settled-reach/docs/sprints/sprint-30/server.md

Sprint 30: Clean Slate — Server Tasks

Goal: Close Phase 1 wiki quality debt — stereotype correction, data integrity fixes, and atlas regressions — so the wiki stands correct before Phase 1 UI work begins.

Branch: server Agents: Dudley (dev), Tyre (arch), Hoshe (QA)

New Tickets

#	Title	Priority	Blocked by
#762	fix(atlas): habitable_planet_count filter uses 'breathable' but proposals use 'standard'	medium	—
#752	Add settings.db creation-on-missing routine + gitignore	medium	—
#744	Add corridor-status command to atlas CLI	medium	—

Use tooling/db/ticket show <id> for full details.

Key Decisions

• decisions/architecture.md — D-094 (chunk size 32, district 256×256), D-139 (composable behavior primitives)
• decisions/scope.md — D-114 (server simulation is Phase 1 infrastructure)

Notes

#762 — habitable_planet_count filter mismatch

File: server/src/bin/atlas.rs, cmd_commit_system function, line ~1293.

The filter reads:

b.atmosphere.as_deref() == Some("breathable")

But all Track B/C/D body catalog proposals use atmosphere: "standard", not "breathable". As a result, habitable_planet_count is written as 0 for every committed system regardless of actual habitability. Fix the filter value to "standard". Verify by querying the committed systems table after the fix: tooling/db/sqlite-query "SELECT gj_id, habitable_planet_count FROM star_systems WHERE habitable_planet_count > 0 LIMIT 10" should return non-zero counts after recommitting a test system.

If both "breathable" and "standard" are valid atmosphere values that should count as habitable, update the filter to accept both: matches!(b.atmosphere.as_deref(), Some("breathable") | Some("standard")).

Do not retroactively fix existing committed systems — the fix applies to new commits. If you want to backfill, do it in a separate migration with a clear comment.

#752 — settings.db creation-on-missing

File: server/src/main.rs lines 158-169, server/src/settings/ module.

main.rs calls SettingsStore::open(&settings_path) but if settings.db does not exist the call logs a warning and continues without settings persistence — it does not crash, but it silently degrades. Improve this:

1. In SettingsStore::open() (likely in server/src/settings/store.rs), if the file does not exist, create it with default schema. Use rusqlite::Connection::open() which creates the file if absent, then run the settings schema CREATE TABLE IF NOT EXISTS DDL on first open.
2. Add settings.db to .gitignore (it is a runtime artifact, not source).
3. Verify: delete settings.db, run the server binary, confirm settings.db is created and the server starts without warnings about missing settings.

Schema reference: Check server/src/settings/types.rs and server/src/settings/store.rs to understand the existing schema and what tables need to be created on first run.

#744 — corridor-status command for atlas CLI

File: server/src/bin/atlas.rs.

The atlas binary already has a subcommand infrastructure (clap-based, see lines 32-40). Add a corridor-status subcommand that shows remaining unfinished systems grouped by geographic_sector and hop_distance. Currently this is done via ad-hoc raw SQL queries.

What to deliver:

• New CorridorStatus subcommand struct in atlas.rs
• Implementation queries star_systems for systems where body catalog is incomplete (no committed bodies, or status != 'complete' if such a field exists)
• Groups output by geographic_sector, then by hop_distance within each sector
• Output format: plain text table, corridor → hop → count of remaining systems, one row per corridor/hop combination
• Add to the atlas binary's help text

Query basis: The equivalent SQL is approximately:

SELECT geographic_sector, hop_distance, COUNT(*) as remaining
FROM star_systems
WHERE body_catalog_complete = 0 OR body_catalog_complete IS NULL
GROUP BY geographic_sector, hop_distance
ORDER BY geographic_sector, hop_distance;

Verify the actual column name against db/schema.sql.

Dependency Chain

#762 (breathable filter fix) — standalone
#752 (settings.db create) — standalone
#744 (corridor-status cmd) — standalone

All three tickets are parallel tracks. No cross-dependencies.

PR Workflow

When ready to submit, create a PR with tea CLI. All flags are required to avoid TTY prompts (see CLAUDE.md "Gitea access" section):

tea pr create --repo jpmschweitzer/settled-reach --login schweitz --title "fix(server): Sprint 30 atlas filter, settings.db init, corridor-status cmd" --description "body" --base main --head server