172 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
sjgandClaude Opus 5	e52c4839c9	[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>	2026-08-06 23:23:48 +02:00
sjg	77b283cb78	[fix](trx-frontend-http): size the nav's labels against the screen, not the font ... CI cut "Bookmarks" off at 360 px where this machine had ten pixels to spare. How wide a platform draws a word varies by more than ten per cent, so a fixed 0.6rem label is a bet on the machine it was measured on -- the same bet ui-core's own comment warns about where it explains why the tab strip reflows by measurement rather than at a width. The long labels are sized with `clamp(0.46rem, 2.2vw, 0.62rem)` now. A tab is a fifth of the viewport, so what fits in it follows the viewport; tying the label to the same thing leaves better than a fifth of the tab spare at every phone width, and holds with the text drawn 30% wider than it is here. The test asked the wrong question too. "Does the label fit" is a question about font metrics, and it will keep answering differently on different machines. It now asserts what actually matters: a label stays inside its own tab, and if it is too long for it, it ends in an ellipsis rather than being cut through a letter. Both hold whatever width the platform draws the words at. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 19:41:45 +02:00
sjg	6284747339	[fix](trx-frontend-http): let the bottom nav's labels fit inside its tabs ... The bottom nav keeps its labels under the icons -- that is what makes it navigation rather than five glyphs -- but the labels did not fit the tabs. On a 360 px screen "Bookmarks" and "Digital modes" were cut off mid-word and ran into each other: "Bookmarks igital mode". The stylesheet already meant to handle it. Three rules shrink the long labels, and a rule twenty lines further down sets the size for all of them; identical specificity, later in the file, so the blanket rule won and nothing was ever shortened. Those rules now come after it. Beyond that, tabs sized themselves to their own labels, so "Map" and "Digital modes" were given the same room. They divide the bar evenly now, which is the shape of every bottom nav, and a label that still runs long ellipsises rather than being cut through a letter. "Digital modes" does not fit at any size worth reading, so the nav shows "Digital" and the button carries the full name as its accessible label; the short form is hidden from assistive tech, which reads the button's name instead. The phone layout test now checks that no label in the nav is cut off at 430, 390 or 360 px, and that the shortened tab still says what it is to something that reads the page rather than looks at it. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 19:14:31 +02:00
sjg	14ad6e241b	[fix](trx-frontend-http): stop the radio controls running off the side of a phone ... On a 390 px screen the transmit controls were laid out at x=400, off the side of a tray 354 px wide: present, invisible, and reachable only by a horizontal scroll with nothing to say it was there. The page scrolled sideways by a dozen pixels as well. Three causes, each in a different place. A container query at the end of the stylesheet re-imposes `flex-wrap: wrap` on a narrow tray's rows. That is right while a row runs left to right; below the phone breakpoint the row is a column, and wrapping a column starts a *second column* — which is what put the transmit controls beside the tray rather than under it. The rule outranks the phone one, two classes to its one, and sits later in the file, so it now excludes itself below that breakpoint rather than being overridden. The tray is a grid, and a grid column sizes to its content. One row wider than the screen — the mode picker, six buttons across — dragged the whole tray out with it. `minmax(0, 1fr)` lets the column be as narrow as the phone and the rows wrap inside it. The wavelength and signal-strength readouts are given the width of their column on narrow screens, but with padding on a content box that is the column's width plus the padding. Those dozen pixels were the page's sideways scroll. They are border-box now. Also: the rig picker was taking 139 px of a 338 px bar, pushing the rest of the top bar into the overflow menu. It is capped and ellipsised on phones, with the full name still in the menu it opens. The map's filter bar, collapsed, keeps both its anchors and shrinks inside them rather than dropping `left` to be sized by shrink-to-fit, and no longer asks for a compositing layer it does not need. An absolutely positioned, backdrop-filtered, composited box sizing itself from its content is the shape of thing that renders as nothing on an engine other than the one it was written against — which is what Edge does with it. Unverified there: this machine has no Edge to test with. tests/mobile-layout.mjs holds the page to it at 430, 390 and 360 px: no sideways scroll, nothing laid out past the right edge, the transmit controls inside the screen, the rig picker within its cap — and the collapsed filter bar still on screen with a button to bring the filters back. The SSTV panel test stamped its picture with a fixed date, which the panel's own retention window dropped once that date was a day old. It uses a recent stamp now. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 19:06:15 +02:00
sjg	06971ff65c	[style](trx-frontend-http): make the spectrum control strip one strip ... The row of controls under the plot held four different control heights, units as loose text beside the field they belonged to, and a quarter of its width as a hole in the middle. Between about 1100 and 1400 px it came apart: the bandwidth cluster wrapped to two lines while the level cluster stayed on one, so the two sat at heights that matched neither each other nor anything else on the page. Every control stays, in its order, with its name and its behaviour. This is the styling and the layout. A field is now one box -- name, value and unit inside a single border -- so a number cannot be read apart from the unit it is in. Fields, buttons, the peak-hold select and the contrast slider are one height, border-box so a button's own border cannot add two pixels to it, and 2.4rem under a coarse pointer where a fingertip needs the room. The contrast readout holds a fixed, tabular slot, so the row no longer twitches between 1.0 and 0.9. The container wraps and a cluster does not: a cluster that will not fit drops whole to the next line and starts it left-aligned. The slack goes to a spacer rather than to `space-between`, which is what opened the hole. Two things this turned up. The select carries `status-input` for other layouts' sake, which drew a box inside the field's box. And the narrow -screen rules lived in a media query earlier in the file than the rules they override -- identical specificity, so the later one won and the phone layout had been overflowing sideways rather than stacking. The narrow rules now sit directly after what they override. The layout test measures the strip at three widths: one height across every control, no overflow, inside the plot, and clusters either sharing a line or each having one -- never one floating against the middle of the other. docs/Spectrum-Controls-Rework.md records what was wrong and what was deliberately left alone: the two different Autos, the settings that do not persist, the one-shot buttons, and Sweet-spot's silence while it retunes the SDR. Those are behaviour, and are for another day. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 01:10:26 +02:00
