PEP 793: Updates from discussion & implementation

peps/pep-0387.rst

@@ -114,6 +114,8 @@ Basic policy for backwards compatibility
   platforms).
 
 
+.. _pep387-soft-deprecation:
+
 Soft Deprecation
 ================
 

peps/pep-0793.rst

@@ -26,6 +26,10 @@ To make this viable, we also specify new module slot types to replace
 
 We also add an API for defining modules from slots dynamically.
 
+The existing API (``PyInit_*``) is soft-deprecated.
+(That is: it will continue to work without warnings, and it'll be fully
+documented and supported, but we plan to not add any new features to it.)
+
 
 Background & Motivation
 =======================
@@ -147,6 +151,27 @@ Unlike types, the import mechanism often has a pointer that's known to be
 suitable as a token value; in these cases it can provide a default token.
 Thus, module tokens do not need a variant of the inelegant ``Py_TP_USE_SPEC``.
 
+To help extensions that straddle Python versions, ``PyModuleDef`` addresses
+are used as default tokens, and where it's reasonable, they are made
+interchangeable with tokens.
+
+
+Soft-deprecating the existing export hook
+-----------------------------------------
+
+The only reason for authors of *existing* extensions to switch to the
+API proposed here is that it allows a single module for both free-threaded
+and non-free-threaded builds.
+It is important that Python *allows* that, but for many existing modules,
+it is nowhere near worth losing compatibility with 3.14 and lower versions.
+
+It is much too early to plan deprecation of the old API.
+
+Instead, this PEP proposes to stop adding new features to the ``PyInit_*``
+scheme.
+After all, the perfect time for extension authors to switch is when they want
+to modify module initialization anyway.
+
 
 Specification
 =============
@@ -264,6 +289,10 @@ If specified, using a new ``Py_mod_token`` slot, the module token must:
 
 (Typically, it should point to a static constant.)
 
+When the address of a ``PyModuleDef`` is used as a module's token,
+the module should behave as if it was created from that ``PyModuleDef``.
+In particular, the module state must have matching layout and semantics.
+
 Modules created using the ``PyModule_FromSlotsAndSpec`` or the
 ``PyModExport_<NAME>`` export hook can use a new ``Py_mod_token`` slot
 to set the token.
@@ -288,8 +317,15 @@ will return 0 on success and -1 on failure:
     int PyModule_GetToken(PyObject *, void **token_p)
 
 A new ``PyType_GetModuleByToken`` function will be added, with a signature
-like ``PyType_GetModuleByDef`` but a ``void *token`` argument,
-and the same behaviour except matching tokens, rather than only defs.
+like the existing ``PyType_GetModuleByDef`` but a ``void *token`` argument,
+and the same behaviour except matching tokens rather than only defs.
+
+For easier backwards compatibility, the existing ``PyType_GetModuleByDef``
+will be changed to work exactly like ``PyType_GetModuleByToken`` -- that is,
+it will allow a token (cast to a ``PyModuleDef *`` pointer) as the
+*def* argument.
+(The ``PyModule_GetDef`` function will not get a similar change, as users may
+access members of its result.)
 
 
 New slots
@@ -333,6 +369,14 @@ via a pointer; the function will return 0 on success and -1 on failure:
     int PyModule_GetStateSize(PyObject *, Py_ssize_t *result);
 
 
+Soft-deprecating the existing export hook
+-----------------------------------------
+
+The ``PyInit_*`` export hook will be
+:ref:`soft-deprecated <pep387-soft-deprecation>`.
+
+
+
 .. _pep793-api-summary:
 
 New API summary
@@ -383,9 +427,14 @@ If an existing module is ported to use the new mechanism, then
 We claim that how a module was defined is an implementation detail of that
 module, so this should not be considered a breaking change.
 
-Similarly, ``PyType_GetModuleByDef`` will not match modules that are not
-defined using a *def*.
-The new ``PyType_GetModuleByToken`` function may be used instead.
+Similarly, the ``PyType_GetModuleByDef`` function may stop matching modules
+whose definition changed. Module authors may avoid this by explicitly
+setting a *def* as the *token*.
+
+``PyType_GetModuleByDef`` will now accept a module token as the *def* argument.
+We specify a suitable restriction on using ``PyModuleDef`` addresses as tokens,
+and non-``PyModuleDef`` pointers were previously invalid input,
+so this is not a backwards-compatibility issue.
 
