chore: release 0.7.0

[FBumann]

Prompt: cut 0.6.3

Prompt: 0.7.0

Note

The following content was generated by AI.

Releases 0.7.0 on merge. The one change is #1857: specsolve requires polars 2.0. That is a minor release, because an upgrade pulls polars 2 into the user's environment, and polars 2 can break their own polars code.

Method and gates

• ## Upcoming version → ## 0.7.0 (2026-10-07), with a two-sentence paragraph on top that names the break and the way out (stay on 0.6.2). An empty ## Upcoming version stays above it.
• Why minor, not patch. The repository rule names a broken model file or import. This release breaks neither, but specsolve users write polars code, and pip install -U specsolve now moves them to polars 2. A user who pinned ~=0.6 would get that through a patch. An environment that caps polars below 2 resolves to 0.6.2 either way.
• v0.6.2..origin/main in src/ is feat(deps): specsolve runs on polars 2.0 and requires it #1857 only. answer_layout.py does not change, so archives written by 0.6.x still read back.
• python -m tools.changelog check → releases 0.7.0 on merge.
• pixi run check passes on 05677de with the same changelog edit under the heading 0.6.3: 5012 passed, 564 skipped, 1 xfailed. The later edit changes only the heading.
• The title follows chore: release 0.6.2 #1875 (chore: release 0.6.2).

🤖 Generated with Claude Code

https://claude.ai/code/session_01EpzN6XKNByxsfi3GhX3v3D

---

Generated by Claude Code

[read-the-docs-community]

Documentation build overview

📚 specsolve | 🛠️ Build #34989980 | 📁 Comparing ebd457e against latest (05677de)

  🔍 Preview build  

1 file changed

± about/changelog/index.html

[codspeed]

Merging this PR will regress 6 benchmarks

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 21 improved benchmarks
❌ 6 regressed benchmarks
✅ 40 untouched benchmarks
⏩ 307 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_emit[fleet-s-specsolve-lp]	26.2 MB	32.1 MB	-18.28%
❌	test_emit[commitment-s-specsolve-lp]	2.6 MB	3.2 MB	-17.81%
❌	test_emit[nodal-s-specsolve-lp]	6 MB	6.9 MB	-12.39%
❌	test_emit[storage-s-specsolve-lp]	21.7 MB	24.5 MB	-11.38%
❌	test_read[commitment-s-specsolve-parquet]	695.3 KB	774 KB	-10.17%
❌	test_read[storage-s-specsolve-parquet]	2.6 MB	2.8 MB	-7.35%
⚡	test_window[sector-s-specsolve-highs-values]	5.5 MB	4.1 MB	+33.36%
⚡	test_window[sector-s-specsolve-highs-cold]	5.3 MB	4.1 MB	+28.6%
⚡	test_window[sector-s-specsolve-highs-one]	5.3 MB	4.1 MB	+28.35%
⚡	test_window[sector-s-specsolve-highs-coefficient]	5.3 MB	4.2 MB	+27.77%
⚡	test_emit[sector-s-specsolve-highs]	5.3 MB	4.2 MB	+27.18%
⚡	test_window[sector-s-specsolve-highs-shape]	5.3 MB	4.1 MB	+26.9%
⚡	test_emit[sector-s-specsolve-lp]	5.3 MB	4.6 MB	+14.8%
⚡	test_window[storage-s-specsolve-highs-values]	13.8 MB	12.9 MB	+7.1%
⚡	test_window[storage-s-specsolve-highs-one]	13.8 MB	12.9 MB	+7.1%
⚡	test_window[fleet-s-specsolve-highs-values]	17.4 MB	16.3 MB	+6.76%
⚡	test_window[fleet-s-specsolve-highs-one]	17.4 MB	16.3 MB	+6.76%
⚡	test_emit[storage-s-specsolve-highs]	16.9 MB	15.9 MB	+6.51%
⚡	test_window[nodal-s-specsolve-highs-values]	4.8 MB	4.5 MB	+6.34%
⚡	test_emit[fleet-s-specsolve-highs]	21.9 MB	20.7 MB	+5.9%
...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing claude/friendly-dirac-t7g8rd (ebd457e) with main (04aced1)²}

Footnotes

1. 307 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 (05677de) during the generation of this report, so 04aced1 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