson-python ninja
+
+Simplest way to build with Meson
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+To build scikit-learn, the simplest way is to run:
+
+.. code-block:: bash
+
+    make dev-meson
+
+You need to do it once after this you can run your code that imports `sklearn`
+and it will recompile as needed.
+
+In case you want to go back to using setuptools:
+
+.. code-block:: bash
+
+    make clean-meson
+
+More advanced way to build with Meson
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+If you can not use `make`, want to do it yourself or understand what goes in
+behind the scenes, you can edit `pyproject.toml` and make sure `build-backend`
+is set to `"mesonpy"`
+
+.. code-block:: toml
+
+    [build-system]
+    build-backend = "mesonpy"
+
+Build with the following `pip` command:
+
+.. code-block:: bash
+
+    pip install --editable . \
+        --verbose --no-build-isolation \
+        --config-settings editable-verbose=true
+
+If you want to go back to using `setuptools`:
+
+.. code-block:: bash
+
+    pip uninstall -y scikit-learn
+
+Note `--config-settings editable-verbose=true` is advised to avoid surprises.
+meson-python implements editable install by recompiling when doing `import
+sklearn`. Even changing python files involves copying files to the Meson build
+directory. You will see the meson output when that happens, rather than
+potentially waiting a while and wondering what is taking so long. Bonus: that
+means you only have to do the `pip install` once, after that your code will
+recompile when doing `import sklearn`.
+
+Other places that may be worth looking at:
+
+- `pandas setup doc
+  <https://pandas.pydata.org/docs/development/contributing_environment.html#step-3-build-and-install-pandas>`_:
+  pandas has a similar setup as ours (no spin or dev.py)
+- `scipy Meson doc
+  <https://scipy.github.io/devdocs/building/understanding_meson.html>`_ gives
+  more background about how Meson works behind the scenes
+
 .. _platform_specific_instructions:
 
 Platform-specific instructions

diff --git a/doc/whats_new/v1.5.rst b/doc/whats_new/v1.5.rst
@@ -11,6 +11,16 @@ Version 1.5.0
 
 .. include:: changelog_legend.inc
 
+Support for building with Meson
+-------------------------------
+
+Meson is now supported as a build backend, see :ref:`Building with Meson
+<building_with_meson>` for more details.
+
+:pr:`28040` by :user:`Loïc Estève <lesteve>`
+
+TODO Fill more details before the 1.5 release, when the Meson story has settled down.
+
 Changelog
 ---------
 

diff --git a/meson.build b/meson.build
@@ -0,0 +1,53 @@
+project(
+  'scikit-learn',
+  'c', 'cpp', 'cython',
+  version: run_command('sklearn/_build_utils/version.py', check: true).stdout().strip(),
+  license: 'BSD-3',
+  meson_version: '>= 1.1.0',
+  default_options: [
+    'buildtype=debugoptimized',
+    'c_std=c99',
+    'cpp_std=c++14',
+  ],
+)
+
+cc = meson.get_compiler('c')
+cpp = meson.get_compiler('cpp')
+
+# Check compiler is recent enough (see "Toolchain Roadmap" for details)
+if cc.get_id() == 'gcc'
+  if not cc.version().version_compare('>=8.0')
+    error('scikit-learn requires GCC >= 8.0')
+  endif
+elif cc.get_id() == 'msvc'
+  if not cc.version().version_compare('>=19.20')
+    error('scikit-learn requires at least vc142 (default with Visual Studio 2019) ' + \
+          'when building with MSVC')
+  endif
+endif
+
+_global_c_args = cc.get_supported_arguments(
+  '-Wno-unused-but-set-variable',
+  '-Wno-unused-function',
+  '-Wno-conversion',
+  '-Wno-misleading-indentation',
+)
+add_project_arguments(_global_c_args, language : 'c')
+
+# We need -lm for all C code (assuming it uses math functions, which is safe to
+# assume for scikit-learn). For C++ it isn't needed, because libstdc++/libc++ is
+# guaranteed to depend on it.
+m_dep = cc.find_library('m', required : false)
+if m_dep.found()
+  add_project_link_arguments('-lm', language : 'c')
+endif
+
+tempita = files('sklearn/_build_utils/tempita.py')
+
+py = import('python').find_installation(pure: false)
+
+# Copy all the .py files to the install dir, rather than using
+# py.install_sources and needing to list them explicitely one by one
+install_subdir('sklearn', install_dir: py.get_install_dir())
+
+subdir('sklearn')
diff --git a/sklearn/__check_build/meson.build b/sklearn/__check_build/meson.build
@@ -0,0 +1,7 @@
+py.extension_module(
+  '_check_build',
+  '_check_build.pyx',
+  cython_args: cython_args,
+  install: true,
+  subdir: 'sklearn/__check_build',
+)
diff --git a/sklearn/__init__.py b/sklearn/__init__.py
@@ -133,6 +133,14 @@
         "show_versions",
     ]
 
+    _BUILT_WITH_MESON = False
+    try:
+        import sklearn._built_with_meson  # noqa: F401
+
+        _BUILT_WITH_MESON = True
+    except ModuleNotFoundError:
+        pass
+
 
 def setup_module(module):
     """Fixture for the tests to assure globally controllable seeding of RNGs"""

diff --git a/sklearn/_build_utils/tempita.py b/sklearn/_build_utils/tempita.py
@@ -0,0 +1,57 @@
+import argparse
+import os
+
+from Cython import Tempita as tempita
+
+# XXX: If this import ever fails (does it really?), vendor either
+# cython.tempita or numpy/npy_tempita.
+
+
+def process_tempita(fromfile, outfile=None):
+    """Process tempita templated file and write out the result.
+
+    The template file is expected to end in `.c.tp` or `.pyx.tp`:
+    E.g. processing `template.c.in` generates `template.c`.
+
+    """
+    with open(fromfile, "r", encoding="utf-8") as f:
+        template_content = f.read()
+
+    template = tempita.Template(template_content)
+    content = template.substitute()
+
+    with open(outfile, "w", encoding="utf-8") as f:
+        f.write(content)
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("infile", type=str, help="Path to the input file")
+    parser.add_argument("-o", "--outdir", type=str, help="Path to the output directory")
+    parser.add_argument(
+        "-i",
+        "--ignore",
+        type=str,
+        help=(
+            "An ignored input - may be useful to add a "
+            "dependency between custom targets"
+        ),
+    )
+    args = parser.parse_args()
+
+    if not args.infile.endswith(".tp"):
+        raise ValueError(f"Unexpected extension: {args.infile}")
+
+    if not args.outdir:
+        raise ValueError("Missing `--outdir` argument to tempita.py")
+
+    outdir_abs = os.path.join(os.getcwd(), args.outdir)
+    outfile = os.path.join(
+        outdir_abs, os.path.splitext(os.path.split(args.infile)[1])[0]
+    )
+
+    process_tempita(args.infile, outfile)
+
+
+if __name__ == "__main__":
+    main()