Proposal to Remove CPython's Global Interpreter Lock (GIL) and Introduce a New Lock in Python 3.12

According to the Python Software Foundation blog, Sam Gross introduced a new proposal at the 2022 Python Language Summit to completely remove the Global Interpreter Lock (GIL) from the CPython interpreter, aiming to improve performance for multithreaded Python programs.

CPython remains the core implementation of Python, and because its memory management is not thread‑safe, the GIL was introduced as a mutex that allows only one thread to execute Python bytecode at a time, preventing race conditions.

While the GIL protects object access, it also limits the ability of multithreaded programs to utilize multiple CPU cores, and many packages have become tightly coupled with the GIL, making its removal a daunting task. Previous attempts, such as Larry Hastings' "Gilectomy" project, failed because they caused significant slowdowns in single‑threaded code.

The new "nogil" project, led by Meta developer Sam Gross, builds on the lessons from Gilectomy and proposes adding a new type of lock—described as a "more lucky" lock—to maintain thread safety without the GIL. Implementing this new lock is challenging, as it may introduce performance regressions in certain areas.

Gross plans to introduce a compiler flag in Python 3.12 that disables the GIL, potentially bringing the new lock mechanism to the official release. The proposal has generated lively discussion among the Python community.

What are your thoughts on this proposal? Share your comments below.