gh-106917: fix super classmethod calls to non-classmethods#106977
Merged
carljm merged 4 commits intopython:mainfrom Jul 24, 2023
Merged
gh-106917: fix super classmethod calls to non-classmethods#106977carljm merged 4 commits intopython:mainfrom
carljm merged 4 commits intopython:mainfrom
Conversation
* main: pythongh-106004: Add PyDict_GetItemRef() function (python#106005)
Member
Fidget-Spinner
left a comment
Fidget-Spinner
left a comment
There was a problem hiding this comment.
Thanks. This is quite the subtlety. LGTM in general. I just have a comment about the tests.
* main: (73 commits) Thoroughly refactor the cases generator (python#107151) Docs: Add missing markup to Argument Clinic docs (python#106876) pythongh-107162: Document errcode.h usage in its comment (python#107177) pythongh-106320: Remove private _PyDict C API (python#107145) Fix PyVectorcall_Function doc versionadded (python#107140) Docs: Remove duplicate word in Argument Clinic howto heading (python#107169) pythongh-107017: Change Chapter Strings to Texts in the Introduction chapter. (python#107104) pythongh-106320: Remove private _PyObject C API (python#107159) Docs: fix typo in os.pwrite docstring (python#107087) pythongh-105291: Add link to migration guide for distutils (python#107130) pythongh-106948: Docs: Disable links for C standard library functions, OS utility functions and system calls (python#107062) pythongh-106320: Remove _PyBytes_Join() C API (python#107144) pythongh-106320: Remove private _PyObject C API (python#107147) pythongh-106320: Remove _PyTuple_MaybeUntrack() C API (python#107143) pythongh-106320: Remove _PyIsSelectable_fd() C API (python#107142) Remove superflous whitespaces in `layout.html`. (pythonGH-107067) pythongh-107122: Update what's news for dbm.*dbm.clear() method (pythongh-107135) pythongh-107122: Add clear method to dbm.ndbm module (pythongh-107126) pythongh-62519: Make pgettext search plurals when translation is not found (python#107118) pythongh-107122: Add clear method to dbm.gdbm.module (pythongh-107127) ...
Fidget-Spinner
approved these changes
Jul 24, 2023
Contributor
|
Thanks @carljm for the PR 🌮🎉.. I'm working now to backport this PR to: 3.12. |
Contributor
|
Sorry, @carljm, I could not cleanly backport this to |
carljm
added a commit
to carljm/cpython
that referenced
this pull request
Jul 24, 2023
…ds (pythonGH-106977). (cherry picked from commit e5d5522) Co-authored-by: Carl Meyer <carl@oddbird.net>
|
GH-107204 is a backport of this pull request to the 3.12 branch. |
carljm
added a commit
that referenced
this pull request
Jul 24, 2023
jtcave
pushed a commit
to jtcave/cpython
that referenced
this pull request
Jul 27, 2023
facebook-github-bot
pushed a commit
to facebookincubator/cinder
that referenced
this pull request
Mar 30, 2025
Summary: Fixes a couple of issues which are causing `test_super` to fail on 3.12: 1) We don't actually respect the global super object at all! We are using the exported Cinder function to implement this on 3.10, but just calling the equivalent of `Ci_Super_Lookup` on 3.12. Because we need to have the `global_super != &PySuper_Type` logic on 3.12 this just brings it over and stops exporting it from Cinder 3.10. 3) We get `no_args_in_super_call` backwards as oparg is effectively the number of arguments. 2) Ports the fix in python/cpython#106977 to the implementation of super in both the JIT and Cinder 3.10. Reviewed By: alexmalyshev Differential Revision: D71638145 fbshipit-source-id: 0f5a189d8cea573034e578c921925ce2988c3bb1
facebook-github-bot
pushed a commit
to facebookincubator/cinderx
that referenced
this pull request
Mar 30, 2025
Summary: Fixes a couple of issues which are causing `test_super` to fail on 3.12: 1) We don't actually respect the global super object at all! We are using the exported Cinder function to implement this on 3.10, but just calling the equivalent of `Ci_Super_Lookup` on 3.12. Because we need to have the `global_super != &PySuper_Type` logic on 3.12 this just brings it over and stops exporting it from Cinder 3.10. 3) We get `no_args_in_super_call` backwards as oparg is effectively the number of arguments. 2) Ports the fix in python/cpython#106977 to the implementation of super in both the JIT and Cinder 3.10. Reviewed By: alexmalyshev Differential Revision: D71638145 fbshipit-source-id: 0f5a189d8cea573034e578c921925ce2988c3bb1
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The fix for #105035 was subtly incomplete. It checked that
cls->tp_getattro == PyObject_GetGenericAttr(as a precondition for attempting the load-method optimization), but what we really need to know is thatPy_TYPE(self)->tp_getattro == PyObject_GenericGetAttr. In the typical case ofsuper(MyClass, instance)...these are effectively the same thing (instancewill be an instance ofMyClass), but in a classmethod super() call (super(MyClass, MyClass)...) they are quite different; the type ofMyClassistype, which has a differenttp_getattro.For actual classmethods this was fine; the load-method optimization still works correctly for classmethods looked up on types. But it doesn't work for regular methods (plain function objects) looked up on types; those are just regular function calls with no bound method behavior.
It's pretty unusual to use classmethod-style
superwhere the call target is a regular function, not a classmethod. But it can occur, e.g. if usingsuperwith__new__and dynamically-assigned__new__methods, as in #106917.cls.__new__andsuper()#106917