Import sklearn._distributor_init first (scikit-learn#15929)

ogrisel · Pan Jan · commit 1ac6586b4be7 · 2020-03-03T10:48:12.000+01:00
diff --git a/sklearn/__init__.py b/sklearn/__init__.py
@@ -70,12 +70,18 @@
     # We are not importing the rest of scikit-learn during the build
     # process, as it may not be compiled yet
 else:
-    from . import __check_build
+    # `_distributor_init` allows distributors to run custom init code.
+    # For instance, for the Windows wheel, this is used to pre-load the
+    # vcomp shared library runtime for OpenMP embedded in the sklearn/.libs
+    # sub-folder.
+    # It is necessary to do this prior to importing show_versions as the
+    # later is linked to the OpenMP runtime to make it possible to introspect
+    # it and importing it first would fail if the OpenMP dll cannot be found.
+    from . import _distributor_init  # noqa: F401
+    from . import __check_build  # noqa: F401
     from .base import clone
     from .utils._show_versions import show_versions
 
-    __check_build  # avoid flakes unused variable error
-
     __all__ = ['calibration', 'cluster', 'covariance', 'cross_decomposition',
                'datasets', 'decomposition', 'dummy', 'ensemble', 'exceptions',
                'experimental', 'externals', 'feature_extraction',
@@ -90,9 +96,6 @@
                'clone', 'get_config', 'set_config', 'config_context',
                'show_versions']
 
-    # Allow distributors to run custom init code
-    from . import _distributor_init  # noqa: F401
-
 
 def setup_module(module):
     """Fixture for the tests to assure globally controllable seeding of RNGs"""
