221 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
sjg	0b1fb005f6	[fix](trx-rs): install binaries under user home ... Default installation and removal to /home/sjg/.local/bin while preserving explicit prefix and binary-directory overrides. Update the service examples and README to match. Assisted-By: OpenAI Codex (GPT-5) Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-18 23:08:08 +02:00
sjg	be9d5c301b	[docs](trx-rs): add safe deployment guide ... Document deployment with a dedicated service account, restricted device access, authenticated network listeners, systemd user services, reverse proxying, verification, upgrades, and rollback. Assisted-By: OpenAI Codex (GPT-5) Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-18 22:53:00 +02:00
sjg	b73c97dd5b	[fix](trx-server): honor per-rig audio bind addresses ... Use each rig's configured audio listener unless --listen explicitly overrides all bind addresses. Keep preflight socket validation consistent with runtime behavior. Assisted-By: OpenAI Codex (GPT-5) Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-18 22:45:33 +02:00
sjg	6276c3feea	[fix](trx-rs): allow systemd hardware discovery ... Do not restrict socket families in the generic user units because SoapySDR and libusb require netlink sockets to enumerate radio hardware. Assisted-By: OpenAI Codex (GPT-5) Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-18 22:44:58 +02:00
sjg	110c0e1d49	[fix](trx-rs): correct systemd network ordering ... Let the client wait for an enabled local server without activating a disabled one, and declare the socket families required by both services for configured listeners and remote connections. Assisted-By: OpenAI Codex (GPT-5) Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-18 22:35:54 +02:00
sjg	05c337b581	[feat](trx-rs): add build/install script and systemd user services ... Provide a one-shot installer and matching systemd *user* units so the server and client can be built, installed system-wide, and run in the background without hand-rolled steps. - script/install.sh: builds all three binaries (trx-server, trx-client, trx-configurator) in release mode and installs them to /usr/local/bin (configurable via --prefix/--bindir/PREFIX, sudo only when the target is not writable). Seeds ~/.config/trx-rs/trx-rs.toml from the example without ever overwriting existing config, then installs the user units with @BINDIR@ substituted to the real path and runs daemon-reload. Flags: --no-sdr, --no-build, --no-systemd, --enable-now. - script/uninstall.sh: stops/disables the units, removes them and the binaries; keeps config unless --purge. - packaging/systemd/{trx-server,trx-client}.service: user units reading the combined config at ~/.config/trx-rs/trx-rs.toml. The client softly depends on the server (Wants/After). KillSignal=SIGINT matches how the binaries shut down cleanly (SIGTERM is not handled). - README: new "Install (optional, Linux + systemd)" section. The web UI assets are embedded in the binary, so an install needs only the binaries plus a config file — no data directory to ship. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-16 23:01:45 +02:00
sjg	c10b5faef4	[feat](trx-rs): redesign SDR noise blanker with tuning profiles ... The old IQ noise blanker tracked a fast running RMS and, on a threshold crossing, replaced the sample with the last clean one. That hard sample-and-hold is a step discontinuity: it splatters energy back across the wideband passband, so after the narrow channel filter it often sounded worse than the noise it removed — especially on SSB, CW and digital. It also had no look-ahead (the impulse leading edge leaked through before the fast RMS reacted), blanked only single samples, and used a fixed 1/128 time constant that did not scale with capture rate. Redesign the blanker around accepted wideband-NB practice and add profiles matched to the interference source: - Noise-floor tracker updated only from clean samples and frozen while blanking, so a burst cannot desensitise detection. - Detection on instantaneous power vs threshold² × noise floor. - Look-ahead delay line so the gate closes *before* the impulse reaches the output, removing the leading edge. - Raised-cosine tapered gate (ramp 1→0→1) instead of a hard hold, which minimises blanker splatter. - Windowed blanking that re-arms on every detected sample to cover the full width of a burst. - All timings expressed in real time and converted to samples at the capture rate, so behaviour is consistent across SDR sample rates. Profiles (`NoiseBlankerProfile`, default `spike`): spike, ignition, powerline, broadband — each selects the blank window, look-ahead, taper and floor time constant. `threshold` stays orthogonal as the sensitivity knob. Core/protocol: - New `NoiseBlankerProfile` in trx-core (serde/parse/u8/TS), re-exported at the crate root; `RigFilterState.sdr_nb_profile` for state sync. - `profile` added to `RigCommand`/`ClientCommand::SetSdrNoiseBlanker`, the trait method, and the command mapping. Config: `[rig.sdr.noise_blanker] profile = "spike"` (regenerated trx-rs.toml.example). SDR backend: `NoiseBlanker` rewritten in the channel DSP; profile wired through `SoapySdrConfig`, the runtime setter, and `filter_state()`. Frontend: an "NB profile" selector in the SDR advanced controls (POST /set_sdr_noise_blanker&profile=…), reflecting server state; the profile rides along with the enable/threshold quick toggle so it is preserved. Regenerated generated.ts and app.js. Tests: profile u8/parse round-trips (trx-core); DSP tests for impulse suppression with no leading-edge leak, strong-steady-signal pass-through, and wider-profile-blanks-longer. Docs: User-Manual NB section rewritten for the new algorithm and profiles. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-16 21:42:14 +02:00