sjg	18107ce07e	[feat](trx-rs): receive SSTV pictures end to end ... Wires the SSTV decoder into the stack, from the audio the server already has to a panel in the browser that shows the picture arriving. Server: a decoder task alongside the WEFAX one, running whenever the decoder is enabled and the rig is in a mode SSTV is sent in. A finished picture is written to the cache as a PNG and sent on as a message; the rows are sent as they decode, so a client can watch two minutes of Martin M1 fill in rather than waiting for it. Pictures join the decode history, are replayed to a client that connects later, and survive a restart. Protocol: SetSstvDecodeEnabled and ResetSstvDecoder, a sstv_decode _enabled flag in the rig state, two audio message types, and Sstv and SstvProgress on DecodedMessage. The history stores the message without its base64 payload -- the picture is already on disk, and a megabyte per entry is not what a history is for. Client: pictures land in their own history, and the PNG the server sent is written to the local cache so /sstv-images/ can serve it back. That endpoint and the WEFAX one now share their filename checks rather than each carrying a copy: no separators, no parent references, .png only. Web UI: an SSTV sub-tab beside WEFAX, with a live canvas the rows paint into at the line number they carry, a card for the last picture, and a filterable history with links to the files. Rows below the one arriving are grey rather than black -- not yet received is a different thing from received as black. A picture is not a spot, so neither pictures nor their progress updates reach the decode statistics; that exclusion list had grown by hand for LRPT and WEFAX and is now one named set. The decoder crate gains what the server needed to hand a picture on: to_png, to_png_base64 and save_png, with file names stamped in UTC so they sort. Panel behaviour is tested with the plugin runtime: rows painting at their own line numbers rather than in arrival order, a completed picture linked by file name alone with no server path in the page, a cut-off picture reported as partial, clearing, and the toggle following the rig state. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-06 00:14:59 +02:00
sjg	a0b0c0ed81	[feat](trx-sstv): decode SSTV pictures ... A new decoder crate covering the modes SSTV is actually sent in: Martin M1/M2, Scottie S1/S2/DX, Robot 36/72, PD50 through PD290, and Wraase SC2-180. The mode comes from the VIS header every transmission opens with, so nothing has to be told what is arriving. Modes are a table rather than code: a list of segments -- sync, gaps, and one scan per colour channel -- plus a colour model and a geometry. The decoder reads the offset of each scan straight off that list, which is what makes fifteen modes cost about as much as one, and a new mode a table entry. The segment lists are checked against the published line durations in a test, because both are transcribed by hand from the same specification and a digit wrong in one is unlikely to be wrong identically in the other. Signal path: band-pass over the SSTV band, Hilbert FIR, instantaneous frequency by phase difference, then a state machine that walks the transmission a line at a time. Each line is looked for where the mode says it should be and nudged into place by the sync pulse found near it -- two sound cards never agree exactly, and over the two minutes of a Martin M1 frame an uncorrected error of a few parts per million shears the picture visibly. Rows are emitted as they decode, so a picture can be watched arriving, which is most of the appeal of the mode. Four things this cost, each now the reason a piece of it is shaped the way it is: The per-sample frequency estimate ripples by ±95 Hz at 1200 Hz, where the Hilbert approximation is weakest, though its mean is exact. Pixels average over their own window and were always right; the VIS bits and the sync detector classify individual samples and were reading the ripple. Both now read short means. Pixels deliberately still do not, so edges stay where they are. Broadband noise cost the whole picture, not part of it: a phase-difference detector answers whatever is loudest, and there was no input filter. Hence the band-pass, which is what every real decoder does first. A sync search window shorter than a sync pulse rejected every pulse arriving late in it, for being short. The first line's sync search locked onto the VIS stop bit -- 30 ms at exactly the sync frequency, immediately before the picture starts. The header already says where the picture begins, so the first line no longer searches. Tests: nine modes are encoded from a test card and decoded back, compared pixel by pixel, alongside silence around the signal, a transmission cut off part way, two transmissions back to back, 20 dB of noise, and a transmitter clock 0.1% fast. The encoder that produces those signals reads the same table as the decoder, so a round trip tests the decoder and not the timings; the timings are held to the published line durations separately. Nothing is wired into the server or the web UI yet: this is the decoder alone. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 23:29:52 +02:00
sjg	a2c630a92b	[feat](trx-frontend-http): put the tuned frequency in the address bar ... A receiver spreads by being linked to, and there was nothing to link to: the routes carried the tab and nothing else, so "listen to this" could only ever mean a screenshot and a frequency typed out in a message. The query string now carries the dial -- rig, frequency, mode and bandwidth -- in both directions. Opening a link selects the rig, sets the mode, tunes, then applies the bandwidth: a mode change brings its own default bandwidth with it, so an explicit bw has to land after it. Frequencies are read the way someone writes them by hand (7074k, 14.074M) and written back as whole Hz, so what comes out of the address bar is the same link in canonical form. After that the address bar keeps up with the dial, which is what makes it copyable at any moment rather than only at load. It is rewritten with replaceState -- tuning is not navigation, and a swept dial would otherwise bury the back button. A link button in the top bar copies the current link; it folds into the overflow menu when the bar is tight. Applying a link changes the radio, so an rx session says so instead of failing control calls one at a time. A tab listening to a virtual channel leaves the address alone rather than publishing a frequency the rig is not on, and bw is skipped in both directions on rigs without filter control, which would only refuse it. The fixture pinned every state frame to 100 MHz plus jitter to keep frames distinct, so no test could observe tuning at all. The jitter moves to the S-meter and the fixture echoes set_freq/set_mode/ set_bandwidth, as it already did for squelch. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 22:31:12 +02:00