 The ``Py_mod_create`` function may now be called with ``NULL`` for the second
 argument.
@@ -412,6 +461,9 @@ Here is a guide to convert an existing module to the new API, including
 some tricky edge cases.
 It should be moved to a HOWTO in the documentation.
 
+This guide is meant for hand-written modules. For code generators and language
+wrappers, the :ref:`pep793-shim` below may be more useful.
+
 #. Scan your code for uses of ``PyModule_GetDef``. This function will
    return ``NULL`` for modules that use the new mechanism. Instead:
 
@@ -425,11 +477,15 @@ It should be moved to a HOWTO in the documentation.
      Later in this guide, you'll set the token to *be* the existing
      ``PyModuleDef`` structure.
 
-#. Scan your code for uses of ``PyType_GetModuleByDef``, and replace them by
-   ``PyType_GetModuleByToken``.
+#. Optionally, scan your code for uses of ``PyType_GetModuleByDef``,
+   and replace them with ``PyType_GetModuleByToken``.
    Later in this guide, you'll set the token to *be* the existing
    ``PyModuleDef`` structure.
 
+   (You may skip this step if targetting Python versions that don't expose
+   ``PyType_GetModuleByToken``, since ``PyType_GetModuleByDef`` is
+   backwards-compatible.)
+
 #. Look at the function identified by ``Py_mod_create``, if any.
    Make sure that it does not use its second argument (``PyModuleDef``),
    as it will be called with ``NULL``.
@@ -464,18 +520,17 @@ It should be moved to a HOWTO in the documentation.
     };
 
 #. If you switched from ``PyModule_GetDef`` to ``PyModule_GetToken``,
-   and/or from ``PyType_GetModuleByDef`` to ``PyType_GetModuleByToken``,
+   and/or if you use ``PyType_GetModuleByDef`` or ``PyType_GetModuleByToken``,
    add a ``Py_mod_token`` slot pointing to the existing ``PyModuleDef`` struct:
 
    .. code-block:: c
 
     static PyModuleDef_Slot module_slots[] = {
         // ... (keep existing slots here)
-        {Py_mod_token, your_module_def},
+        {Py_mod_token, &your_module_def},
         {0}
     };
 
-
 #. Add a new export hook.
 
    .. code-block:: c
@@ -489,9 +544,55 @@ It should be moved to a HOWTO in the documentation.
     }
 
 The new export hook will be used on Python 3.15 and above.
-Once your module no longer supports lower versions, delete the ``PyInit_``
-function and any unused data.
+Once your module no longer supports lower versions:
+
+#. Delete the ``PyInit_`` function.
+
+#. If the existing ``PyModuleDef`` struct is used *only* for ``Py_mod_token``
+   and/or ``PyType_GetModuleByToken``, you may remove the ``Py_mod_token``
+   line and replace ``&your_module_def`` with ``module_slots`` everywhere else.
+
+#. Delete any unused data.
+   The ``PyModuleDef`` struct and the original slots array are likely to be
+   unused.
+
+
+.. _pep793-shim:
+
+Backwards compatibility shim
+----------------------------
+
+It is possible to write a generic function that implements the “old” export
+hook (``PyInit_``) in terms of the API proposed here.
 
+The following implementation can be copied and pasted to a project; only the
+names ``PyInit_examplemodule`` (twice) and ``PyModExport_examplemodule`` should
+need adjusting.
+
+When added to the :ref:`pep793-example` below and compiled with a
+non-free-threaded build of this PEP's reference implementation, the resulting
+extension is compatible with non-free-threading 3.9+ builds, in addition to a
+free-threading build of the reference implementation.
+(The module must be named without a version tag, e.g. ``examplemodule.so``,
+and be placed on ``sys.path``.)
+
+Full support for creating such modules will require backports of some new
+API, and support in build/install tools. This is out of scope of this PEP.
+(In particular, the demo “cheats” by using a subset of Limited API 3.15 that
+*happens to work* on 3.9; a proper implementation would use Limited API 3.9
+with backport shims for new API like ``Py_mod_name``.)
+
+This implementation places a few additional requirements on the slots array:
+
+- Slots that correspond to ``PyModuleDef`` members must come first.
+- A ``Py_mod_name`` slot is required.
+- Any ``Py_mod_token`` must be set to ``&module_def_and_token``, defined here.
+
+It also passes ``NULL`` as *spec* to the ``PyModExport`` export hook.
+A proper implementation would pass ``None`` instead.
+
+.. literalinclude:: pep-0793/shim.c
+   :language: c
 
 
 Security Implications
@@ -507,6 +608,8 @@ In addition to regular reference docs, the :ref:`pep793-porting-notes` should
 be added as a new HOWTO.
 