sjgandClaude Opus 4.8	79adc8d5c6	[chore](trx-rs): run CI on self-hosted runner ... Switch every CI job from ubuntu-latest to the self-hosted runner. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UiK871ht2uPFBHtMbxy3wD Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-16 13:22:44 +02:00
sjgandClaude Opus 4.8	c33e3caedb	[style](trx-rs): cargo fmt ... Apply rustfmt to the wefax and DIG-sideband changes. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UiK871ht2uPFBHtMbxy3wD Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-16 12:50:12 +02:00
sjgandClaude Opus 4.8	5084c19899	[fix](trx-frontend-http): move passband overlay to resolved DIG sideband ... The spectrum passband overlay is drawn one-sided per mode, but DIG was hardcoded to upper sideband — so switching the DIG sideband (or tuning DIG/Auto across 10 MHz) left the overlay on the wrong side of the carrier even though the backend had flipped the demodulator. Resolve the DIG overlay direction from the current sideband policy and dial frequency, mirroring the backend: usb → upper, lsb → lower, auto → upper at or above 10 MHz and lower below. The overlay repaints immediately on a policy change (optimistically on the selector, and on confirmed filter state) and tracks frequency as it already did. Non-SDR backends keep the historical upper-sideband overlay. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UiK871ht2uPFBHtMbxy3wD Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-16 10:24:30 +02:00
sjgandClaude Opus 4.8	df7483fe30	[feat](trx-rs): configurable DIG sideband for SDR (auto/USB/LSB) ... DIG has no inherent sideband; on the SDR backend it was always demodulated as USB. Make it resolve to USB or LSB via a policy that defaults to the amateur SSB/data convention (USB at/above 10 MHz, LSB below) and can be overridden globally from the advanced radio controls or per-bookmark. Design: the logical DIG mode is kept in RigState (display, decoder gating) while the SDR pipeline is handed a concrete USB/LSB demodulator resolved from (policy, dial frequency). Resolution happens at the boundary — the rig for the primary channel and the virtual-channel manager for vchans — so the hot DSP/demod path is untouched. The resolved sideband is only re-pushed when it actually changes (e.g. tuning DIG/Auto across 10 MHz), keeping ordinary tuning glitch-free. Core/protocol: - New `DigSidebandPolicy { Auto, Usb, Lsb }` with `resolve(freq)` and an `effective_demod_mode()` helper (trx-core), re-exported at the crate root. - `RigCommand::SetSdrDigSideband`, `RigSdr::set_sdr_dig_sideband`, and a `RigFilterState.sdr_dig_sideband` field for state sync; wired through the ClientCommand mapping. Config: `[rig.sdr] dig_sideband = "auto"` (regenerated trx-rs.toml.example). SDR backend: the vchan manager owns the shared policy (atomic); the rig applies it to the primary channel and, on `set_sdr_dig_sideband`, re-resolves all DIG virtual channels. Frontend: a mode-gated "DIG sideband" selector in the SDR advanced controls (POST /set_sdr_dig_sideband), reflecting server state; bookmarks gain an optional `dig_sideband` field (form selector shown only for DIG) that, on apply, sets the global policy before switching to DIG. The scheduler honours it for automated bookmark activation too. Tests: policy resolution / effective-mode / u8+parse round-trips (trx-core); a vchan integration test asserting a DIG channel resolves to LSB below 10 MHz, flips with the policy, and still lists as DIG. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UiK871ht2uPFBHtMbxy3wD Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-16 00:26:45 +02:00
sjgandClaude Opus 4.8	fe3b414fba	[fix](trx-wefax): stop fragmenting one transmission into many images ... The WEFAX decoder decoded content but chopped a single chart into many short PNG "chunks". Two heuristics fought each other: in State::Receiving the carrier-loss watchdog finalized the image after only 30 low-correlation scan lines (~15 s at 120 LPM), and the Idle variance auto-start then re-triggered on the still-present carrier ~3 s later, starting a fresh image. Ordinary HF fading (QSB) of 5-20 s therefore split every chart into a stream of tiny images. Reference decoders (fldigi) keep one continuous image per APT cycle up to a large line cap and only stop on the APT stop tone or genuine signal loss. Align with that model: - Raise the end-of-transmission watchdog to 120 low-correlation lines (~60 s at 120 LPM) so normal fades ride through within one image. - Gate the variance-based auto-start to fire at most once per session (auto_start_used). After the first image a new one starts only on an APT start tone or an explicit reset, so trailing carrier/noise can no longer spawn a second image. reset() re-arms it. - Cap a single image at 3000 lines to bound memory on an open carrier. - Require a 2 s (was 1 s) APT tone sustain, cutting false Stop detections on busy image content that momentarily hits ~450 transitions/s. Also make line slicing drift-free: 120 LPM at 11025 Hz is 5512.5 samples per line, and slicing on the rounded integer accumulated a fractional-sample error every line (slow horizontal slant). Boundaries are now derived from the exact fractional line length (samples_per_line_f64) so the error never accumulates. Adds regression tests for the single-auto-start invariant and zero slicer drift over 1000 lines. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UiK871ht2uPFBHtMbxy3wD Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-15 23:35:09 +02:00
sjg	1829e3d17b	[fix](trx-frontend-http): make Guest role composable	2026-08-13 02:16:59 +02:00
sjg	b037225a05	[fix](trx-frontend-http): block Guest password changes	2026-08-11 22:47:55 +02:00