sjg	09634eb851	[fix](trx-frontend-http): tidy up the map's filter bar ... The bar explained itself in prose: "All bands visible by default" sat between the chips and the next group, taking width the bar could not spare and reading as a stray line of text. An "All" chip says the same thing in a chip's width and gives the selection somewhere to be undone. Band chips also came up dimmed at the very moment every band was on the map -- an empty selection is no filter at all, so nothing is dimmed until something is picked. The path toggles drop their "On"/"Off" suffix, which cost most of a row and only repeated what their own highlight already said; state moves to aria-pressed and the tooltip. The rest is alignment. The rule dividing the buttons from the filters is drawn on the button block's edge, and a centred block left it floating as a stub beside a two-row bar; stacked, it lay down the left of a block that sits underneath. The labels sat at their natural widths, so each row's first control started somewhere different, and the two pairs of phase buttons differed in width, so the groups after them missed each other by four pixels. One gutter for every label, one width for both pairs, and the search field moved last where it can take the room the fixed-width groups leave. The map layout test now covers the chips, the divider's height and the rows' shared start. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 21:43:57 +02:00
sjg	90ab7781ad	[fix](trx-frontend-http): stop the browser offering saved values for frequency ... The tuned and centre frequency readouts are text inputs, so the browser keeps what has been typed into them and offers it back in a dropdown -- Edge does this out of the box, dropping stale frequencies from other sessions over the reading. Turn autofill off on both, along with autocorrect and spellcheck, which have no business near a number either. Fixes #39 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 19:36:22 +02:00
sjg	88d04253ca	[fix](trx-frontend-http): move the map's fullscreen and filter toggles into the bar ... Fullscreen and Hide Filters floated in their own block over the map's top-right corner, separate from the filter bar they sit beside. Put them at the right-hand end of the bar, behind a separator. What made this awkward before is that Hide Filters cannot live inside the thing it hides, so the collapse now applies to the filters alone: the bar keeps its two controls and shrinks to them at the map's right edge, leaving the whole map visible and the way back one click away. Fixes #38 Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 19:34:09 +02:00
sjg	08005c5c07	[feat](trx-frontend-http): lay the map filters out as a bar across the top ... The filters were a 30rem column parked in the bottom-right corner, covering a third of the map they filter. Lay them out horizontally instead: one row per group -- label beside its control, thin rules between -- across the top of the map, spanning ~87% of its width at 1600px and wrapping to a second row as it narrows. It starts clear of Leaflet's zoom buttons and stops short of the corner controls, which stay outside it: the button that hides the filters cannot live inside the thing it hides. The bottom-left band legend keeps its place. The sentence explaining the two path toggles would have swallowed the bar, so it moves to their tooltips and is shown inline only in the stacked narrow-screen layout. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 07:54:21 +02:00
sjg	026f816ddb	[fix](trx-frontend-http): replay stored decodes onto the map when it loads ... The map module is lazy: it arrives when the Map tab is first opened, which is normally long after startup restored the decode history. Until then aprsMapAddStation, aisMapAddVessel and vdesMapAddPoint are undefined, and the decoders' `if (lat != null && ... && fn)` guards quietly dropped every restored position. Nothing replayed them once the module did arrive, so the map came up empty and filled in only from decodes heard afterwards -- a station heard once was never plotted at all. A second reload appeared to fix it because the cached module then loaded early enough to win the race against the history fetch. Give DecoderPlugin an optional syncMap(), implement it for APRS, AIS and VDES over the history each already retains, and have map-core call trxPluginRuntime.syncMapAll() as it attaches. The add functions are keyed by callsign, MMSI and point, so replaying updates in place and cannot duplicate a marker; the replay runs oldest-first so tracks are rebuilt in the order they happened. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 07:34:45 +02:00
sjg	d31b6f545f	[test](trx-frontend-http): watch the history progress from inside the page ... The replay-progress check polled the overlay from the test every 100ms. A replay that starts and finishes between two polls is never sampled, and the test then reports that no progress was shown at all -- the source of the intermittent "no progress was shown while the history loaded" failure. Record the samples from a MutationObserver installed before the page's own scripts run, so a fast replay is observed rather than missed. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 07:10:57 +02:00
sjg	39f551c914	[fix](trx-frontend-http): hold the APRS symbol column open for frames without one ... renderLocalAprsSymbol() returns nothing when a packet carries no symbol table or code, so those rows lost the icon's 24px slot and every column after it -- callsign, type badge, summary -- slid left against the rows around them. Frames that do carry a symbol then read as indented. Render an empty slot of the same size instead, so a list mixing position reports with messages and telemetry still lines up. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-05 07:10:33 +02:00
sjg	6d25ecdc11	[feat](trx-frontend-http): give the AIS list the same log shape ... A message was three stacked lines — time and name, then MMSI and route, then motion, distance, position and age — so a screen held eight of them. It is one line now: time, vessel, message type, and what the message says, opening in place for the MMSI, the channel frequency, the route, the age, the fix and a jump to the map. Twenty-two fit where eight did. What a message says depends on what it is. Position reports give the fix and the motion; the static and voyage reports that carry no fix give the callsign and where the vessel is bound. Both fall back to whatever fields are present rather than showing nothing. The row vocabulary the APRS list introduced is no longer APRS-specific — the classes are decode-line and decode-expanded now, shared by both, and identity sits in fixed columns so the summaries line up down the list instead of starting wherever the callsign happens to end. The three summary cards above the list go the way of the APRS ones. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 23:24:09 +02:00
