Book Image

Speed Up Your Python with Rust

By : Maxwell Flitton
5 (2)
Book Image

Speed Up Your Python with Rust

5 (2)
By: Maxwell Flitton

Overview of this book

Python has made software development easier, but it falls short in several areas including memory management that lead to poor performance and security. Rust, on the other hand, provides memory safety without using a garbage collector, which means that with its low memory footprint, you can build high-performant and secure apps relatively easily. However, rewriting everything in Rust can be expensive and risky as there might not be package support in Rust for the problem being solved. This is where Python bindings and pip come in. This book will help you, as a Python developer, to start using Rust in your Python projects without having to manage a separate Rust server or application. Seeing as you'll already understand concepts like functions and loops, this book covers the quirks of Rust such as memory management to code Rust in a productive and structured manner. You'll explore the PyO3 crate to fuse Rust code with Python, learn how to package your fused Rust code in a pip package, and then deploy a Python Flask application in Docker that uses a private Rust pip module. Finally, you'll get to grips with advanced Rust binding topics such as inspecting Python objects and modules in Rust. By the end of this Rust book, you'll be able to develop safe and high-performant applications with better concurrency support.
Table of Contents (16 chapters)
1
Section 1: Getting to Understand Rust
5
Section 2: Fusing Rust with Python
11
Section 3: Infusing Rust into a Web Application

Summary

In this chapter, we went through the basics of building a simple catastrophe model. We then broke down the logic and converted it into steps so that we could build the catastrophe model in Rust. This included taking in paths, loading data from files, including data in our package, and building a Python interface so that our users do not have to know about what is going on under the hood when constructing a model. After all of this, we tested our module and ensured that we kept increasing the data size of the test to see how it scales. We saw that, initially, our Rust solution was faster because Rust is faster than Python and pandas. However, our implementation did not scale well, as we did a loop within a loop for our merge.

As the data size increased, our Rust code ended up being slower. In previous chapters, we have shown multiple times that Rust implementations are generally faster. However, this does not counteract the effects of bad code implementation. If you are relying...