sjg	e978cf8a84	[feat](trx-frontend-http): separate transmit permission ... Assisted-By: Codex (GPT-5) Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-11 18:36:33 +02:00
sjg	44870bc941	[feat](trx-frontend-http): add restricted Guest role ... Assisted-By: Codex (GPT-5) Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-11 16:04:51 +02:00
sjg	5e9dae02c7	Complete managed account lifecycle	2026-08-11 07:49:53 +02:00
sjg	34507ffa17	Add composable HTTP access roles	2026-08-11 01:06:57 +02:00
sjg	36c1e56efa	Protect the final administrator	2026-08-11 00:39:25 +02:00
sjg	e4cce9a004	Add Users settings tab	2026-08-11 00:37:59 +02:00
sjg	3c3fc69542	Refactor HTTP account system	2026-08-10 23:47:51 +02:00
sjgandClaude Opus 5	d539ff96e5	[docs](trx-rs): write down how a contest and a confirmation are worked ... The manual gains the contest exchange, the Cabrillo entry and the confirmations. The plan marks all five phases done, corrects the API paths to the /api/logbook they were built under -- /logbook is the page, and the bookmarks API already shadows its own page that way -- and records the two decisions phase 5 settled: that the Cabrillo header comes from the operator because no log can derive it, and that a confirmation counts from whichever bureau answered. The plan itself travels with this branch: it was written on a branch of its own that was never proposed for merge, and the earlier attempts to update it from the implementation branches were silent no-ops, because the edits did not assert that they had found what they were replacing. Refs #54 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 21:31:06 +02:00
sjgandClaude Opus 5	ca5cd65c85	[docs](trx-rs): settle the logbook's remaining questions ... The clock is the server's, as asked: it is the machine at the radio, where the browser may be a phone in another timezone with a clock nobody checked. When the two disagree by more than a second the panel says so, rather than logging a time the operator did not expect. The rest, decided against how logging is actually done: One station log, not one per rig. DXCC, WAS and LoTW count the callsign, not the radio, and a station worked on the second rig is still worked. The rig goes on the QSO as MY_RIG. Station location does follow the rig, though — these rigs can be in different places, so MY_GRIDSQUARE comes from the one that made the contact, which is what LoTW's station locations expect too. The operator is a per-QSO field set once per session. ADIF separates the callsign used on the air from the person at the key, and multi-operator stations rotate people through one station callsign. It defaults from the configured callsign, so a single operator never touches it. It cannot come from the session: the auth roles are control and rx, with nobody's name on them. The log file is configurable, defaulting to the user's data directory. Bookmarks sit in the config directory because they are settings and decode logs in the cache directory because they are disposable; a QSO log is irreplaceable, and cache directories get swept. Import collisions match on callsign, band, mode and a two-minute window. Loggers rarely agree to the second on the same QSO — one stamps the contact, the other the entry — so an exact-minute key duplicates half of what it is asked to merge. Two minutes absorbs that without swallowing a legitimate re-work, since contest rules forbid a second contact on the same band and mode. Times compare as instants so midnight matches, and modes are normalised or an imported SSB would miss our USB. That normalisation is now written down: a rig mode is not an ADIF mode. DIG is the one the rig cannot answer — a rig in DIG is in FT8 or FT4 depending on what is decoding — and WSPR never opens an entry at all, because hearing a beacon is not a contact. Refs #54 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 21:30:10 +02:00
sjgandClaude Opus 5	a9e1e86fdc	[docs](trx-rs): settle the logbook's panel, its prefill, and its formats ... Three answers from the issue, folded into the proposal. The logbook is a panel of its own rather than a strip on the radio page, and it stands in every layout: a log can be kept without adopting the ham layout, and read while another is selected. The ham layout is then the arrangement that starts from it, with the radio controls around it. Prefill is exactly six fields — frequency, mode, rig name, time, callsign and locator — and nothing else. A signal report in particular stays empty: an FT8 SNR is not what was sent, and prefilling one would put a number in the log that nobody exchanged. The station's own callsign and locator are not per-entry fields at all; they are station identity, shown once at the top of the panel and written into the QSO from configuration. The file format was left to me. ADIF stays, because it is not one option among several: LoTW, eQSL, Club Log, QRZ and every other logger read it and nothing else, so a log that cannot write it cannot be uploaded, confirmed or moved. Nothing on disk is ADI regardless — the store is JSON Lines. The second format is Cabrillo 3.0, which ADIF cannot replace: contest logs are submitted in it and rejected in anything else. It lands with the contest exchange fields, since without a serial or a zone it has nothing to write. Refs #54 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 21:30:10 +02:00
sjgandClaude Opus 5	76e33a91bb	[docs](trx-rs): propose the logbook and the ham radio layout ... Issue #54 asks for a ham radio layout and an ADIF logbook, with no detail behind either. This is what they would be: what the logbook has to hold, how it is stored, where ADIF is read and written, and which of the existing parts each half hangs off. Two things it settles before any code is written. The log is append-only JSON Lines rather than the whole-file dump the bookmarks use, because a station with forty thousand QSOs would rewrite megabytes to log one contact and lose the lot if the power went mid-dump. And a decode is not a QSO: the decoders only ever heard something, so a decoded callsign pre-fills an entry and never writes one. The layout is a fifth entry in the operator layouts that already exist, gated on the rig being able to transmit. Refs #54 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 21:30:10 +02:00