sjg	37987b2779	[feat](trx-frontend-http): make the APRS list a log, and read the payloads ... A frame was a card five lines tall — timestamp, a meta line, the information field as it arrived on the air, three buttons, and a Details panel repeating the four things already on the row — so five frames filled the panel and the payload was left to be decoded by eye. A frame is one line now: time, station, type, and what the frame says. It opens in place for the path, the CRC, the raw field, its bytes and the actions. Twenty-one frames fit where five did. And the information field is read rather than echoed. Weather reports give temperature, wind, humidity and pressure; telemetry gives its sequence and channels; a message gives its addressee and text; a position gives the fix, course and speed, and the comment the station wrote. Anything that cannot be summarised falls back to the raw field, which is in the expanded view either way. HF APRS had a copy of the same forty lines of markup, differing by one badge, so both now build their rows from one function in the shared module — the CSS is shared between them and this would have broken it otherwise. Both headers lose their three summary cards for a line of counts beside the filters, which frees another fifth of the panel. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 23:12:02 +02:00
sjg	67a7bace4e	[fix](trx-frontend-http): let the decode lists fill their panel ... FT8, FT4, FT2 and WSPR size their list against the panel with flex, and the sidebar layout made the panel a grid item aligned to the start of its row — sized to its own content. The lists collapsed to their 120px minimum with several hundred pixels of the page empty underneath. The panel stretches to the row now and the sidebar keeps its own height. The marine lists were sized a different way, by formula: 100vh minus a guess at everything above them. That guess stopped matching the moment the panel changed shape, so they left a few hundred pixels unused as well. They fill the panel like the rest now, and so does CW, which had a 360px ceiling. HF APRS had no container styling at all — no scroller, no frame, no height — so its packets ran down the page. It gets what the other packet lists have. The smoke test measures each list against its panel and requires it to scroll on its own. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 22:49:28 +02:00
sjg	b48cc23d6e	[fix](trx-frontend-http): keep the rig names through the state stream ... The picker and the header showed each rig's lowercase id instead of its configured name. applyRigList takes the names as a parameter defaulted to an empty map, and the state-update path passes only the rig ids — names come from /rigs, not from a state frame — so that call landed on the default and the body, which treats "an object" as "here are the names", cleared them. One frame after load the names were gone for the rest of the session. Omitted now means no news rather than no names. The fixture is why this was invisible: it pushed an identical status payload every tick and the client skips a frame equal to the last, so render never ran and neither did the call that did the damage. Its event stream varies between frames now, as a real one does. Which immediately caught a second fault: state frames arrive continuously, and one sent before the server applied a new squelch threshold snapped the line back to where it had just been dragged from. A local change outranks the echo for two seconds, the same idea as the optimistic frequency guard beside it. The fixture also records what /set_sdr_squelch sets and reports it back afterwards — the drag test had been passing against a server that ignored the write. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 22:38:21 +02:00
sjg	c1899229a0	[fix](trx-frontend-http): stop the decode history replay giving up at 20s ... Reloading a second time sometimes showed history the first load did not, and the safety valve is why: it called one function that both released the buffered live decodes and tore the history worker down, so any load where the replay had not finished inside twenty seconds — a large backlog, a cold cache, a slow link — dropped whatever had not arrived, without a word. A reload got another go at it, and the second one is faster because everything is cached by then. Those are two separate things now. At the timeout the live decodes are released so the panels are not held back, the replay carries on, and the progress says so. The fallback's error path retries once and then says "Decode history unavailable" rather than leaving the operator to guess whether there was anything to see. The progress is no longer a scrim. It was fixed to the whole viewport with a wash over the page — the waterfall, the decode panels, all of it — for the length of the replay, which is exactly when there is something worth watching. It is a corner card with a bar: indeterminate while the payload is on the wire, then filling as N of M messages replay. None of this was reachable from a test. /decode/history answers in CBOR and the worker reads the body as CBOR unconditionally, but the fixture served JSON, so every browser run had been exercising the client's retry path and never its history path. It encodes CBOR now, including the 64-bit form the millisecond timestamps need, and decode-flow serves 1200 records and holds the client to restoring all of them on the first load, showing progress while it does, and never covering the page with it. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 22:14:13 +02:00
sjg	84a99a3636	[fix](trx-frontend-http): load the decoders that own the digital modes panels ... AIS, VDES and both APRS decoders were listed under the map plugin group alone, so opening Digital modes and clicking AIS or APRS gave an empty panel reading "Connected, listening for packets" while the decodes piled up unprocessed in the plugin runtime. They appeared only if something had opened the Map tab first, which flushed the queue. There is also a map-data group naming exactly those four that nothing loads: the loader is called with tab names and no tab is called map-data. They load with the tab whose panels they fill now. map-core stays lazy, since their calls into it are optional and the Map tab can go on paying for Leaflet by itself. tests/decode-flow.mjs follows a decode from the wire to the map: an AIS vessel and an APRS beacon arrive on /decode, and it asserts both panels fill with the map module confirmed absent, the mini view names the vessel and offers a pin, following that pin lands on /map centred on the vessel, and both decoders leave a marker. Nothing exercised any of this before — the fixture served an empty decode stream, which is how the map links came to be broken for every decoder at once. The fixture stamps decodes as it sends them, since the client prunes anything outside the retention window, and repeats them, since the views collapse by vessel and need more than one frame to behave. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 21:31:30 +02:00
