Overrides the legacy 55ms analog poll interval (e.g. --poll 20 at
31250 -> ~50Hz analog). Summary's expected-poll-slots math follows.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Bench falsified the v1 retransmit (testlogs/riomash-patched-62500-retx):
resends tracked NAKs 1:1 (321/321) yet lamps still stuck/missed — under
mash bursts the NAK arrives after a newer command is already "latest",
so the wrong packet was resent; and total shreds never NAK at all.
Now commands are stop-and-wait: ONE in flight (_commandGate), resolved
by ACK, NAK, or AckTimeout (50ms); NAK/timeout retransmits THE SAME
packet up to CommandRetransmitLimit (2), then drops (idempotent - the
next state update supersedes). Attribution is exact by construction and
timeouts catch silent shreds. Control-byte replies bypass the gate so
board traffic is never delayed; a request's own reply (analog/version/
check) also resolves the wait, so request/reply exchanges never burn
the timeout even if the board sends no explicit ACK.
7 tests (same-packet resend, timeout retry+drop, ACK completion,
serialization, reply-resolves-request, stray-NAK no-op, disable);
282 green. Mash summary now reports NAK/timeout resends.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The board ACK/NAKs every inbound packet; we were fire-and-forget, so
any corrupt arrival became a permanent state error (stuck-bright /
missed lamps at 62500, where ~6.5% of 4-byte lamp commands lose bytes
to RX-ISR overrun). Now a NAK control byte triggers a resend of the
most recent command packet, bounded by NakRetransmitLimit (default 2;
0 restores fire-and-forget).
Design notes: the wire has no sequence numbers, but every PC->RIO
command is idempotent (lamp state, analog request, reset), so resending
the latest command is safe even in the rare race where the NAK belonged
to an earlier packet. Single control-byte replies (our ACKs) never
participate. NakRetransmits counter surfaced in the mash summary.
6 new tests (retransmit, limit, budget reset, ACK no-op, control-byte
exclusion, disable switch); 281 green.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
127s mash at 31250: zero wedges — but 5655 framing resyncs, duplicate
replies (215% of poll slots) and AbandonCount=65 reveal the board's
reply ACK-wait scales with byte time: ~5+ms at 9600 (USB beats it,
zero framing) vs ~1.6ms at 31250 (USB cannot). All 65 retry
exhaustions self-recovered through the patched give-up path — the
strongest field proof of the latch fix yet; unpatched firmware would
have wedged on the first. Verdict recorded in ANALYSIS.md: cabinets
run the 9600 patched chip; 31250 stays a bench branch until the retry
window is widened in firmware.
Tool: before-snapshot now retries once (can lose the post-DTR boot
race, as this run showed); evidence log archived.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Stopping early is a real operator workflow (finger fatigue); previously
an interrupt died summary-less and could orphan a process holding the
COM port. CancelKeyPress now breaks the run loop, snapshots the after-
counters, and prints the summary with the actual elapsed duration.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
On a wedged (fully silent) board the receive loop sits in a pending
net48 serial read that ignores cancellation; awaiting the link before
closing the port hangs the tool at run end and loses the summary -
exactly what happened on the first real baseline run (the board wedged
at 0.62s and stayed dead, so no byte ever completed the read). Close
the port first (same order RioCoordinator.Teardown uses); applies to
both monitor and --mash modes. Selftest unaffected (its fake transport
honors cancellation, which is why it never caught this).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
make_patch.py gains --baud31250: one byte beyond the wedge patch — the SCI
init operand at $D62B ($30 -> $02). BAUD $30 = /13 prescale, /1 divider
(2MHz E / 208 = 9615); $02 = /1, /4 -> 31250 exactly, 3.3x faster. The
init write also confirms the 8MHz crystal, which is why 19200/38400 are
unreachable and why 62500/125k are left alone pending an ISR cycle count.
- 24 bytes changed vs original (sha256 9f866cf3...); re-disassembly diff
vs the classic patched image shows exactly the one operand line, and
the classic build still reproduces 3fc8170c... (script regression-safe).
- PC side: SerialPortTransport takes an optional baudRate (default 9600,
runtime untouched); RioSerialMonitor + --mash accept --baud 31250.
- Caveats documented in README/ANALYSIS: non-standard rate (FTDI-class
adapters only, no 16550s); native games still speak 9600 so this chip
is bench/RIOJoy-only; validate the wedge patch at 9600 first.
275 tests green; mash --selftest regression unchanged (FAIL/exit-1).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
New --mash mode (tools/RioSerialMonitor/MashTest.cs) mechanizes the
wedge-patch validation plan from RIOv4_2-ANALYSIS.md:
- Runs the live link with the app's >5s reset-recovery DISABLED so a
board wedge stays observable, and echoes lamps on every press
(lamp/reply collisions are the wedge trigger).
- Gap timing uses ANY AnalogReply packet (0xFE sentinels included -
a sentinel still proves the reply path is alive); logs a gap
histogram + top-10 longest gaps with timestamps.
- WEDGE detector: analog silent past the threshold (default 2s) ->
beep + banner; on resume, classifies self-recovered (patched
expectation) vs button-revived (button event within 300ms of resume,
the unpatched signature) vs unresolved at run end.
- Board self-reported RestartCount/AbandonCount/FullBufferCount
snapshotted before/after via CheckRequest, delta printed
(7-bit wrap-aware).
