fix(reference): preserve station evidence and rollback schema

This commit is contained in:
jb
2026-09-19 09:53:46 -04:00
parent f7b604f40f
commit 24660edb09
8 changed files with 83 additions and 40 deletions
+2 -2
View File
@@ -21,7 +21,7 @@ bun workers/reference.ts --production
bun workers/reference.ts --production --offline --archive-dir data/reference-live
```
The default production archive and artifact directory is `data/reference-live` (ignored by git); production publishes `data/reference-live/manifest.json`, while fixture builds remain under `data/reference`. An explicit `--directory` wins over `REFERENCE_DIR`, which wins over these defaults. `REFERENCE_LIVE_DIR` controls archive storage only. Compose explicitly publishes to `/app/data/reference`, the shared worker/web volume; the production web reader rejects fixture-profile artifacts. Runtime images must provide the `xz` executable (Debian package `xz-utils`) for streaming `.tar.xz` decompression; ZIP deflate uses the Bun runtime. No npm archive dependency is required. Production validation requires more than 5,000 systems, at least one station, and imports CCP's `mapSolarSystems`, `mapConstellations`, `mapRegions`, `mapStargates`, and `npcStations` JSONL members. A production candidate missing `npcStations.jsonl` or containing no station rows fails closed rather than claiming station coverage. The importer parses official station ID/system/type/owner/position facts; names remain unknown unless a separately verified name-bearing adapter is selected. The current worktree does not download or promote a production artifact, so deployed station coverage remains unverified until an isolated replay validates the member and its joins.
The default production archive and artifact directory is `data/reference-live` (ignored by git); production publishes `data/reference-live/manifest.json`, while fixture builds remain under `data/reference`. An explicit `--directory` wins over `REFERENCE_DIR`, which wins over these defaults. `REFERENCE_LIVE_DIR` controls archive storage only. Compose explicitly publishes to `/app/data/reference`, the shared worker/web volume; the production web reader rejects fixture-profile artifacts. Runtime images must provide the `xz` executable (Debian package `xz-utils`) for streaming `.tar.xz` decompression; ZIP deflate uses the Bun runtime. No npm archive dependency is required. Production validation requires more than 5,000 systems, at least one station, and imports CCP's `mapSolarSystems`, `mapConstellations`, `mapRegions`, `mapStargates`, and `npcStations` JSONL members. A production candidate missing `npcStations.jsonl` or containing no station rows fails closed rather than claiming any station rows. The importer parses official station ID/system/type/owner/position facts; names remain unknown unless a separately verified name-bearing adapter is selected. The production profile records station coverage as explicitly `incomplete` until a measured release replay proves source-consistent coverage; `stationsMinimum: 1` is only a non-empty safety gate, not a completeness claim. The current worktree does not download or promote a production artifact, so deployed station coverage remains unverified until an isolated replay validates the member, its joins, and its measured count.
`fixtures/reference/` contains small schema-faithful EVE Ref keyed records for categories, groups, types, and blueprints (including types 34 Tritanium, 35 Pyerite, 587 Rifter, and blueprint 687). The Rifter recipe is synthetic and is labelled as such. Hoboleaks `industryactivities`, `repackagedvolumes`, and `meta` are independent raw maps. `systems.json` is explicitly an ESI-universe fixture: it preserves the official system fields and XYZ coordinate object and is not presented as a Reference Data `/systems` record.
@@ -59,7 +59,7 @@ The lock prevents concurrent builds. A crash can leave a lock, staging directory
## Rollback and checks
```sh
REFERENCE_DIR=data/reference bun workers/reference.ts --rollback
REFERENCE_DIR=data/reference bun workers/reference.ts --fixture --rollback
REFERENCE_DIR=data/reference bun workers/reference.ts --check --fixture
bun workers/reference.ts --help
```
+1 -1
View File
@@ -166,7 +166,7 @@ Hoboleaks metadata also carries per-file `md5`, `revision`, `stale`, and `deprec
The CCP JSON Lines SDE includes `npcStations.jsonl`; CCP's [schema changelog](https://developers.eveonline.com/static-data/tranquility/schema-changelog.yaml) records the station-schema migration and the [Map Data guide](https://developers.eveonline.com/docs/guides/map-data/) establishes the map hierarchy. A bounded inspection of the build-3503375 station member found 5,210 records with `_key`, `solarSystemID`, `ownerID`, `typeID`, and nested `position` (`x`, `y`, `z`); none carried a `name` field. The third-party schema page was used only as a cross-check, not as authority; this worktree did not download or promote the production archive.
The importer therefore treats CCP station IDs, system relationships, owner/type IDs, and coordinates as source facts, while station names remain unknown for the official SDE. The normalized station contract permits an optional `name` only when a separately verified name-bearing adapter supplies it; the production CCP adapter ignores all name-like fields and never derives a name from an ID or type. A station record must reference a system in the same candidate; malformed position/IDs or mismatched supplied system station lists fail closed. The selected production CCP parser requires `npcStations.jsonl` and the production validation profile requires at least one parsed station row; a missing member or empty station set fails closed instead of degrading to fabricated or incomplete station coverage.
The importer therefore treats CCP station IDs, system relationships, owner/type IDs, and coordinates as source facts, while station names remain unknown for the official SDE. The normalized station contract permits an optional `name` only when a separately verified name-bearing adapter supplies it; the production CCP adapter ignores all name-like fields and never derives a name from an ID or type. A station record must reference a system in the same candidate; malformed position/IDs or mismatched supplied system station lists fail closed. The selected production CCP parser requires `npcStations.jsonl` and the production validation profile requires at least one verified station row; a missing member or empty station set fails closed instead of degrading to fabricated coverage. This is explicitly an incomplete station-coverage profile until an isolated replay measures a source-consistent release count; it is not a universal completeness threshold.
This worktree has not downloaded or replayed the production ZIP, so actual release member presence and station row coverage remain an acceptance blocker for deployed NPC station coverage.