 
+.. _pep793-example:
+
 Example
 =======
 

peps/pep-0793/examplemodule.c

@@ -49,9 +49,9 @@ PyDoc_STRVAR(examplemodule_doc, "Example extension.");
 static PyModuleDef_Slot examplemodule_slots[] = {
     {Py_mod_name, "examplemodule"},
     {Py_mod_doc, (char*)examplemodule_doc},
-    {Py_mod_exec, (void*)examplemodule_exec},
     {Py_mod_methods, examplemodule_methods},
     {Py_mod_state_size, (void*)sizeof(examplemodule_state)},
+    {Py_mod_exec, (void*)examplemodule_exec},
     {0}
 };
 

peps/pep-0793/shim.c

@@ -0,0 +1,68 @@
+#include <string.h>     // memset
+
+PyMODINIT_FUNC PyInit_examplemodule(void);
+
+static PyModuleDef module_def_and_token;
+
+PyMODINIT_FUNC
+PyInit_examplemodule(void)
+{
+    PyModuleDef_Slot *slot = PyModExport_examplemodule(NULL);
+
+    if (module_def_and_token.m_name) {
+        // Take care to only set up the static PyModuleDef once.
+        // (PyModExport might theoretically return different data each time.)
+        return PyModuleDef_Init(&module_def_and_token);
+    }
+    int copying_slots = 1;
+    for (/* slot set above */; slot->slot; slot++) {
+        switch (slot->slot) {
+        // Set PyModuleDef members from slots. These slots must come first.
+#       define COPYSLOT_CASE(SLOT, MEMBER, TYPE)                            \
+            case SLOT:                                                      \
+                if (!copying_slots) {                                       \
+                    PyErr_SetString(PyExc_SystemError,                      \
+                                    #SLOT " must be specified earlier");    \
+                    goto error;                                             \
+                }                                                           \
+                module_def_and_token.MEMBER = (TYPE)(slot->value);          \
+                break;                                                      \
+            /////////////////////////////////////////////////////////////////
+        COPYSLOT_CASE(Py_mod_name, m_name, char*)
+        COPYSLOT_CASE(Py_mod_doc, m_doc, char*)
+        COPYSLOT_CASE(Py_mod_state_size, m_size, Py_ssize_t)
+        COPYSLOT_CASE(Py_mod_methods, m_methods, PyMethodDef*)
+        COPYSLOT_CASE(Py_mod_state_traverse, m_traverse, traverseproc)
+        COPYSLOT_CASE(Py_mod_state_clear, m_clear, inquiry)
+        COPYSLOT_CASE(Py_mod_state_free, m_free, freefunc)
+        case Py_mod_token:
+            // With PyInit_, the PyModuleDef is used as the token.
+            if (slot->value != &module_def_and_token) {
+                PyErr_SetString(PyExc_SystemError,
+                                "Py_mod_token must be set to "
+                                "&module_def_and_token");
+                goto error;
+            }
+            break;
+        default:
+            // The remaining slots become m_slots in the def.
+            // (`slot` now points to the "rest" of the original
+            //  zero-terminated array.)
+            if (copying_slots) {
+                module_def_and_token.m_slots = slot;
+            }
+            copying_slots = 0;
+            break;
+        }
+    }
+    if (!module_def_and_token.m_name) {
+        // This function needs m_name as the "is initialized" marker.
+        PyErr_SetString(PyExc_SystemError, "Py_mod_name slot is required");
+        goto error;
+    }
+    return PyModuleDef_Init(&module_def_and_token);
+
+error:
+    memset(&module_def_and_token, 0, sizeof(module_def_and_token));
+    return NULL;
+}