Pyston Takes A Big Step Forward In Programming Languages

 The Python programming language has a lot of potential for machine-learning and data science, but it is slow.

Anaconda provides Python for data scientists as a top-rated distribution. Now, they want to make that change by supporting Pyston, a new Python implementation that reduces debugging for speed.

Pyston, a Python program created by Kevin Modzelewski was released open-source in May. The promise of a 30% speedup in Python code was made. Modzelewski was an engineer for Dropbox. Dropbox was a large user of Python. Guido van Rossum hired Modzelewski to help improve its code.

Anaconda hired Modzelewski along with Marius Wachtler, the lead developer, to build the project’s community of users and contributors. This will help ensure that it is sustainable for many years.

Modzelewski released a statement saying that Anaconda support would allow them to deliver Pyston faster than ever. Anaconda claims that it has over 25 million users.

Anaconda claims that Pyston executes Python programs between 20% and 50% faster than standard Python.

Dropbox created the Python implementation between 2014-2017. It was launched in 2020 as Pyston, v2.

Pyston is a derived project from the Python Software Foundation's official CPython. It will remain open-source. Anaconda will help improve Python's compatibility with the legion of Python packages that have helped it become a dominant data science and machine-learning tool.

Anaconda explains how the Pyston 2.x Series is a complete rewrite starting with a fork CPython 3.8.

Python is a popular choice for machine-learning programs running on powerful hardware but it isn't well-known in JavaScript-dominated web and mobile apps.

Peter Wang, Anaconda co-founder and CEO, recently told ZDNet that using Python to build or distribute applications that actually have graphical user interfaces was difficult.

He explained that Python is not the primary language of an operating system on desktops. Instead, it must use third-party frameworks like Qt, wxPython, or both.

Python's strengths lie not only in data science but also in connecting backend systems.

Microsoft employee Van Rossum is now working on Python version 3.11. It is one of three Python branches, with another planned for 2022. Version 3.9.7, the latest stable Python version, is now available.

Anaconda has been involved in Python optimization, Scalability and Performance projects.

Anaconda's oldest open source project is the Numba compiler. This LLVM based JIT compiler for numerical Python operations runs on the CPU/GPU. Because of this, we've been thinking a lot about Python compilers. We see the potential for Pyston for quickly bringing faster Python to a more mainstream audience.

"Numba does a good job in many numerical cases, but cannot optimize entire programs. And it doesn't address the wider Python use cases. Pyston takes a different approach to the Python compilation challenge. Despite the differences in ancestry, Numba works well with the CPython program. Both systems can be used simultaneously within the same program. Pyston is able to improve the performance and speed up certain functions by as much as 2-10x.

Anaconda is also of the opinion that Pyston improvements can easily be upstreamed from CPython to CPython. These developments are in line with van Rossum’s Microsoft plans to significantly accelerate Python.