chore: release 0.2.1

[FBumann]

Prompt: "Merged. Lets prepare a release" … "We want 1774 in there too!"

Note

The following content was generated by AI.

Merging this releases 0.2.1. It is a patch: three fixes for wrong answers that raised no error (#1784, #1779, #1776), the sign of a dual stated on Result.dual (#1774), and no break of a file, archive or import. The diff is CHANGELOG.md alone.

What the section says, and what was checked

• ## Upcoming version became ## 0.2.1 (2026-09-29), with a paragraph on who is affected by each fix, a paragraph on the dual's sign, and then the four pull requests, each listed once. A new, empty ## Upcoming version sits above it.
• Why a patch. No construct was renamed or refused, no import moved, and nothing a result, sweep or archive writes to disk changed: relational/parquet.py is unchanged since v0.2.0, so LAYOUT is too. pyproject.toml changed only tool pins in the dev and docs groups. The fixes change values that were wrong, and docs(api): a dual is the rate at which the optimal objective rises with its row's right side, on every sink #1774 changes no value.
• python -m tools.changelog check prints releases 0.2.1 on merge. python -m tools.changelog notes 0.2.1 prints the section as the release notes. tests/test_changelog.py passes.
• Left out, as their types owe no line: test: every expression is read at one chosen point on both lanes, which sees a term in a row that never binds #1781 (test), test: a program that leaves a side of a bound open builds with that side unbounded in both engines #1737 (test), chore: bump the python group across 1 directory with 6 updates #1761 (chore) and ci: bump the actions group across 1 directory with 4 updates #1760 (ci).
• On merge, release.yaml tags v0.2.1, opens the GitHub release, builds, and waits for a reviewer to approve the pypi environment before it uploads.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Q7hhaASVuhgSVvaXkmUpEh

[codspeed]

Merging this PR will not alter performance

✅ 24 untouched benchmarks
⏩ 82 skipped benchmarks¹

---

_{Comparing claude/nice-davinci-wduwk5 (6d86d6b) with main (2ff1cff)²}

Footnotes

1. 82 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

2. No successful run was found on main (5ce2d56) during the generation of this report, so 2ff1cff was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[read-the-docs-community]

Documentation build overview

📚 specsolve | 🛠️ Build #34825111 | 📁 Comparing 6d86d6b against latest (854c6b4)

  🔍 Preview build  

2 files changed

± about/changelog/index.html
± reference/api/index.html