sjgandClaude Opus 5	ee185b98bc	[feat](trx-logbook): work a contest, and record what came back ... Phase 5 of the logbook: the exchange, the entry sponsors take, and the confirmations an award counts. Contest fields go on the contact — the contest, the serials both ways as numbers and as words, and the zones — because an exchange is not always a number: a zone, a section or a name goes in as written. The serial sent and the contest stay between contacts, since they belong to the session and not to the contact just logged, and the serial counts on by itself rather than being retyped forty times an hour. Cabrillo 3.0 is written because ADIF cannot do this job: sponsors take Cabrillo and reject everything else. Its shape is not ADIF's either — the frequency is kilohertz below 30 MHz and a band designator above it, the modes are CW, PH, FM, RY and DG, and the contacts go oldest first, as a contest log is read. The header cannot be derived from a log — how many operators, how much power, what the score is claimed to be — so it comes from the operator, with single-op, low power, all bands and mixed behind it. QSL, LoTW and eQSL states are held as ADIF's single letters, and anything else is refused rather than written: a log that grew states of its own would be one no other program could read. A contact is confirmed when any one of the three says so — an award wants a card or an electronic match, not one of each, and counting them separately would tell the operator they were short of what they have. The bands report counts contacts, distinct stations and confirmations per band, ordered by wavelength as a band plan reads. Closes #54 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 21:28:40 +02:00
sjgandClaude Opus 5	18b2d0efe6	[feat](trx-rs): keep a station log, and a layout to work the bands from ... The logbook of issue #54, in the shape the proposal settled on. A new crate, trx-logbook, holds the contact, the ADIF reader and writer, the file, and the rules for telling one contact from two. ADIF because it is the only thing the ecosystem reads: LoTW, eQSL, Club Log, QRZ and every other logger take it and nothing else, so a log that cannot write .adi cannot be uploaded, confirmed or moved. The reader is forgiving in the ways real files are irregular -- lowercase tags, CRLF, missing header, unknown fields, a declared length that is the only thing ending a value -- and carries what it does not model through to the export, so a round trip does not strip what another program wrote. The file is JSON Lines, appended one line per contact. A log is the one thing here that cannot be regenerated, and the bookmark store's whole-file dump would rewrite megabytes to log one contact and lose all of them if the power went halfway; an append costs the record being written and no more, which a test tears a line in half to prove. Edits append revisions, deletes append tombstones, and the file compacts when the superseded outnumber the live. The panel is its own tab and stands in every layout. An entry opens with six fields and no more -- frequency, mode, rig name, time, and the callsign and locator of whatever decode it was started from. A report stays empty: an FT8 SNR is not what was sent. Times come from the server, because the browser may be a phone in another timezone, and the panel says so when the two disagree by more than a second. Worked-before answers as a callsign is typed. A decode is not a contact, so the Log button on an FT8 or APRS row opens an entry and logs nothing by itself. The ham layout is the fifth operator layout, opening on the logbook with the radio controls around it, offered only where the rig can transmit. Two bugs found on the way, both in code written here: a frequency of a whole number of megahertz ending in a zero rendered as a tenth of itself, in Rust and in TypeScript alike, because trimming trailing zeros from "20.000000" walks back through the point. The API also sits under /api/logbook rather than /logbook, so it cannot shadow its own page the way /bookmarks does. Closes #54 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 21:12:03 +02:00
sjgandClaude Opus 5	a19633e81f	[fix](trx-frontend-http): rebuild the background decode panel around one list ... The panel listed the same bookmarks twice: a status card at the top naming the selected ones and what each was doing, and a checklist below naming all of them with the checkboxes. Neither list said which row in the other it meant, so choosing what to decode and reading what was being decoded were two passes over the same names. The checklist itself sat squeezed against the right-hand edge, because the row it lives in sets `align-items: flex-end` and an inline `flex-direction: column` turned that into "push everything right". It is one list now. A row carries its checkbox, its name, its frequency, mode and decoder, and its own state, so picking and watching happen in the same place. What the rig can hear moves up beside the switch, where it explains why a selected bookmark reads out of span, and the selection adds itself up under the list. The states lose the ✓/△/✗ they carried next to a dot that already said as much, and say what they mean: "Out of span", "Scheduler has it", "Nobody listening", each with the reason on hover. The empty list now says which of the two reasons it is empty for, and what to do. Save was live from the moment the panel opened, with nothing to save; it now offers itself only when something has changed. Two races behind it, both of which left the panel useless rather than ugly: * The rig was whatever the panel was handed at load. Loading before the rig list arrived handed it null, and the next telling only came when the operator switched rigs — so the panel stayed empty and silent. Every rig list refresh now passes the rig on, and both panels ignore one they have. * The settings panels are wired once the session is up, but their modules import asynchronously and the wiring skipped whichever had not arrived. A panel that missed it kept dead buttons for the rest of the session: no filter, no Select All, no Save. Wiring runs again when the modules land. Closes #52 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 19:54:36 +02:00
