We are supported by readers, when you click & purchase through links on our site we earn affiliate commission. Learn more.

Python implementation: Pyston team packs packages for Conda

The Pyston team has presented a roadmap for the further development of the performance-oriented Python implementation. At the top of the list is providing precompiled packages to pick up more developers. Then a connection to 64-bit ARM is pending, and later implementations for Windows and macOS will follow. Other performance optimizations are also on the list as a constant companion.

The roadmap can also be seen as a sign of life, as it was quiet around Pyston for a while and the implementation was supposedly about to end four years ago. Dropbox first presented the project in 2014, which, inspired by JavaScript implementations such as Google’s V8 engine, relies on JIT (just-in-time) compilation.

In 2017, Dropbox withdrew from development, with which some saw the end of Pyston, especially since other projects such as PyPy and Cython are also preparing to optimize the performance of Python code. In October 2020, however, version 2.0 was initially released as a closed source. Pyston has been an open source project since May and the team partnered with Anaconda in August.

The declared goal for further development is to get more developers on board. Apparently the biggest hurdle is the availability of Python packages. In contrast to CPython, only a few packages are available precompiled. The team has therefore announced that it will compile numerous packages and make them available via the Conda package manager. The team then plans a Pyston implementation for 64-bit ARM. In addition, it wants to set up a CI / CD system (Continuous Integration, Continuous Delivery) for further development.

In the long-term planning, porting to macOS and Windows is pending. In addition, Pyston is to work together with the JIT compiler Numba, which translates numerical functions as a subset of Python and NumPy into machine code. Finally, improved multithreading is on the roadmap. In addition, the team is considering an “opt-in” feature that should enable changes to the semantics.

Future Pyston releases will also always be designed for a specific Python version. It is currently in Python 3.8.12, and the move to Python 3.10 is planned for early 2022. In addition, the team wants to incorporate some of the performance efforts in Python 3.11 as a backport into the implementation.

Let more details join the Pyston blog and the Wiki entry on the roadmap remove. Of the Source code for Pyston can be found on GitHub. There are in the Release section ready-made packages for download ready.

(rme)

To home page