Commit graph

7 commits

Author SHA1 Message Date
680ffa8cf2
test: silence Postgrex sandbox disconnect noise
When a test process exits while still owning a sandboxed Postgrex
connection, Postgrex.Protocol logs `[error] ... disconnected:
DBConnection.ConnectionError client #PID<...> exited`. It's benign
sandbox cleanup but at :error level it bypasses `capture_log: true`
(which only captures the test process's logs) and clutters every
test run.

Pin Postgrex.Protocol + DBConnection.Connection to :critical so the
noise drops without masking genuine DB error reporting (we never log
at :critical from those modules).
2026-05-08 10:43:51 -05:00
363646f34f
test: enable :slow tests by default + add coverage tests across modules
Removed the `exclude: [:slow]` from test_helper.exs — the slow-tagged
wgrib2 fixture tests pass with the binary on PATH and pull
Microwaveprop.Weather.Grib2.Wgrib2 from 23% to 90%+ in coverage runs.

Additional tests:
- Pskr.Client: tcp/tcp_closed/tcp_error stale-socket fall-throughs
- Weather context: weather_point_detail/3, latest_grid_valid_time/0,
  available_weather_valid_times/0, available_hrdps_valid_times/0
- PathCompute.compute/4: with HRRR profile near the path
- RoverPathProfileWorker: real Path row + skip-when-complete
- RoverLive: URL stations encoding, add_station error paths
- PathLive: rover_path_id branches (404, malformed UUID)

3599 tests, 0 failures.
2026-05-08 09:28:51 -05:00
89e8bb2267
test: use Mox for Valkey tests without real Redis (55%→88%)
- Create Microwaveprop.Valkey.Adapter behaviour with command/3 + pipeline/3
- Create Microwaveprop.Valkey.RedixAdapter as the production impl
- Inject adapter via Application config; mock with Mox in tests
- Cover all Valkey operations: get, mget, set, mset_with_ttl, zadd,
  zrevrange, zrangebyscore, zrem, del, scan_match (multi-cursor)
- Full round-trip tests for encode/decode, error paths, edge cases

Coverage: 78.93% → 79.07%
2026-05-07 13:48:08 -05:00
65e97bec6b
feat(aprs): add Aprs context for read-only queries against aprs.me 2026-05-01 12:47:35 -05:00
b2b8ddc1c4
test: silence expected warnings/errors during test runs
Passes `capture_log: true` to ExUnit.start/1 so log output emitted
during a passing test stays in the per-test capture buffer instead of
stdout. Drops ~360 lines of noise from the suite: Postgrex teardown
disconnects, expected worker failures, NexradClient 404 stubs,
NotifyListener warm-skip messages, SNMP poll failures, PromEx tag-
drop warnings, and OTel-handler boot errors. Failing tests still
surface their full logs on report, so debugging is unchanged.

Side-effect fixes:
- LoggerFormatTest opts out via @moduletag capture_log: false — it
  patches the :default :logger handler and capture_log hot-swaps the
  same handler, so the setup lookup would 404.
- TelemetryTest.start_link/1 unlinks the supervisor before the brutal
  kill so the exit signal doesn't propagate to the test process.
- Drop unused `import Ecto.Query` in a top_hours mix-task test.
- Drop unused default-arg `attrs \\ %{}` on create_contact/1 in
  contact_map_live_test (every caller passes attrs explicitly).

Remaining noise is ~3 lines from oban_pro vendored-dep @impl warnings
(require an upstream patch) plus rare intermittent async teardowns.
2026-04-24 09:27:17 -05:00
8f89bbdc02
Exclude slow GRIB2 fixture tests by default (81s -> 16s) 2026-03-31 10:00:13 -05:00
bc674f0d54
initial 2026-03-28 11:28:47 -05:00