Nikolay Miroshnichenko 07352cdb61 suggest: split multi-output channels by vendor object marker; drop device diagnostics ...

- _subunit()/_split_subunits(): '[1] Switch On/Off' style markers separate the outputs a vendor
  packs into one ETS channel; each sub-unit is classified on its own (ids and secondary_info carry it)
- diagnostics (error/communication/firmware/version/bus voltage/reset, EN+DE+RU) never become sensors
  and are excluded from the name-based fallback; counted in hints.diagnostics_skipped
- real-project effect (3646 GAs): entities 341->513, sensors 752->367, platform agreement 92->95 %,
  address-key agreement 94->95 %
- tests: multi-output split (two lights, statuses consumed, no sensor fallout) + diagnostics filter

2026-09-12 09:09:08 +02:00

..

test_appprog_parser.py

feat: exact device models — local catalog for decompose_device + parse_devices_from_project (v0.7.0)

2026-07-02 06:48:19 +02:00

test_council_fixes.py

fix: council-review noise cuts — scenes/central-macros exempt, language-matched repairs

2026-07-15 09:09:57 +02:00

test_cover_pairing.py

harden(#11): operation!=domain gate, deterministic tiebreak, no central-stop steal (LLM-council P0 + gate-1)

2026-08-29 15:01:25 +02:00

test_device_catalog.py

feat: exact device models — local catalog for decompose_device + parse_devices_from_project (v0.7.0)

2026-07-02 06:48:19 +02:00

test_domain_classifier.py

fix(P5): dogfood round 2 — presence is a sensor, 9.001 soft, passive-range rule

2026-07-15 11:33:20 +02:00

test_explain.py

feat(P5): contextual domain classification — DPT is one signal, not the verdict

2026-07-15 10:44:04 +02:00

test_explainable_aggregates.py

feat(P7b): explainable aggregate scores — expose denominator + formula

2026-07-15 10:15:40 +02:00

test_param_check.py

feat(param_check): significance layer — a focus list of config-value outliers

2026-07-14 22:28:32 +02:00

test_pipeline.py

feat: matter/energy/diff/iot/completeness + climate & expose (v0.6.0)

2026-07-01 20:31:14 +02:00

test_policy.py

check_policy: example profile seeded from the loaded project's own main groups (issue #13)

2026-09-09 20:03:32 +02:00

test_rm_status.py

fix(#12 residual): recognise a shutter Function by its 1.008 move, not only a role token

2026-09-04 21:33:25 +02:00

test_role_completeness.py

feat(P2): role-aware feedback completeness — catch a value command with no value status

2026-07-15 09:23:02 +02:00

test_room_library.py

room-library: Theben-benchmark corrections (dogfood on 5 official ETS examples)

2026-07-16 11:21:18 +02:00

test_root_cause.py

fix(P3): root-cause suppression — an empty-name GA no longer cascades

2026-07-15 09:29:54 +02:00

test_safexml.py

feat(P7a): harden parsing of untrusted .knxproj (zip-bomb / XXE defense)

2026-07-15 10:11:38 +02:00

test_suggest.py

suggest: split multi-output channels by vendor object marker; drop device diagnostics

2026-09-12 09:09:08 +02:00

test_theben_corrections.py

room-library: Theben-benchmark corrections (dogfood on 5 official ETS examples)

2026-07-16 11:21:18 +02:00

test_topology.py

feat: check_topology — per-line device count + individual-address validity (KNX canon)

2026-07-24 21:41:14 +02:00

test_v08_fixes.py

feat: positional status pairing + self-reporting commands, app-program parser v2 (ComObjectRef merge, version pick), authoritative main names, grader range-name evidence (v0.8.0)

2026-07-07 12:24:24 +02:00