sjgandClaude Opus 5	c8b6f2d536	[fix](trx-frontend-http): stop freezing the page in the browser cache ... index.html, the stylesheets and the entry bundles are all served from fixed URLs and answered with `public, max-age=31536000, immutable`. Nothing in those URLs changes when the bytes behind them do, and immutable tells the browser not to ask, so a client that visited once could go on running the page it downloaded then — for a year, with the build-stamped ETag never consulted. That is how a layout fix ships and one browser still shows the old behaviour while every other one has it: not a rendering difference, a copy of last week's stylesheet. Only the shared chunks are content-addressed — esbuild hashes their names — so only they can be kept forever. Everything served from a stable URL now answers `no-cache`, which asks and gets a 304 in the ordinary case, at the cost of one conditional request per asset per load. Vendored files with a version in the URL stay immutable; Leaflet, whose URL does not name its version, revalidates with the rest. Covered both ways: a unit test on the policy each asset gets, and endpoint tests that the page, the stylesheet and app.js come back revalidating with an ETag, and that an unchanged one answers 304 under the same policy. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 10:50:26 +02:00
sjgandClaude Opus 5	17300170cc	[test](trx-frontend-http): read the statistics counters past their formatting ... The new assertion parsed the decode counter with Number() on its text. The counters are written with toLocaleString(), so the 1220 records the history fixture serves arrive as "1,220" and the parse gave NaN. It passed here and failed on CI because the count in the local repro was 18 — below the point where grouping appears — and the runner's locale groups where mine did not. Read the digits and ignore the separator, whichever one the locale picks. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 09:43:43 +02:00
sjgandClaude Opus 5	e41c13917d	[fix](trx-frontend-http): put HF APRS on the map, under its own source ... The HF APRS list never plotted anything. Its plugin ships in the map plugin group and its packets carry positions, but nothing ever handed one to the map, so a station heard on 30 m appeared in the panel and nowhere else — and unlike the replay gaps around it, no reload brought it back. Plot it, and not as more VHF APRS. HF is a different band and a different path, and lumping the two together would leave no way to tell them apart or to look at one without the other, so it goes on as a source of its own: its own colour, its own chip in the map's Show filter, its own entry in the source legend, and its own clear. Station entries are keyed by source and callsign rather than callsign alone, so a station worked on both bands keeps a marker for each while the popups, the search text and the tracks still show the callsign as heard. decode-flow feeds an HF beacon alongside the VHF one and checks all of it: both reach the map under their own sources, the Show row offers HF APRS next to APRS, and turning that chip off takes the HF station off the map while the VHF one stays. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 09:05:26 +02:00
sjgandClaude Opus 5	ae7df31d91	[fix](trx-frontend-http): replay what arrived before a lazy view loaded ... Opening the Map or Statistics page showed only what had been decoded since the moment it was opened, and a reload — landing straight on the tab, so its module loads at startup ahead of the history — was the only way to see the rest. Two things were being thrown away. The decode log the Statistics page counts lives in the map module, which is lazy. Recording into a module that is not loaded yet is a no-op, and unlike the map markers nothing replayed the log when it finally arrived, so every decode heard before the first visit was simply never counted. Hold those records in the client and hand them over when the module attaches. The map's own replay covered APRS, AIS and VDES, whose plugins implement syncMap, but not the grid squares: the FTx family and WSPR plotted locators as decodes arrived and had no replay at all, so everything they heard before the map loaded was lost, and the unique-grid counter with it. Both plot through a helper now, which their syncMap replays oldest first. A replayed WSPR spot carries the frequency it was heard on rather than one worked out against wherever the dial has moved to since. Pinned in decode-flow, whose history fixture gains FT8 and WSPR spots: after a first visit the statistics count every stored decode and every grid square, which before this change were 0 and 0. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 08:56:06 +02:00