sjg	84bdf2593c	[fix](trx-frontend-http): measure whether the tab bar fits, and watch for it ... CI put the tab strip 9px into the controls at 1440px with no scaling at all, on a bar that had every degradation step available to it and used none of them. It used none because nothing thought anything was wrong: the fit test was an arithmetic estimate — identity + nav.scrollWidth + actions.scrollWidth + a 48px allowance for the gaps — and on a platform whose fonts run wider than the one it was written on, that allowance no longer covered what it stands for. An estimate that says "fits" stops the ladder before its first rung. It reads the geometry now: the controls have to stay inside the bar, and no tab may reach them. That is the same measurement the test makes, so the two cannot disagree about any platform's metrics. The tabs are the subject rather than the nav's box because the nav shrinks below its content — the box gets smaller while the tabs keep their width and slide underneath the controls. A second fault turned up while probing this: the strip only reflowed on window resize. The rig name arriving from the server, the style picker filling in, a font swapping in wider metrics — each changes what fits without touching the window, and the bar sat there as it was through all of them. A ResizeObserver on the bar and the controls covers those, and document.fonts.ready covers the swap. The guard sweeps text scales and adds a station name too long for the bar, but it should be said plainly: it passes against the old code too. Nothing here reproduces on this machine — a 4px viewport sweep from 1080 to 1500, three wide font stacks and scales from 1.0 to 3.0 all failed to make the old estimate lie. What is fixed is the mechanism that could. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 20:32:38 +02:00
sjg	c527f20a88	[feat](trx-frontend-http): give digital modes a decoder sidebar ... Thirteen decoders in a horizontal strip needed a scroller on anything but a wide window, and the open one was marked by a single underline among thirteen. They are a list down the left now, all visible at once, each keeping the state dot it already carried, with the panel for the selected one filling the rest of the width. No script changed: the sub-tab wiring, the aria roles, the decoder picker and the state-dot observers all work on the same markup, so this is layout only. Below 760px the sidebar gives way to the picker that already existed there. That path needed align-content: the tab panel fills the page height and a grid stretches its rows to match, which handed the picker a 218px row and left a 189px gap under it. The tab icon was the signal-strength bars, which is what the S-meter shows two rows above it; a pulse train says digital modes instead. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 08:01:22 +02:00
sjg	9f495021f0	[style](trx-frontend-http): name and fence the three audio groups ... The row carries three unrelated things — how loud it is, whether there is any audio at all, and how much is arriving — and only the middle one was named. Each is a group now: VOLUME over the two sliders, SQL on its own switch, LEVEL over the meter, with a hairline between each. The rules are drawn only while the row is one line, measured against the row rather than the viewport: what fits depends on whether the rig transmits and whether it has a squelch at all. A rule divides what sits either side of it, so once a group wraps the wrap is the division and the rule would just be a mark at the start of a line — which is what it was at 900px before this. The labels carry the grouping on their own below that width, and the groups stack whole on a phone. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 07:46:56 +02:00
sjg	33aa7b807c	[style](trx-frontend-http): fence the squelch off from the volume controls ... The squelch sat in the audio row on the same gap as everything else, so it read as a continuation of the volume sliders. It decides whether there is audio at all, which is not the same kind of control as how loud it is, and a hairline says so. The rule belongs to the squelch block, so it leaves with it on a rig that has none, and it stands down where the row stacks: there the line break separates them already and a leading rule would just start a line. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-04 07:36:53 +02:00
sjg	2c56d82a81	[feat](trx-frontend-http): make the SQL label the squelch switch ... Clicking SQL turns the squelch on and off. The label and the button beside it said the same thing twice — one naming the control, the other reading "On" or "Off" — where the name itself is the obvious target, and the dot already carries the state: grey when off, green while the gate passes, amber while it holds. The pressed state is on the label, so the switch reads the same to a screen reader as it looks. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 23:41:52 +02:00
sjg	0fc2115973	[fix](trx-frontend-http): make map links work before the map has loaded ... The AIS and APRS mini views link each position to the map, and neither did anything: the map module installs itself lazily, and it was the one defining window.navigateToAprsMap, so until something had opened the Map tab the global did not exist. AIS calls it inline from onclick and threw "not a function"; APRS guards the call and so failed silently. The grid links on FT8, FT4, FT2 and WSPR rows went the same way through navigateToMapLocator. The app owns both globals now, installed at startup. They record the target, switch tabs through navigateToTab — the only path that materialises the panel from its template, loads the module and updates the history entry, none of which the module's own hand-rolled tab switch did — and the target is applied once the module reports ready. The module keeps the focusing, which is its job, and exposes it as focusMapPosition and focusMapLocator. The smoke test now calls the link from a cold page, asserting the map module is not loaded first so the check cannot pass by accident. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 23:41:09 +02:00
sjg	2c1df75d19	[fix](trx-frontend-http): measure auto squelch from the meter ... Auto took the spectrum's noise floor and added 6 dB, but the threshold is compared against the channel level the meter reports, and the two sit a long way apart: the gap is set by the FFT size and window, the channel bandwidth, the decimation, and peak-versus-mean statistics. Measured on white noise it runs +22.1 dB at 48k/8k/3k, +18.7 dB at 240k/24k/12k and -1.2 dB at 1.92M/24k/12k — a 23 dB swing across ordinary configurations. Only the last of those is anywhere near right, so on a narrow span Auto set the gate some 20 dB below the noise and it never closed. It now reads the same number the DSP compares: the 20th percentile of the meter over the last ten seconds, plus 5 dB. The percentile keeps a burst of traffic inside the window from dragging the estimate up, and 5 dB clears the meter's own jitter, which measured 0.9-1.6 dB. Nothing in it converts between scales, so no part of the signal chain can put it out again. With no history yet — a fresh connection, a rig switch — it listens for a moment rather than refusing. The fixture gained a streaming /meter, without which there is nothing to measure, and the spectrum test pins auto to the meter it serves. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 23:32:18 +02:00
