GoKawiil
Tech News
A project using Opus 5.5 has generated 613,318 lines of Lean code that mirror CPython's internals function-by-function, testing each rewrite against real Python before attempting formal proofs. A second AI agent independently verifies each proof before it counts as accepted, and 4,071 functions have passed this check so far, yielding 18,898 theorems and lemmas across 1,539 files.
The maintainers of CPython's experimental just-in-time compiler have withdrawn PEP 836, a proposal meant to satisfy the Python steering council's June request for a standards-track plan addressing maintenance, compatibility, and success metrics. The JIT, added by Brandt Bucher in 2024 under the informational PEP 744, had been paused on CPython's main branch pending that approval. With the withdrawal of PEP 836, work on the JIT project appears to have stopped.
APSW (Another Python SQLite Wrapper) is an open-source library that exposes the complete SQLite C API to Python, including features like full text search, sessions, virtual tables, VFS, JSON, and CArray. It supports CPython 3.10 and later, works with synchronous code and async frameworks, and is distributed via PyPI and GitHub under an OSI-approved open source license.
At the Python Language Summit, software engineer David Hewitt presented a proposed timeline and success criteria for incorporating Rust into CPython, following up on discussions from last year's PyCon. The effort, led by core developers Kirill Podoprigora and Emma Smith, now involves roughly 60 contributors including some Python core developers and Rust project members, and includes a draft PEP. The plan calls for Python distributors to test optional Rust support starting with Python 3.16, due October 2027.
PyPy has shipped v8.0.0, updating its Linux build infrastructure to manylinux_2_28 images based on AlmaLinux 8, glibc 2.28, and gcc14, replacing the older gcc5 toolchain. The release also lays groundwork for supporting cp312-abi3 wheels by restructuring how PyPy's internal PyObject extension field is stored and by no longer renaming exported limited-API function names.
An investigation into Python signal handling shows that Python signal handlers, unlike C ones, run outside the strict low-level signal context but can still be re-entered mid-execution if another signal arrives. A test script that rapidly sends SIGUSR1 signals to itself while calling print() inside the handler triggers a RuntimeError, since Python's buffered stdout object detects the reentrant write and refuses to proceed.
The official CPython documentation lays out Big O complexity tables for core operations on built-in types like list and tuple, covering actions such as append, insert, slicing, concatenation, and membership testing. It notes these figures apply specifically to CPython's exact built-in types, and that subclasses or other Python implementations may behave differently.