sjgandClaude Opus 5	bf3bcc8a84	[fix](trx-frontend-http): show one rig at a time on the digital modes page ... A client connected to several rigs decodes all of them at once, and the decode stream carries every rig's traffic to every browser. The decoder panels listed all of it: a station a background rig copied on another band appeared in the APRS list next to the selected rig's, the vessel counts and the "latest seen" lines counted both, the status lines said "Receiving" because some other rig was, and the CW pane interleaved two rigs into one stream of text that read as neither. The page is about the rig the operator selected — the one whose spectrum is on screen and whose audio is playing — so each panel now shows what that rig heard: rows, counts, latest-seen, status, the live picture a WEFAX or SSTV frame is painting, and the CW pane. Nothing is dropped on the way in. The map is the whole station's view, has its own rig filter, and would empty out if the plugins stopped feeding it, so the histories still hold every rig and the map still plots them. That also means a switch loses nothing: the runtime gained a rerender hook, which the rig switch calls, and switching back brings the other rig's traffic up again. The CW pane is the exception — a running stream of text cannot be unpicked after the fact — so it starts empty on the rig switched to. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 08:11:13 +02:00
sjgandClaude Opus 5	86dd36312e	[test](trx-frontend-http): type the frequency instead of filling it ... tune-links drove the dial with Playwright's fill(), which writes a value into the field without a keystroke. The app arms its guard against its own refreshes on the first keydown, so a filled field stays unguarded: any state update landing between the fill and the Enter rewrites the field with the frequency the radio is already on, and the Enter then re-applies that. The window is a few milliseconds wide on a developer's machine and wide enough to lose on a loaded CI runner, where the test failed claiming the tuning had landed on the frequency it started from. Type it the way an operator does: select the field, then send the characters as keystrokes. The select arms the guard before a single character changes. Under CPU throttling that reproduced the failure — 1 in 6 runs with fill(), on this branch and on main alike — typing came through 8 runs clean. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 02:47:04 +02:00
sjgandClaude Opus 5	15ff686542	[fix](trx-frontend-http): keep the mini views to the rig on screen ... The decode SSE stream and the history behind it are not rig-scoped: every rig's decodes reach the browser, each carrying the rig that heard it. The panels on the decoder tabs want that — they aggregate the whole station — but the mini views over the waterfall caption the spectrum underneath, and they were reading the same unfiltered histories. A background rig copying APRS on another band put its frames over the active rig's waterfall. The mode gate did not help: it reads the mode of the rig on screen, so those frames appeared whenever that rig happened to be in PKT. Filter each overlay on the rig it belongs to, through one shared predicate that compares a decode's rig_id with the per-tab active rig already driving the spectrum and the audio. A decode that names no rig, and a session that has not learnt its rig list yet, still show everything. The FTx normalizer was dropping rig_id on the floor, so it now keeps it. CW needed more than a filter: its lines accumulate character by character, so two rigs copying at once braided their text into one unreadable line. Lines in progress are now kept per rig. The bar repaints in render() move into refreshDecodeBars(), which the rig switch calls as well — otherwise the outgoing rig's frames stayed on screen until the next state update — and which finally includes the CW bar. Closes #49 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 00:46:39 +02:00
sjgandClaude Opus 5	b78c4a4dd4	[feat](trx-rs): make spectrum affordable over a slow link ... Spectrum dominates the server↔client connection, and all three things that govern its cost were working against a poor link. **It was polled, one round trip per frame.** The client asked for a frame every 50 ms on a dedicated connection and waited for the reply, so the frame rate was capped at 1/RTT — on a 200 ms link, five frames a second no matter what was configured. Add SubscribeSpectrum alongside the existing SubscribeMeter: the server pushes frames from a per-rig broadcast that rig_task fills only while somebody is subscribed. A server too old to know the command answers with an error and leaves the connection usable, so the client falls back to polling on the same connection without reconnecting. **Bins were JSON floats.** 1024 bins spelled out as decimal text is around 10 KB a frame, ~200 KB/s at full rate — while the very next hop, client to browser, already sends the same information as base64 i8 in about 1.4 KB. Bins now travel base64-encoded whole dBFS, the resolution the display draws at anyway. Decoding still accepts the old array form. **Nothing was tunable.** [sdr].spectrum_fft_size and [sdr].spectrum_interval_ms replace the compile-time FFT size and cadence; [[remotes]].spectrum_interval_ms lets the client ask for less. 512 bins at 5 frames/s is roughly 3.5 KB/s against roughly 200 KB/s before. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 00:15:50 +02:00
