- Vendor maplibre-gl 4.7.1 (js+css) and serve from /vendor instead of the unpkg CDN — no external dependency/SRI gap for the core map.
- Projector skips duplicate (imei, occurred_at) history rows via NOT EXISTS (parked devices re-report the same gpsTime each poll); migration 23 dedupes existing rows and adds a unique index.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Migration 20: collapse `Nairobi`/`nairobi` in domain.vehicles → 'nairobi'
- Remove the SLO panel from the top band (filter + tile rows stay)
- Offline vehicles render as solid grey instead of dim-cost-centre tint;
opacity now only differentiates moving (1.0) vs parked (0.75) vs
offline (0.55) so colour carries identity + state cleanly
PG function overloading is type-strict: integer→bigint isn't implicit. The
prior signature (integer, date) didn't match callers passing vehicle_id
directly. Also prepends a DROP for the integer-signature in case the
previous migration ran and left a now-orphan function around.
Calibrated against a full-day legacy report for vehicle KDE 638J:
without this rule we collapsed 15 dispatcher-visible trips into 3
(the algorithm waited for explicit ACC_OFF + stationary fixes that never
came in the polled data — the device just stops reporting between trips).
New rule: if mid-trip and the next fix arrives >=5 min after the previous
one (but <=30 min, which is still long_gap), close the trip at the prior
fix with end_reason='nofix_stop'.
Validation:
638J full day: legacy 15 trips, mine 15 trips, perfect alignment
728K half day: legacy 33 noisy trips, mine 9 clean trips
(stop-and-go traffic still consolidates because fixes
keep coming, just slow/stationary — not silence)
Also commits scripts/simulate_trips_from_legacy.py which runs the same
state machine in Python against REPORTS/*.json dumps for offline tuning.
Migration 19: serve.fn_vehicle_trips(vehicle_id, date) — PL/pgSQL state
machine that walks state.position_history for one vehicle on one EAT day
and emits the trip breakdown. Rules:
- reporting_time = first ACC_ON of the day
- trip starts at ACC_ON (or first fix if already ACC_ON / moving)
- trip ends:
* ACC_OFF + stationary (<5 km/h) for >=5 min → end_reason 'work_stop'
* fix gap >30 min → end_reason 'long_gap'
* end of day's data → end_reason 'day_end'
- within a trip, ACC_ON + stationary >=5 min is logged as an idling
segment (no trip split — engine still on)
- distance only accumulates when speed >= 5 km/h (excludes GPS jitter)
- falls back to movement-only segmentation when acc_state is null
across the day (has_acc_data=false in the response)
Returns one jsonb document: vehicle, date, reporting_time, day totals
(distance, driving/idling/stopped/unknown minutes), data_quality flags,
trips[] with start/end/duration/distance/idling/end_reason/stops/path
where path is a GeoJSON LineString ready for the map.
New endpoints (read:fleet, rate-limited):
GET /api/views/vehicle/{id}/trips?date=YYYY-MM-DD JSON
GET /api/views/vehicle/{id}/trips.csv?date=YYYY-MM-DD one row per trip
Defaults date to today in EAT (UTC+3) regardless of host TZ.
Migration 18: ops.contract_check_log table — append-only log of probes
against the Tracksolid Pro endpoints we depend on.
New worker app/workers/contract_check.py — per run:
- jimi.oauth.token.get (token refresh succeeds)
- jimi.user.device.location.list per configured target (parse first item
with JimiPollFix)
- jimi.device.location.get with a sample IMEI from the list (parse first
item with JimiPollFix)
Each probe logs success or {error_class, error_detail, sample}; failures
are recorded, not raised.
slo_metrics now also computes contract_drift_days = days since the most-
recent successful probe of the laggard endpoint. With threshold 1d (from
mig 5), a single failed daily run flips the badge red within 24h.
cron entrypoint registers the check daily at 02:00 UTC plus once on
startup, gated on TRACKSOLID_APP_KEY + a configured target.
The CSV-based roster import (mig 15+16 and scripts/import_csv_roster.py)
merged vehicle rows that differed only by _Track / _CAM suffix, dropping
the active fleet count from 144 to 124. Reverting the whole thing.
Mig 17 in one transaction:
- Re-splits devices by parsed plate from device_name (same regex as
mig 14, preserving _Track as separate vehicle)
- Restores serve.fn_live_view to its v3 body (no d.driver_name/phone
refs that would break once the columns are gone)
- Drops the six CSV-only columns from domain.devices
- Deletes schema_migrations rows for the deleted 15/16
- Logs final counts via RAISE NOTICE
Apply on VPS: psql -f db/migrations/20260601000017_rollback_csv_import.sql
