Explore Pyodide, a Python distribution for the browser and Node.js based on WebAssembly, in this PyCon US talk. Discover how to run Python applications in the browser, learn about porting existing Python packages, and understand the criteria for determining project suitability. Delve into Pyodide's components, including CPython 3.9 ported to WebAssembly/Emscripten, and its robust Javascript ⟺ Python foreign function interface. Examine examples of Python utils from JavaScript, random sampling, and using the fetch API from Python. Investigate the Emscripten host environment, client-only architecture, and the growing ecosystem of notebook environments and educational applications. Learn about deploying machine learning models, including scikit-learn models in Pyodide, and explore challenges such as function pointer cast handling and making Python package sizes web-friendly. Gain insights into Pyodide's roadmap and its potential impact on serverless Python apps for the web.
Overview
Syllabus
Intro
Serverless Python apps for the web?
What is WebAssembly?
The emscripten build toolchain
Pyodide Components
Upstream CPython WASM work
Related projects
Pure Python packages with micropip
Packages with binary extensions
Supported Python packages in Pyodide
Foreign function interface (JS # Python)
Example: Python utils from JavaScript
Example: random sample
Examples: fetch API from Python
Examples: Buffers
Emscripten Host Environment
Client-only Architecture
Client-only Web Apps in Python
A growing ecosystem
Notebook environments
Pyodide in Education
Deploying machine learning models Classical workflow
Deploying scikit-learn models in Pyodide Use pickle?
Classifier decision boundary example
Packaging Scipy and Fortran
Function Pointer Cast Handling
Getting http.client to work (WIP)
Asyncio in the browser
Download sizes for packages
Make Python package sizes web friendly
Roadmap
Acknowledgement
Taught by
PyCon US
