Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
82 changes: 82 additions & 0 deletions _projects/2026/astropy/astropy_core_improvements.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,82 @@
---
name: Hardening astropy's core stability
# Add a short one line description of your project
desc: Improvements to astropy low level extension modules (C/C++/Cython)
# Student requirements:
requirements:
- at least one low level language (C or C++ prefered)
- testing in Python
# How difficult would you consider the project to be: low, medium, high
difficulty: medium to high, depending on candidate
# Related issues (if any) to this project.
# Ideally you want at least one that describes the problem it is trying to solve
issues:
- https://github.com/neutrinoceros/astropy-APEs/pull/1
- https://github.com/astropy/astropy/issues/17760
- https://github.com/astropy/astropy/issues/19249

# First person in contact; mentors may change before project starts.
# GitHub or GitLab handles
mentors:
- neutrinoceros
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will there be a second person?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

possibly. I'm reaching out for another mentor, but I can't confirm anything yet.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it a hard requirement?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

While I don't think its one for GSoC, they do recommend strongly to have multiple mentors and we like to try and enforce two people. It is common that one mentor can be busy and it is very useful for a student to have someone else to fall back on.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It is absolutely a hard requirement. If you cannot convince 1-2 other people from the project to co-mentor this idea; then I'm afraid we won't run it.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

fair enough. What would be the deadline on this ?

# The programme under this project wish to run. At the moment only GSOC is available
initiatives:
- GSOC
project_size:
- 175 h
# Different technologies needed
tags:
- python
- meson-python
- C
- C++
- Cython
- pytest
# suborganisation(s) to which this project belongs.
collaborating_projects:
- astropy
---

# Hardening astropy's core stability

## Description

Astropy is a mixed-language code base. Most of it is pure Python, but many hotpaths are
written in lower level languages (Cython, C and C++). The former is where the vast
majority of past and current development is conducted, while the latter in much more
slowly paced. Meanwhile, the low-level layer overwhemingly dominates other important
aspects like complexity of *building* and distributing astropy releases, and takes an
estimated 20 to 50% share of astropy CI time for recompilation *of code that almost never changes*.

It would be possible to decouple the bulk of the code base from its low level layer,
for instance by splitting the latter into a separate package, but this new package would
be insufficiently tested on its own, as much of astropy's test suite currently only
checks the behavior of the public API, which doesn't intersect with the low level layer.

We want to build a more fundamental test suite for this low level layer, making it much
more independent from astropy's existing tests.

This project would be highly related to (although separable from) the [following draft
APE](https://github.com/neutrinoceros/astropy-APEs/pull/1), which also contains more details on why the separation is desirable

As such, it would also provide opportunities for exploratory works like
- migrating to a modern, externally maintained build system like Meson
- rewritting, or incorporating `rust`-based extensions
- exploring PEP 803/809 Limited API for Python 3.15

All of the above would be in scope for this project, depending on the candidate's own
interests and time allocated.

## Project Milestones

### Coding starts

* TBD

### 1st evaluation

* TBD

### Final evaluation

* TBD