feat(acquire): name the station-selection release, and file v0.4.0 as shipped

[sgjholt]

Important

Merge this with a merge commit. Do not squash. Squashing is what broke the last one: it replaces both subjects with the pull request title, and a merge commit keeps them so release-please sees feat(acquire): and docs(roadmap): as separate entries. A squash still produces the right version — the Release-As: 0.5.0 footer survives into the body — but collapses the two changelog lines into one.

Two commits. One restores a release record that a squash destroyed; the other files v0.4.0 on the roadmap.

1. feat(acquire): — the release record, and the docs it owed

#45 was squash-merged, which replaced fix(acquire): ask for waveforms by name… and feat(acquire): rank co-sited instruments… with the PR title. release-please could not parse it:

commit could not be parsed: 2af57e8 Make the station selection mean what the
  config says: radius, and one instrument per station (#45)
error message: Error: unexpected token ' ' at 1:5, valid tokens [(, !, :]
commits: 0
✔ Considering: 0 commits
✔ No commits for path: ., skipping

So the station-selection work is invisible to the release machinery: no bump, and no changelog entry for a release that ships it. The message on main cannot be repaired in place, so the record is restored here where release-please reads it, with Release-As: 0.5.0 pinning the version rather than leaving it to the type.

0.5.0, not 0.4.1 — feat(acquire): rank co-sited instruments is a feature, and with bump-minor-pre-major: true and bump-patch-for-minor-pre-major: false a feature bumps the minor while on 0.x.

The commit carries the documentation that change owed regardless. docs/upgrading.md had no entry for:

• a radius that now cuts against true WGS84 distance instead of a fixed 111.195 km/degree, which can select a different set of stations for the same config;
• min_radius_km beginning to work on its own, having previously been sent only alongside max_radius_km;
• max_radius_km = 0, a negative radius and an inverted annulus now raising instead of silently selecting nothing;
• channel_priorities / location_priorities.

It was also missing for 0.3 → 0.4, where [smoothing] method began being applied — the same shape of break, where the call is unchanged and the numbers move. That section is added too.

2. docs(roadmap): — v0.4.0 filed as shipped

The roadmap had no v0.4.0 section a day after v0.4.0 released, and its closing paragraph still said nothing was merged and unreleased while the acquire fixes sat on main.

TestTheRoadmapKeepsUpWithTheReleases compares the page against the second-newest minor in CHANGELOG.md, because the newest is release-please's proposal rather than something a reader can install. That deliberate lag is what stops the check blocking its own release PR — and it also means the missing v0.4 section would have gone unnoticed until v0.5 shipped, then failed on main. Verified by injecting a simulated ## [0.5.0] section: 19 passed with this change.

The 1.0 criterion is updated too, since a release has now gone past it. v0.4.0 moved no name and no signature, but [smoothing] method changed what it does, so a config setting konno_ohmachi or none gets different numbers from the same call. That is a weaker promise than 1.0 makes and still one that was broken, so the page counts it rather than claiming a clean release. Worth a second opinion — it is a judgement about your release, not a fact I can check.

Verification

pytest tests/test_release_config.py tests/test_api_surface.py   99 passed, 14 skipped
  test_release_config with a simulated 0.5.0 in CHANGELOG.md    19 passed
sphinx-build                                                    clean (bar blocked intersphinx here)

Docs only — no source file changes in either commit.

🤖 Generated with Claude Code

[read-the-docs-community]

Documentation build overview

📚 SpecMod | 🛠️ Build #34635876 | 📁 Comparing b460d19 against latest (2af57e8)

  🔍 Preview build  

3 files changed

± roadmap.html
± upgrading.html
± tutorial/specmod-tutorial.html