Astropy: Hardening Astropy's Core Stability (Dhruv Yadav)

[dhruv1955]

Requirements

• I've read OA contributor application guide;
• I confirm that this is my work and not made by LLMs;
• I've introduced myself to the sub-organisation I'm applying for;
• I've created a blog post and wrote something about the project I'm interested;
• I've contributed at least one patch to the sub-organisation I'm applying for and it's my work and not made by LLMs
  (Check for issues marked "easy" or "package-novice" in each organization. Don't be afraid to contact the development mailing list if you have questions or run into trouble setting up a development environment.):
• BUG: Fix missing dereference in use_mad_std allocation check in fast_sigma_clip.c astropy/astropy#19457
• TST: Add direct tests for _convolveNd_c Cython extension astropy/astropy#19455
• TST: add direct tests for _parse_times C extension in time astropy/astropy#19410
• TST: add direct tests for scalar_inv_efuncs Cython extension in cosmology astropy/astropy#19407
• TST: Add direct tests for ks_2samp Cython extension in stats._stats astropy/astropy#19378
• TST: Add direct tests for _sigma_clip_fast C extension astropy/astropy#19377
• DOC: add examples for masked=False behavior in sigma_clip astropy/astropy#19376
• BUG: fix pickling of SigmaClip when Bottleneck is installed astropy/astropy#19372
• BUG: Fix missing goto fail in create_parser after ValueError in parse_times.c astropy/astropy#19368
• BUG: Fix numpy.ediff1d returning incorrect unit for magnitudes astropy/astropy#19360
• BUG: Fix AssertionError for numpy bool index in _handle_index_argument astropy/astropy#19358
• tests(io.registry): replace TODO-skipped get_formats tests with real coverage astropy/astropy#19357
• BUG: Fix return vs raise in ShapedLikeNDArray.take() when out is passed astropy/astropy#19351

[dhruv1955]

@dpshelio could you regenerate the PDF with the latest changes? The proposal has been updated with additional PRs, timeline dates, and other comments section.

Also, should I submit this auto-generated PDF to the GSoC portal, or is there a different version I should use?

[dhruv1955]

@neutrinoceros @astrofrog @nstarman - tagging you here as my GSoC 2026 proposal mentors. The proposal reflects work already underway - 9 of 14 PRs merged, including direct extension tests for _parse_times.c (#19410) and scalar_inv_efuncs.pyx (#19407), and a C-level bug fix (#19457) found by reading source directly. Happy to answer any questions.

[neutrinoceros]

Hi @dhruv1955, first of all, thank you for all your contributions so far. However I need to say that we don't want to select a candidate based on existing contributions explicitly made towards the project, for fear that it would create an incentive for applicants to essentially do the work for free before the project even starts, as well as bias our selection towards candidates that have more free time than others.

It's a tough spot to be in. You're obviously very motivated and the record shows your work is very much appreciated, but I need to temper your expectations.

[dhruv1955]

@neutrinoceros thank you for being direct - I appreciate the transparency. I understand the concern about creating unfair incentives.

My contributions were driven by curiosity about the codebase rather than as a strategy for selection, but I understand why that distinction can be difficult to assess externally.

I'll let the proposal speak for itself, and I'm looking forward to the evaluation process.