PEP 489: Multi-phase extension module initialization

Known limitations of the current implementation:

- documentation changes are incomplete
- there's a reference leak I haven't tracked down yet

The leak is most visible by running:

  ./python -m test -R3:3 test_importlib

However, you can also see it by running:

  ./python -X showrefcount

Importing the array or _testmultiphase modules, and
then deleting them from both sys.modules and the local
namespace shows significant increases in the total
number of active references each cycle. By contrast,
with _testcapi (which continues to use single-phase
initialisation) the global refcounts stabilise after
a couple of cycles.

1 files changed, 4 insertions, 3 deletions

diff --git a/Python/dynload_win.c b/Python/dynload_win.c
index 9ed9eea333..96f1a09f8a 100644
--- a/Python/dynload_win.c
+++ b/Python/dynload_win.c
@@ -186,8 +186,9 @@ static char *GetPythonImport (HINSTANCE hModule)
     return NULL;
 }
 
-dl_funcptr _PyImport_GetDynLoadWindows(const char *shortname,
-                                       PyObject *pathname, FILE *fp)
+dl_funcptr _PyImport_FindSharedFuncptrWindows(const char *prefix,
+                                              const char *shortname,
+                                              PyObject *pathname, FILE *fp)
 {
     dl_funcptr p;
     char funcname[258], *import_python;
@@ -201,7 +202,7 @@ dl_funcptr _PyImport_GetDynLoadWindows(const char *shortname,
     if (wpathname == NULL)
         return NULL;
 
-    PyOS_snprintf(funcname, sizeof(funcname), "PyInit_%.200s", shortname);
+    PyOS_snprintf(funcname, sizeof(funcname), "%20_%.200s", prefix, shortname);
 
     {
         HINSTANCE hDLL = NULL;