sjgandClaude Opus 5	f396e8f235	[fix](trx-frontend-http): give satellite passes their own page ... Pass predictions were a third view inside the Weather Satellite Decoder card, under Digital modes — a planning tool filed behind a decoder toggle, beside the FT8 and WEFAX panels it has nothing to do with. Nothing about knowing when a bird comes over belongs there. Move them to /satellites, reached from Tools alongside Statistics, Recorder, Settings and About: occasional destinations that live behind that menu rather than taking a slot in the operating strip. Adding a sixth strip button wrapped the phone nav onto two rows and cost the desktop strip its labels at 1280px, so the tab is hidden from the strip exactly the way its four peers already are — the nav is byte-for-byte what it was. The prediction code moves out of sat.ts into its own plugin that loads with the page, so the decoder card no longer carries it. Countdowns stop when the page is hidden and each visit reloads, since passes go stale while it is closed. The server grows a /satellites index route so a deep link or a refresh serves the SPA shell rather than a 404. Closes #47 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-07 00:06:00 +02:00
sjgandClaude Opus 5	46c9827e8a	[fix](trx-config): keep the generated example off the machine that made it ... The example is generated from the config defaults, and [decode_logs].dir defaults to the running user's cache directory. So the file rendered /Users/sjg/Library/Caches/trx-rs/decoders on the machine that generated it and /root/.cache/trx-rs/decoders in CI, and the up-to-date test failed for everyone but its author. Pin dir to an illustrative /var/lib/trx-rs/decoders in the example config; omitting the key still falls back to the per-user directory. A new test asserts the rendered example contains none of this machine's home, cache or config directories, so the next environment-derived default cannot slip through the same way. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 22:08:06 +02:00
sjgandClaude Opus 5	0fc977f19f	[style](trx-config): apply rustfmt ... Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 21:44:17 +02:00
sjgandClaude Opus 5	084f629b5b	[docs](trx-rs): record the trx-config crate and its commands ... Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 21:39:09 +02:00
sjgandClaude Opus 5	bc63ded583	[feat](trx-config): warn about deprecated configuration keys ... Several keys quietly stopped doing what they look like they do, and nothing said so: [remote] and the flat per-rig sections are ignored outright once [[remotes]] / [[rigs]] exist, [frontends.rigctl].port and --rigctl-port have been dead since rig_ports replaced them, [frontends.audio].rig_ports is superseded by rig_urls, and default_rig_id was renamed to default_rig_name. Warn once at load, naming the replacement. Defaults are indistinguishable from explicit values after deserialization, so the loader now records which key paths the file actually set and the checks work off that — no warning for a setting the user never wrote. The single-rig flat layout is not deprecated: it is the documented simple form, and only draws a warning when [[rigs]] is silently shadowing it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 21:35:52 +02:00
sjgandClaude Opus 5	cfaeb6ee15	[docs](trx-rs): generate the example config and correct the manual ... trx-rs.toml.example was maintained by hand and had fallen well behind: no [[rigs]], no [[remotes]], no [timeouts], no bandplan or decode-history settings, and a [frontends.http].default_rig_id that had been renamed. Generate it from the config structs instead, so a new field shows up the moment it exists, and add a test that fails when the checked-in copy drifts: cargo run -p trx-config --example generate_example Section comments come from a small table; a section without an entry is still emitted, so forgetting a comment can never drop a setting from the example. The manual was wrong about the basics. It listed five config search paths, none of which the loader has ever looked at (the real order is ./trx-rs.toml → XDG → /etc), called --print-config output "fully commented" when it carries no comments at all, and documented a TRX_PLUGIN_DIRS variable no code reads. It also still described [frontends.rigctl].port as the bind port years after rig_ports replaced it. Fixed, and the new configuration features are written up alongside. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 21:32:46 +02:00
sjgandClaude Opus 5	76bcce8c54	[feat](trx-config): let secrets live outside the config file ... Tokens and passphrases had exactly one representation: plain text in trx-rs.toml. That is awkward for config-management tools, for a config kept in a private repo, and for anything shared between machines. Two alternatives: - ${VAR} anywhere in a config string, expanded from the environment at load. An unset variable is an error rather than an empty string — a silently blank passphrase is how authentication gets disabled by accident. - A *_file sibling for every credential: [listen.auth].tokens_file, [[remotes]].auth.token_file, [frontends.http.auth].rx_passphrase_file and .control_passphrase_file, [frontends.http_json.auth].tokens_file. Setting both forms is an error rather than a guess about which wins. Plus a nudge: a config file that holds credentials inline and is readable by group or others gets a warning naming the chmod that fixes it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 21:27:18 +02:00
sjgandClaude Opus 5	88ed3da6cc	[feat](trx-server): make the decoder set configurable per rig ... Every rig started nine decoders — APRS, HF APRS, CW, FT8, FT4, WSPR, LRPT, WEFAX, SSTV — whether or not anyone ever looked at the results. Two rigs on a Pi meant eighteen decoder tasks chewing CPU for modes the operator does not run. Only the SDR virtual channels had a decoder list; the analog path had no say at all. Add [decoders] per rig: [decoders] enabled = ["cw", "ft8", "wspr"] output_dir = "/var/lib/trx-rs" using the same decoder names as [sdr.channels].decoders, so there is one vocabulary. enabled defaults to every decoder, so upgrading changes nothing. An unknown name is a config error rather than a silently ignored entry. output_dir also replaces the hard-coded cache paths for the decoders that write images, so SSTV, WEFAX and LRPT output can live somewhere the operator chooses. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 21:22:03 +02:00
