Cue-sheet caching so each route is enriched once #30

New issue

Closed

opened 2026-08-31 17:14:46 +02:00 by robert · 1 comment

robert commented

2026-08-31 17:14:46 +02:00

(Migrated from git.butzei.de)

Goal

Enrichment is slow and possibly online. Do it at import, never during a ride.

Acceptance criteria

Cue sheet serialised to disk, keyed by route id and enricher version
Cache invalidated when the route or the enricher version changes
Ride start loads from cache with no network and no BRouter round-trip
Which backend produced the cached sheet is recorded and displayed
Manual re-enrich available, for example after installing BRouter

Files

companion/.../route/enrich/CueSheetCache.kt

## Goal Enrichment is slow and possibly online. Do it at import, never during a ride. ## Acceptance criteria - [ ] Cue sheet serialised to disk, keyed by route id and enricher version - [ ] Cache invalidated when the route or the enricher version changes - [ ] Ride start loads from cache with no network and no BRouter round-trip - [ ] Which backend produced the cached sheet is recorded and displayed - [ ] Manual re-enrich available, for example after installing BRouter ## Files - `companion/.../route/enrich/CueSheetCache.kt`

robert added a new dependency

2026-09-03 19:01:19 +02:00

#26 CueSheetEnricher interface and canonical direction enum

robert added a new dependency

2026-09-03 19:01:25 +02:00

#55 Cue-sheet review on the phone before riding

robert referenced this issue from a commit

2026-09-04 14:03:17 +02:00

Cache cue sheets so each route is enriched once (#30)

robert referenced this issue from a pull request that will close it,

2026-09-04 14:03:42 +02:00

Cache cue sheets so each route is enriched once (#30) #89

robert closed this issue

2026-09-04 14:04:45 +02:00

robert referenced this issue from a commit

2026-09-04 14:04:46 +02:00

Merge pull request 'Cache cue sheets so each route is enriched once (#30)' (#89) from area/cue-sheet-cache into main

robert commented

2026-09-04 14:04:59 +02:00

Owner

Closed by PR #89 (merged): CueSheetCache (companion/core, pure JVM, 13 new host tests incl. one real end-to-end run against GpxCueEnricher + the real bikerouter-style-cues.gpx fixture) keyed by route id + a single pipeline-wide CURRENT_ENRICHER_VERSION constant \u2014 covers both a tier's own logic changing and a new tier slot being wired in later, no per-tier versioning needed. Backed by a new Room cue_sheets table with a real Migration(1,2) (checked field-for-field against CueSheetEntity: routeId/enricherVersion/cuesEncoded NOT NULL, tier nullable \u2014 matches). Which backend won is recorded via the existing RouteLibrary.updateEnrichment() hook, so RouteLibraryScreen's status line now shows real data with no new UI needed there; a manual re-enrich button was added.\n\nHonestly flagged, not papered over: raw GPX text (needed by tier 0) isn't persisted anywhere in the route library yet \u2014 GpxImporter discards it after parsing. getOrEnrich/forceReEnrich take it as optional; absent, tier 0 fails closed (same behaviour it already has for malformed input) and falls through to tier 1+. Not a regression \u2014 no caller anywhere can supply it post-import today anyway \u2014 documented as a separate follow-up (persisting rawGpx) rather than solved here.\n\nVerification split, same pattern as every companion PR tonight: :companion:core (the cache/codec logic) is host-tested and green. :companion:route (Room codegen, the migration actually running, UI wiring) is hand-reviewed only \u2014 no Android SDK in this sandbox.

Closed by PR #89 (merged): CueSheetCache (companion/core, pure JVM, 13 new host tests incl. one real end-to-end run against GpxCueEnricher + the real bikerouter-style-cues.gpx fixture) keyed by route id + a single pipeline-wide CURRENT_ENRICHER_VERSION constant \u2014 covers both a tier's own logic changing and a new tier slot being wired in later, no per-tier versioning needed. Backed by a new Room `cue_sheets` table with a real Migration(1,2) (checked field-for-field against CueSheetEntity: routeId/enricherVersion/cuesEncoded NOT NULL, tier nullable \u2014 matches). Which backend won is recorded via the existing RouteLibrary.updateEnrichment() hook, so RouteLibraryScreen's status line now shows real data with no new UI needed there; a manual re-enrich button was added.\n\n**Honestly flagged, not papered over**: raw GPX text (needed by tier 0) isn't persisted anywhere in the route library yet \u2014 GpxImporter discards it after parsing. getOrEnrich/forceReEnrich take it as optional; absent, tier 0 fails closed (same behaviour it already has for malformed input) and falls through to tier 1+. Not a regression \u2014 no caller anywhere can supply it post-import today anyway \u2014 documented as a separate follow-up (persisting rawGpx) rather than solved here.\n\n**Verification split, same pattern as every companion PR tonight**: :companion:core (the cache/codec logic) is host-tested and green. :companion:route (Room codegen, the migration actually running, UI wiring) is hand-reviewed only \u2014 no Android SDK in this sandbox.

robert referenced this issue from a commit

2026-09-04 14:33:48 +02:00

Fix two real compile errors found by the first actual Android build

robert referenced this issue from a commit

2026-09-05 23:29:57 +02:00

Cue-sheet review screen before riding (issue #55)

robert referenced this issue

2026-09-05 23:32:48 +02:00

Cue-sheet review on the phone before riding #55