I'm working on submitting my Google Summer of Code 2026 proposal to OpenAstronomy this week, and I wanted to write a quick post about it.
The project is called Hardening Astropy's Core Stability. The short version: Astropy has C and Cython extensions powering its most performance-critical code, and none of them have direct tests. They're only tested indirectly through the public Python API, which means a bug deep in the compiled layer can sit undetected for a long time. My project is to build a test suite that targets that layer directly.
It's also groundwork for something bigger — there's a proposal in the Astropy community to eventually split the compiled extensions into a separate package. That can't happen safely without standalone tests first.
I've been contributing to Astropy since December 2025 — trying to fix the core and complex issues. That work gave me enough codebase context to write a proposal grounded in real problems I already ran into, not just ideas on paper.
If it goes well, I'll be spending the summer reading C source code, mapping calling conventions, and making a part of Astropy that almost nobody sees a lot more reliable. That sounds exactly like the kind of work I want to be doing.
Fingers crossed. 🤞
Top comments (0)