sjg	aefd36c4b1	[feat](trx-frontend-http): set the squelch on the spectrum, in dB ... The threshold is in dB, and since the squelch fix that is the scale the spectrum axis and the S-meter are labelled in — so the control belongs on the plot, at the level it gates. A dashed line spans the spectrum at its threshold with a grip that reads it out, dragged like the bandwidth edges, green while the signal is above it and amber while it gates. Arrow keys move it a dB at a time for anyone not using a mouse. The audio row keeps a compact version: the dB, an indicator lit from the same meter the DSP compares against, Auto, and an enable toggle that no longer doubles as the level. The slider ran 0-100% over that dB range, which gave the operator a number with nothing on screen to relate it to, and zero meant "disabled", so turning the squelch off to listen threw the threshold away. Auto now says which level it picked. Two things the browser could only show once it was on the plot: the grip landed underneath the split control at the right edge, which swallowed its pointer, and dragging to the foot of the axis hid the line — and the grip with it — instead of pinning it where it could be dragged back. The fixture could not exercise any of this: /audio answered 404, which hides the audio row and the control inside it, and the status carried no filter block, which is what tells the client the rig has a squelch at all. Both now look like an SDR, and the spectrum test drives the line. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 23:20:19 +02:00
sjg	25b9c31c9b	[fix](trx-backend-soapysdr): squelch against the meter, not the post-AGC level ... The threshold was compared against the block level measured after the IQ AGC. Holding that level at a setpoint is the AGC's entire purpose, so for every mode that has one — FM, PKT, AIS, AM, SAM, which is to say the modes anyone squelches — the comparison was against a near-constant. With FM's 12 dB of gain a weak signal reads some 12 dB hotter than it is, and the value never had the decimation correction the meter applies on top of that: around 20 dB adrift at 48k/8k, more as decimation grows. The threshold arrives in the other scale entirely. The slider maps its percentage onto -120..-30 dB and Auto takes the spectrum noise floor plus 6, both of which are what the meter and the spectrum display show. So a gate set just above the noise sat open on it. It now reads last_signal_db, which is already computed each block before the AGC and corrected for decimation — the same number the meter shows. The post-AGC measurement had no other consumer. The test feeds one signal twice and takes the threshold from the channel's own meter: 6 dB above must gate it, 6 dB below must pass it. Nothing there depends on the absolute scale, only on the two agreeing. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 23:03:46 +02:00
sjg	d68a84f7f9	[chore](trx-client): apply cargo fmt ... Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 22:57:33 +02:00
sjg	4fa191e9b6	[fix](trx-frontend-http): serve the band plan to every session ... /bandplan.json needed the control role. Route access is decided by suffix for static assets — .js, .css, .png and so on — and ".json" is not among them, so the band plan matched nothing and fell through to the catch-all. It is compiled into the binary and identical for every user, so it is public now, like the rest of them. Two things followed from that. Read-only sessions never saw a band plan at all. And since the page asks for it during startup, the request can land before the session is established: that 401 was swallowed by an empty catch and never retried, which is why the allocations sometimes only appeared after a manual reload. So the client no longer hides the failure, retries once the auth gate clears — which is exactly when a startup 401 becomes fixable — and schedules a draw when the data lands, since the strip is painted from the spectrum draw and a rig sitting between frames would stay blank. The fixture can now refuse the first request the way the server did, and the spectrum layout test holds the client to recovering from it. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 22:50:49 +02:00
sjg	c073d03ffb	[feat](trx-frontend-http): reorder the controls tray sections ... The radio's own settings now come first, then audio, then the scheduler: Advanced radio controls, Audio controls, Scheduler controls. The advanced section is not in the markup — ui-core builds it at runtime and gathers the SDR settings, virtual channel and TX limit rows into it, appending the result, which put it last however the markup was ordered. It is inserted ahead of the audio section instead. The signal readout and the TX meters stay where they are, between the controls and the sections: they are readouts rather than a section, and on an SDR the spectrum covers them anyway. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 22:41:08 +02:00
sjg	730f129404	[fix](trx-frontend-http): let the header identity give way before the tabs ... CI still put the tab strip into the controls, now at 1100px — the narrowest bar in the app, since the bookmark gutters take 9.5rem a side above that width and leave 756px against 871px at 900px. With the controls already in the overflow menu and the tabs already down to icons, nothing else could give, and what gives by default is the strip: it is the one item allowed to shrink below its content, so its tabs keep full width and slide under the controls, out of reach. The identity block takes the squeeze instead, ellipsised. A clipped station name is still readable; a destination hidden underneath the controls is not. The guard that was supposed to catch this scaled only the tabs and the controls, not the title and subtitles — which is exactly what runs out of room — and skipped 900px. It now scales every piece of text in the bar and checks all four widths. Measured across text scales from 1.0 to 3.0 at each width, the bar keeps its 16px allowance everywhere; before this, 1.6 and above overlapped at 1100px. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 22:40:36 +02:00
sjg	f95f3d0104	[feat](trx-frontend-http): centre the radio controls, split off the mode row ... The controls every rig has — mode, wheel, tune step, transmit — now sit as a centred block rather than packed against the left edge. What the current mode adds moves out from among them: WFM's six controls stretched the row sideways whenever it was active, pushing the wheel and the step pickers off centre, and SAM did the same on a smaller scale. They get a row of their own below a divider, which appears and leaves with them — an empty one would still take a track and a gap in the tray and draw its divider under controls it has nothing to do with. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 22:29:31 +02:00