- Fixed-layout summary teed to riomash-<label>-<stamp>.log so
baseline-vs-patched runs diff directly. Exit 0 = no wedge, 1 = wedge.
--mash --selftest drives the whole instrument against a scripted
in-memory board (SelftestTransport) that goes silent at t=4.0s and
revives 200ms after a button at t=6.5s: verified end-to-end - alarm at
6.0s, wedge classified button-revived (2.75s), counter delta +4/+0/+1,
verdict FAIL, exit 1. Use it to sanity-check the alarm at the cabinet.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Prepares RioJoy.Core for the net40 (Windows XP) target, which has no
System.Memory, ValueTask, or System.Text.Json:
- IRioTransport and the whole protocol/framing layer now use byte[] +
Task (RioPacket.Payload, PacketParser/Builder, RioChecksum, replies,
AnalogReport, RioHidReport). At 9600 baud Span bought nothing; the
SerialPortTransport bridge copies disappear entirely.
- ConfigStore/OverlayTemplateStore switch to Newtonsoft 13 with the
same conventions (indented, PascalCase, string enums, null-skipping);
verified against the real STJ-written config.json and regions.json
(load + round-trip). System.Memory and System.Text.Json packages
dropped.
275 tests green.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Documentation now describes the .NET Framework 4.8 stack (was .NET 8):
- README build prerequisites + framework-dependent/no-runtime-install note,
test count 136 -> 241.
- PLAN.md architecture diagram, stack decision (with PolySharp/shims note),
suite total 136 -> 241.
- deploy/README-DEPLOY.txt: app is net48 framework-dependent (4.8 is in-box on
Win10/11), and build prerequisites.
Also migrate the three tools (RioJoySmokeTest, RioSerialMonitor, XcfRegionExtract)
to net48 so they keep building against the now-net48 RioJoy.Core (a net8 project
cannot reference a net48 one). Added PolySharp + System.Memory/System.Text.Json
shims and replaced net-core-only APIs (string.Contains/IndexOf with comparison,
Array.Fill, generic Enum.IsDefined, int.TryParse(span)). All three build clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A console diagnostic that opens the RIO COM port, runs the real serial link,
and logs every event (buttons, keypad, axis, version/check, control bytes,
framing). Flashes all lamps once to prove the PC->RIO path and echoes a lamp
on each button press. Verified end-to-end against a partial RIO on COM1
(firmware 4.2; MFD + keypad inputs and the throttle axis all decode).
dotnet run --project tools/RioSerialMonitor -- COM1 30
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Overlay generator (pure, tested) — RioJoy.Core/Overlay:
- FontFitter ports the legacy calc-fontsize auto-fit search (validated against a
brute-force oracle); OverlayLayoutEngine ports create-data-layer's fit + h/v
justification and adds per-region 90 deg CCW rotation; OverlayTemplate/Region +
OverlayTemplateStore hold cell geometry/colour/rotation (regions.json).
- GoobieDataImporter parses the legacy .data label sheet into label rows.
- src/RioJoy.Overlay: SkiaSharp rasterizer (SkiaTextMeasurer shared with the engine
so measured layout == drawn output; SkiaOverlayRenderer -> PNG;
ProfileWallpaperGenerator). Verified end-to-end on the real cockpit art
(regions.json + riojoy.png + TEST.data) by OverlayRenderIntegrationTests.
- RioJoy.Tray/WallpaperApplier applies via SystemParametersInfo; RioCoordinator
generates+applies on profile activation (opt-in via AppConfig.OverlayTemplatePath).
Region geometry — tools/XcfRegionExtract parses riojoy.xcf (a GIMP-format binary
reader, no GIMP needed) into the 119-cell regions.json: per-layer offsets/size/
font/colour, with 90 deg CCW rotation on a-00 + b-10..b-1F. Base image riojoy.png.
NOTE: the wallpaper positions are a VGA chroma-split display artifact (6 displays),
not the cockpit's logical layout.
Mapping editor (first cut) — RioJoy.Tray/Editor/ProfileEditorForm renders the
config sheet's logical grid (SheetLayout parses the sheet CSV export). The iRIO
word is edited via ButtonBinding <-> RioMapEntry.Create (no hex bit-twiddling), with
a context-sensitive picker: keyboard keys by name (KeyCatalog VK names), joystick
Button N, hat direction, mouse/RIO-command enum names; modifiers only for keyboard.
Opened from the tray ("Edit profile..."). The sheet-grid layout is a starting point
that needs rework from a better-formatted source.
~220 xUnit tests green. Docs (PLAN.md, README) updated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The RioGamepad driver now installs and enumerates, so the user-mode
feeder path is verifiable. Add tools/RioJoySmokeTest, a standalone
on-cabinet utility that drives the real HidFeederJoystickSink (open the
device, submit reports via IOCTL_RIO_SUBMIT_REPORT) and reads the gamepad
back through winmm joyGetPosEx, asserting axes (min/mid/max), buttons,
and the POV hat all surface to the OS. Verified: all checks pass against
the installed driver.
Update docs to match reality (PLAN.md predated the HidFeederJoystickSink
commit): Phase 1 is now test-signed/installed/verified with the VHF
LowerFilters requirement noted; the stale "NullJoystickSink placeholder"
remainders in Phases 3 and 5 are corrected to reflect the wired,
verified feeder. driver/README.md notes the end-to-end verification.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
CydandClaude Fable 5