sjgandClaude Opus 5	7c69e0de08	[feat](trx-rs): add --check-config to the server and client ... Validating a config meant starting the daemon and reading the first error it died on, fixing that, and repeating. Add --check-config, which loads the config through the real loader, reports every problem at once and exits 0/1: $ trx-server --check-config --config trx-rs.toml trx-rs.toml warning: unknown config key 'listen.prot' (did you mean 'listen.port'?) error: [general].log_level 'verbose' is invalid (expected one of: ...) error: [rig.access].baud must be > 0 for serial access error: [audio].frame_duration_ms must be one of: 3, 5, 10, 20, 40, 60 error: [listen] and rig "default" [audio] would both bind 127.0.0.1:4530 Validation grows validate_all()/validate_resolved_all() alongside the existing first-error entry points; validate() is now the first element of validate_all(). Sockets are built by one helper shared by startup and the check, so the two cannot disagree about what --listen overrides. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 21:08:20 +02:00
sjgandClaude Opus 5	fbc4f6e398	[feat](trx-config): add a resolved-config validation phase ... Some things can only be checked once CLI overrides have been folded in and the rig/remote lists are final, so nothing checked them at all: - The client's per-rig maps (rigctl.rig_ports, audio.rig_urls, audio.rig_ports, decode_history_retention_min_by_rig, http.default_rig_name) are keyed by a remote's short name. A typo used to spawn a rigctl listener that injected a rig_id no remote answered to, without a word in the log. - Nothing noticed two listeners claiming one socket. [listen].port and a rig's [audio].port could both be 4530; on the client, http, http_json and each rigctl rig port could collide freely. Add validate_resolved() to both configs, run after argument parsing, plus a shared socket-conflict check that treats a wildcard address as conflicting with any address on the same port and ignores port 0. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 21:02:26 +02:00
sjgandClaude Opus 5	d42ca4f030	[fix](trx-config): validate every rig, not just the legacy flat one ... ServerConfig::validate() checked the flat [rig]/[audio]/[behavior] fields and gave [[rigs]] entries only an id/audio-port uniqueness pass, and validate_sdr() returned early unless the *flat* access type was "sdr". A multi-rig SDR station therefore got no Nyquist, stream_opus, duplicate-decoder or tx_enabled checking at all, and a rig entry with frame_duration_ms = 7 or a missing baud rate started and failed at runtime. Move the per-rig rules into validate_rig_instance() and validate_sdr_instance() and run them over resolved_rigs(), which already synthesises the flat layout as a single entry. Both layouts now go through the same code, and multi-rig messages name the rig they came from. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 20:57:59 +02:00
sjgandClaude Opus 5	335922fecc	[feat](trx-config): report unknown configuration keys ... Every config struct is #[serde(default)], so a misspelled key was dropped in silence and the setting kept its default. Writing `prot = 9999` under [listen] started the server on 4530 without a word. Collect the ignored key paths with serde_ignored and pair each with the closest known key at the same level: WARN unknown config key 'listen.prot' (did you mean 'listen.port'?) Warnings by default, so a config written for a newer version still runs on an older binary; --strict-config makes them fatal for CI. Logging now starts before validation so these warnings are actually visible. trx-configurator --check drops its hand-maintained key lists and re-implemented range checks in favour of the real loader and validators, so it no longer passes configs the binaries reject. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 20:55:01 +02:00
sjgandClaude Opus 5	bede2e34fe	[fix](trx-config): accept both sectioned and bare config files ... trx-configurator wrote standalone configs with [general]/[rig] at the root while the loader required a [trx-server] section header, so every config the wizard generated with --type server or --type client was rejected by the binary it was generated for: $ trx-server --config trx-server.toml Error: ParseError("trx-server.toml", "missing [trx-server] section") Teach the loader to fall back to the document root when no section header is present, so hand-written standalone files keep working, and have the wizard emit the same sectioned shape --print-config does. A file carrying only the *other* component's section still reports the missing section rather than silently loading defaults. Round-trip tests now load every document the wizard can generate through the real loader and validator. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 20:46:59 +02:00
sjgandClaude Opus 5	da58a004fe	[refactor](trx-config): extract client/server config into a shared crate ... The setup wizard, the server and the client each carried their own idea of what a valid config looks like: trx-configurator validated with hand-written toml_edit key lists while the binaries validated with serde plus their own validate(). Nothing kept the three in sync. Move ServerConfig, ClientConfig, the section loader, the shared validators and the endpoint-URL parsing into a new trx-config crate that all three depend on, so there is one definition of the config to drift from. The binaries keep a thin crate::config re-export so their internal paths are unchanged. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SyX26FCpMQxiBoC7r5K1A7 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 20:44:42 +02:00