sjg	e70e82c8c0	[feat](trx-frontend-http): rebuild the general radio controls row ... Mode was a full-width select: 483px of the row to display "FM". The modes are three or four characters and there are at most twelve, so they become a segmented group like the Unit and Step Scale pickers beside them — a third of the width, and one click instead of two. The <select> stays as the mode's value. A dozen call sites and several plugins read #mode.value, so replacing it outright would have reached much further than a layout change should; it is hidden from sight and from assistive tech, the buttons write to it, and everything downstream runs unchanged. Every writer re-syncs the buttons, the plugins through a new trxCore.syncModePicker. The row itself was a grid with a track per column, but the WFM, SAM and transmit columns are hidden on most rigs, so it ended in some 500px of hole. It packs left now. Same fault one level down: the power buttons sat in three fixed tracks, so a rig with neither transmit nor lock kept two empty ones and left its label chip stranded at the far edge. Unit and Step Scale move out of the frequency row and in beside the wheel and the +/- they modify, which were some 600px away. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 22:09:08 +02:00
sjg	23cc0db1fa	[fix](trx-frontend-http): drop the tabs to icons when the bar runs out of room ... CI put the tab strip 9px into the controls at 1440px on a run that passed locally: its system font is wider, and the bar had no move left to make. Controls are moved into the overflow menu until the bar fits, but once all of them were in the menu nothing else gave — the nav may shrink below its content, so the tabs kept full width and ran under the controls, leaving the destinations nearest them unclickable. Labels dropping to icons was the other half of the answer, but it hung off a max-width:1360px media query and so was unavailable at 1440px. That class now goes on by measurement, as the last step after the menu is exhausted, which is the same reasoning the controls' own fit test already uses: how much fits depends on the rig name and on how wide the platform draws the labels, not on the viewport. The class is cleared before measuring so the decision cannot ratchet, and icon widths are fixed, so it always buys back the labels' width. Labels now stay put between 1100px and 1360px while they fit, with the style picker and theme toggle behind the overflow menu instead. The suite could not have caught this: it passed on the fonts of the machine that wrote it. The layout section now repeats its fit check with the bar's text scaled up, which reproduces a wider system font anywhere — with this fix reverted it fails on macOS too. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 21:36:55 +02:00
sjg	657f952a2c	[fix](trx-frontend-http): hold the strips' place above the spectrum ... Tuning across a band edge moved the whole page under the cursor: the band plan strip is in flow, so a range with no allocations collapsed it from 18px to nothing and dragged every element below it up (measured at 1600x950: overview top 118 to 100, footer 1026 to 1008). It now keeps its height whenever a band plan could be drawn at all, and only gives it back when the feature is off, has no data, or there is no spectrum. The bookmark rail gets the same treatment for consistency, though it never moved anything — it is absolutely positioned over the top of the overview. It stays up and blank rather than vanishing. Which exposes something the rail was already doing wrong: it covers the top of the plot, and a bare div still takes pointer events, so whenever bookmarks were in range that band of the overview could not be dragged or scrolled. Only the chips are targets now. tests/spectrum-layout.mjs covers this: it streams spectrum frames, tunes between a band with bookmarks and allocations and one with neither, and asserts nothing moves and that the rail lets clicks through. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 21:26:19 +02:00
sjg	1f256cbb68	[refactor](trx-frontend-http): extract the browser test fixture ... browser-smoke.mjs carried its static server inline, which made it the only browser test that could exist: a second one would have had to copy 180 lines of routes to change a single capability flag. The server moves to tests/web-fixture.mjs behind startWebFixture(), with the rig's spectrum support, bookmarks and band plan as options. Serving a rig with a spectrum matters because that is where the layout actually lives — the panel, the strips above it and the waterfall are all gated on filter_controls, and the existing fixture reports a CAT-only rig, so none of it has ever been rendered under test. No change to what the smoke test checks. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 21:26:06 +02:00
sjg	92fbdb692c	[feat](trx-frontend-http): lay the scheduler controls out in one row ... The controls were a column — release, then the step buttons, then the status line, then the entry on air last — which read bottom-up and left the entry that is actually transmitting furthest from the buttons that change it. They now run left to right: step through the entries, hand the rig back, then the current entry behind a separator. The separator is a pseudo-element on the current-entry block rather than an element of its own, because that block is display-toggled whenever fewer than two entries are active; a standalone rule would be left hanging with nothing after it. No ids move, so the enable/disable logic in the scheduler plugin and the release polling in vchan bind exactly as before. The smoke test asserts the row's order, which is also what keeps the separator in place. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 20:58:31 +02:00
sjg	b2fbcb318d	[fix](trx-frontend-http): stop Tools lighting up on every refresh ... navigateToTab marked the Tools button by asking whether the destination tab was displayed, which is the right question at the wrong moment: the first route navigation runs while the card is still behind the loading state, where every tab computes to display:none. Refreshing or deep linking to any page therefore lit Tools alongside the real destination, and nothing re-evaluated it once the page appeared. Membership of the Tools menu answers the same question without needing anything laid out, and still reads the grouping ui-core installs rather than a second copy of it. The smoke fixture now serves the SPA shell for route paths the way the server's per-tab index handlers do, so a deep link no longer 404s and the case is testable at all; two of them are asserted. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 20:51:04 +02:00
sjg	44721b579c	[fix](trx-frontend-http): fill the map column down to the footer ... The windowed map was capped at 75% of the viewport height and at a width-derived aspect ratio, which left a dead band under it: 69px at 1600x950, and on a 420px-wide phone a 270px map on an 800px screen. Neither cap was doing useful work now that the stage spans the full width, so the map fills the column down to the footer instead. Growing into the footer needs a bound: once the column is tall enough to push the footer below the fold, using its position would push it further on every pass, so the bottom edge is clamped to the viewport. Growth then consumes the column's spare height and settles in one pass. Also drops three mapIsFullscreen() branches in the windowed path that could never be taken — the fullscreen case returns above them. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 20:50:48 +02:00
