2026-07 backlog sweep: RELAY v1.11, diagnostic/sporadic frames, CLI/Docker/doc fixes

[SoundMatt]

Full sweep of the 21 issues that were open at the start of this pass (2026-07-27), triaged and actioned as below.

Issues fixed here

RELAY spec conformance

• [P2] go.mod / CI pin RELAY v1.10.0, one minor behind the current spec (v1.11); relay versions reports ALIGNED=false #42 — bump github.com/SoundMatt/RELAY to v1.11.0 (go.mod + CI relay@ pins); relay versions now reports ALIGNED=true
• [P2] go-lin's optional send/subscribe CLI commands don't implement the section 11.2 flag signatures #41 — send/subscribe now implement the spec §11.2 flag signatures (send --id <uint> --data <hex>, subscribe --format json --count N)
• [P1] Docker images don't ship the RELAY-conformant go-lin CLI and lack required io.relay.* labels (spec section 13.5) #39 — Docker: new go-lin image target with the spec §13.5 io.relay.* labels, published as ghcr.io/soundmatt/go-lin
• [P2] README documents the legacy non-conformant lintool CLI and omits cmd/go-lin, the actual RELAY-conformant binary #40 — README documents cmd/go-lin (the RELAY-conformant CLI) as primary; cmd/lintool marked legacy

Bug fixes

• [P1] virtual.Bus.CloseWithDrain never returns ErrTimeout on context expiry, and a test asserts this wrong behavior as correct #37 — virtual.Bus.CloseWithDrain now returns lin.ErrTimeout (and counts undelivered frames into DropCount) on ctx expiry instead of silently returning nil; the test that had certified the wrong behavior as correct now asserts the right one
• [P1] lin.FromMessage does not wrap ErrInvalidFrame, breaking errors.Is for out-of-range message IDs #38 — lin.FromMessage now wraps ErrInvalidFrame so errors.Is works for out-of-range/malformed message IDs
• [P2] master.Node.SetSchedule rejects an empty schedule, contradicting the documented MasterBus.SetSchedule contract #43 — master.Node.SetSchedule now accepts an empty schedule, matching the documented lin.MasterBus.SetSchedule contract (this was also a formal ASIL-B requirement, REQ-MASTER-010 — updated its description/rationale rather than just the code, since Run independently guards the actual busy-loop hazard via REQ-MASTER-009)

New capabilities

• feat: diagnostic frames — master request (0x3C) and slave response (0x3D) #2 — diagnostic frames: lin.MasterRequestFrame/lin.SlaveResponseFrame (LIN 2.x §4.2.3 single-frame wire encoding) + master.Node.Diagnostics
• feat: sporadic frames — master selects frame based on flags #4 — sporadic frames: master.Node.SetSporadicGroup/SetPending (LIN 2.x §2.3.2.4 priority-ordered candidate selection, correct slot-skip when nothing pending)
• feat: LDF signal encoding (write direction) #6 — ldf.DB.Encode: write-direction signal encoding (inverse of Decode)
• feat: statistics — bus load, frame counters, per-ID metrics #10 — new stats package: frames/sec, per-ID counters, estimated bus load
• chore: transport/ and coverage ≥ 80% #9 — go-FuSa CLI bumped v0.30.0 → v0.33.3 (verified all CI gate commands work against the repo before bumping); coverage part of this issue is already-resolved (see below)
• docs: add pkg.go.dev examples and godoc improvements #11 — runnable Example* functions across lin/virtual/master/ldf/stats

Docs

• [P2] ROADMAP.md is stale -- describes v0.1.0 as 'next' while the repo has already shipped through v1.2.0; no CHANGELOG.md exists #44 — ROADMAP.md replaced the stale pre-RELAY plan with actual v0.1.0–v1.2.0 history; added CHANGELOG.md

All new safety-relevant code carries //fusa:req///fusa:test annotations and corresponding .fusa-reqs.json entries (13 new requirements: REQ-MASTER-014/015/016, REQ-STATS-001..008); gofusa trace -req-coverage 100 and -sec-tested 100 both pass at 100%.

Closed as already-resolved (not part of this PR's commit — closed directly with evidence)

• safety: FMEA review and ASIL-B argument completion #12 (FMEA review/ASIL-B completion) — gofusa trace already reported 100% requirement traceability and 100% test coverage on main before this pass, and has been continuously CI-gated since v1.1.0 (§20 continuous conformance), superseding the original one-time-review ask. See issue comment for full evidence.

Left open (out of scope for this pass — comment posted on each explaining why)

• RELAY spec v0.2 — please review and raise feedback on the RELAY repo #19 — standing invitation for RELAY spec feedback; not a code change here by design
• feat: serial/UART transport layer (transport/) #1 (serial/UART transport) — hardware-facing; needs real-hardware/HIL validation before landing in an ASIL-B library
• feat: sleep/wakeup frame sequences #3 (sleep/wakeup) — wakeup pulse gen/detection is a physical-layer concern blocked on feat: serial/UART transport layer (transport/) #1; the go-to-sleep command frame itself is tractable as a small follow-up
• feat: event-triggered frames and collision resolution #5 (event-triggered frames + collision resolution) — multi-slave collision semantics need a dedicated design/test-vector pass, not a backlog-sweep addition
• feat: bridge/can — LIN-over-CAN gateway #7 (bridge/can), feat: bridge/mqtt — LIN frames over MQTT #8 (bridge/mqtt) — both need a new external dependency + a design decision (ID-mapping / topic-and-encoding scheme) that shouldn't be guessed at in a conformance sweep

Verification

• gofmt -l . — clean
• go vet ./... — clean
• go test ./... -race -count=1 — all pass (221 tests)
• Per-package coverage: all library packages ≥ 89% (CI floor is 85%)
• go test -fuzz=... -fuzztime=2000x for all three fuzz targets — clean
• relay conform --strict / relay interop --protocol LIN against the built cmd/go-lin binary with RELAY v1.11.0 — both PASS
• Full gofusa gate suite (check, trace -req-coverage 100, trace -sec-tested 100, cyber, vuln, qualify, verify) — all green with go-FuSa v0.33.3