sjg	889d00007d	[feat](trx-frontend-http): box the selected tab instead of underlining it ... The desktop strip marked the current page with a 2px underline while the mobile bottom nav already boxed it, so one navigation model looked like two. The box now sits on both: a transparent 1px border on the base reserves it, so switching pages moves no neighbours, and hover fills a fainter version of the same shape. Tools carries it too — that button is marked active for the destinations the strip hides. Dropping the mobile rule's border-bottom:none, which only existed to cancel the old desktop underline, closes the bottom edge its active box had been missing. The smoke test checks all four edges. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 20:37:31 +02:00
sjg	6f53592041	[feat](trx-frontend-http): rework the page footer ... The footer floated in space below the content with no rule to close the page, its two clusters sat on a text baseline that left the source pill hanging, and the status hint was a plain line of text a size larger than the attribution beside it. Now a hairline closes the page the way .tab-bar opens it, the clusters centre on one line, and the attribution drops the opacity it stacked on top of --text-muted, which had put it below a readable contrast ratio. The status hint becomes a pill with a state dot: green when ready, amber while a command is in flight, red on connection loss. The colour comes from a data-state attribute, so every hint now goes through setPowerHint instead of assigning textContent directly. --status-ok carries the indicator green; .about-status-on picks it up too, which darkens it on light themes where the old value was barely legible. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 20:37:12 +02:00
sjg	709939f60b	[feat](trx-frontend-http): span the map across the full viewport width ... The map stage broke out of the centred .card column: negative inline margins cancel the card's centring offset and its side padding, so the stage reaches both viewport edges at every width without hardcoding either value. Its rounded corners and left/right borders go with it — edge to edge, the panel reads as a band rather than a floating card. The browser smoke test now measures the stage against the viewport, and checks that the full-bleed width does not push the page sideways. Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 20:36:50 +02:00
sjgandClaude Opus 5	5b7dd493d4	[feat](trx-frontend-http): draw APRS symbols from the sprite sheets ... Resolve a table/code pair to a sprite cell in aprs-shared, and use it from both the packet lists and the map markers, which had each been printing the raw symbol character in a bordered box. A table identifier of / or \ selects the primary or alternate sheet directly. Anything else is an overlay character, which the APRS spec draws on top of the alternate symbol -- so those stack the overlay sheet over the alternate one rather than picking a sheet. Codes outside 0x21..0x7E have no cell and keep the old character box. The sheet URLs stay in the stylesheet so a min-resolution query can swap in the retina sheets; only the cell offset is computed and set inline. Map markers share the helper through the plugin chunk, so the map stays free of any remote symbol fetch. Verified in a browser against the real stylesheet and sheets: /> is a car, /_ a WX circle, /& an igate diamond, \n a red triangle, and the overlays S> and 7# carry their character on the alternate symbol. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018huL1ELyr86yVqfAabtioA Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 19:55:50 +02:00
sjgandClaude Opus 5	02fe492dbf	[feat](trx-frontend-http): serve the vendored APRS symbol sprites ... Embed the six sheets alongside the other vendored assets and serve them under /vendor with the same immutable cache headers. The browser computes a symbol's cell from a 16x6 grid of 24px cells, so a re-vendored sheet at any other size would shift every station onto a neighbouring icon -- wrong on every packet, and invisible unless you know which glyph to expect. Pin the geometry by parsing each embedded PNG's IHDR in a test. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018huL1ELyr86yVqfAabtioA Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 19:55:50 +02:00
sjgandClaude Opus 5	1bc2c88e2f	[chore](trx-rs): vendor the APRS symbol sprite sheets ... The web UI had no symbol graphics at all, so a position report rendered its raw symbol character in a bordered box. Vendor rev H of the hessu/aprs-symbols set: three 24px sheets (primary, alternate, and the overlay characters) plus their retina variants. The set carries no single license. Individual symbols are variously vectorizations of the original WA8LMF bitmaps with unknown terms, new CC BY-SA work by OH7LZB, public-domain or CC sources, and a handful of brand logos owned by their companies. Record that as LicenseRef-APRS-Symbols with the upstream per-symbol catalogue copied verbatim, and carry the attribution pointer upstream asks for. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018huL1ELyr86yVqfAabtioA Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 19:36:27 +02:00
sjgandClaude Opus 5	863a6d8fd4	[fix](trx-frontend-http): restore decode history from the stored records ... Replay required every restored record to carry a string `type`, and stored records do not have one: an AIS entry holds mmsi, lat, lon, crc_ok and its decoder's own fields, nothing more. The filter therefore discarded all of them, and did it silently — the fetch returned its full payload, the worker decoded it, and no error was logged, so the history simply never appeared. That field identifies live SSE frames, which do carry it, which is why only replay was affected. History arrives already grouped and the group's kind is delivered alongside the messages, so `type` was never needed to route them. Require only that a record is an object. Confirmed against a live server: the first restored group is AIS, and its records expose their decoder fields with `type` undefined. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01GdyUjuXejCEfiub675z6cz Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 18:39:13 +02:00
sjgandClaude Opus 5	b252717f6b	[test](trx-frontend-http): serve a realistic decoder registry to the smoke test ... The fixture answered /decoders with an empty list, which hid most of the application from the only test that runs it in a browser. The decoder sub-tabs, their panels, the decode toggles and the bookmark decoder checkboxes are all built from that registry, so the run exercised three of thirteen sub-tabs and none of the decoder UI. Finding this needed route interception, because nothing in the suite could see it. Serve eleven decoders covering the modes the real registry spans. The run now builds 13 sub-tabs and 11 bookmark decoder checkboxes — the same checkboxes whose construction a recent fix changed without any test reaching them — and still reports no runtime errors. It also makes an existing fault observable: at 1100px the decoder sub-tab bar hides 195px of itself with no scrollbar or fade, the same silent truncation the top strip had. No assertion for it here, since that would fail until the truncation is fixed. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01GdyUjuXejCEfiub675z6cz Signed-off-by: Stan Grams <sjg@haxx.space>	2026-08-03 17:46:52 +02:00