Repository navigation
Conversation
The collapsed peek looked cramped and lopsided. The 120.dp peek height was a flat constant that happened to sit 6.dp past the bottom of the 72.dp Call CQ / Hunt row, while the row itself butted straight against the 42.dp handle strip: 0.dp of padding above the buttons, 6.dp below, and the buttons effectively jammed onto the tab bar. The status line above had no gap either, so the sheet's rounded top edge read as clipping it. - Derive OperateDrawerPeekHeight from its parts (handle + top gap + the shared OperatePrimaryRowHeight + bottom padding) instead of hard-coding it, so the peek can neither clip the primary row nor reveal the row below it when any of those change. Row spacing is now larger than the bottom padding, which is what keeps the next row hidden; both invariants are unit-tested. - 10.dp above the primary row, 14.dp below it (was 0 / 6), and a 10.dp gap between the docked status line and the sheet's top edge. - Shrink the handle strip 42 -> 36.dp and stop the "More controls" hint bobbing. A caption drifting up and down directly above fixed buttons reads as uneven spacing, not as an affordance; the up-chevron already says "pulls up". Measured on an emulator (1080x2400, xhdpi): clearance under the Call CQ row goes from 6.1.dp to 14.1.dp. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Lbd9brnZDMCzipgsu9n8Xg
The Map redesign already loads us_states.json and strokes it, but behind a `scale >= 3` zoom gate. That gate hides the layer in exactly the view a US operator sees: with a grid set and no decodes yet, the map auto-frames the operator at scale ~2.9 — a hair under the threshold — so the state lines never appear. Zoom was the wrong measure anyway. The map COVERs its card, so the same `scale` means very different detail in the two projections: the standard map spends ~6.3 px per degree of longitude at zoom 1 (the lower 48 are ~370 px across — their borders read cleanly), while the azimuthal disc squeezes the whole globe into the card at ~2.8 px per degree, where state outlines really are noise. Gate on that density instead, and fade rather than snap so a pinch doesn't pop a whole layer into existence: stateBorderAlpha() ramps 0 -> 1 between 3 and 6 longitude px per degree, fed by EquirectViewport.pxPerLonDegree for the standard map and azimuthalPxPerLonDegree() for the disc. Net effect: state borders are on throughout the standard projection, still off on the azimuthal world view, and the stroke stays a shade under the national borders so the basemap hierarchy holds. Verified on an emulator: state lines visible at the standard map's default zoom (screenshot in the PR). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Lbd9brnZDMCzipgsu9n8Xg
The clock-sync pill showed the raw arithmetic MEAN of a slot's decode DTs, re-posted on every decode pass. On a rig working stations perfectly it would announce "+2.5 s" and turn red, which reads as an unstable app. Three separate reasons it did that: - A mean has no defence against one bad decode. DT is per-station: a station transmitting late, or a marginal decode placed a couple of seconds off, carries its own large DT. Three good decodes at +0.1 s plus one at +7 s average to +1.8 s — a red "clock off" verdict produced entirely by somebody else's clock. - Every pass re-posted. A slot is decoded several times (fast, then deep and late passes that dig out the weakest signals — exactly the ones with the sloppiest DT), so the number visibly jumped several times per slot. - No slot-to-slot smoothing at all. Meanwhile the estimator that actually corrects the clock (ClockSelfSync) has always been robust — median, MAD rejection, multi-slot confirmation — so the display and the correction openly disagreed: the pill flashed "clock off" while auto-sync, reading the same decodes, correctly did nothing. New DecodeDtDisplay puts the display on the same footing: MAD-rejected median per slot, passes accumulating into their slot instead of replacing the reading, the median of the last 3 slots, and a 0.05 s publish deadband so the last digit stops twitching. A deliberate clock change (manual, GPS/NTP or self-sync) drops the window so the reading re-acquires instead of averaging across a correction. A real, sustained error is still followed within ~2 slots. Also logs "DT: raw=... shown=... n=..." once per slot, so a debug.log shows directly whether a wandering raw mean is being smoothed or a clock is genuinely drifting. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Lbd9brnZDMCzipgsu9n8Xg
Even out the collapsed controls drawer's spacing
Draw US state borders on the standard map view
Stop the DT indicator jumping on a well-synced rig
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## staging #916 +/- ##
=============================================
+ Coverage 41.50% 41.51% +0.01%
Complexity 252 252
=============================================
Files 280 280
Lines 33973 33993 +20
Branches 4033 4034 +1
=============================================
+ Hits 14100 14112 +12
- Misses 19584 19592 +8
Partials 289 289
Flags with carried forward coverage won't be shown. Click here to find out more.
🚀 New features to boost your workflow:
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Promotes the current
devintostaging— three UI/behaviour fixes merged since the last promotion (#913, #914, #915).What's in it
Collapsed controls drawer spacing (Even out the collapsed controls drawer's spacing #913) — the peek's
120.dpheight was a flat constant that left 0 dp above the 72 dp Call CQ / Hunt row and 6 dp below it, so the buttons read as jammed onto the tab bar and the docked status line looked clipped by the sheet's rounded top edge. The peek height is now derived from its parts (handle + gap + the shared primary-row height + bottom padding → 10 dp above, 14 dp below, 10 dp under the status line), with unit-tested invariants so it can't clip the primary row or leak the row below. Handle strip 42 → 36 dp, and the "More controls" hint no longer bobs — a caption drifting against fixed buttons reads as uneven spacing rather than as an affordance.US state borders on the standard map (Draw US state borders on the standard map view #914) — the state layer existed but sat behind a
scale >= 3zoom gate, and the map auto-frames a grid-holding operator at ~2.9, a hair under it, so the lines never appeared. Zoom was the wrong measure anyway: the map COVERs its card, so the samescalemeans very different detail in the flat map vs the azimuthal disc. Now gated on basemap density (longitude px per degree) with a 3 → 6 px/deg fade: on throughout the standard projection, still off on the azimuthal world disc, fading rather than snapping into existence on a pinch.DT indicator stability (Stop the DT indicator jumping on a well-synced rig #915) — the clock-sync pill showed the raw per-slot mean decode DT, re-posted on every decode pass, so one station transmitting late (or a marginal deep-pass decode) could make a rig that was working stations perfectly announce "+2.5 s" in red.
ClockSelfSync, which actually corrects the clock, has always used a median + MAD rejection + multi-slot confirmation, so the display and the correction openly disagreed. NewDecodeDtDisplayputs the display on the same footing: MAD-rejected median per slot, passes accumulating into their slot instead of replacing the reading, median of the last 3 slots, a 0.05 s publish deadband, and a window reset when the clock is deliberately corrected. A genuine drift is still followed within ~2 slots.Test plan
stagingdebug.log— the newDT: raw=… shown=… n=…line per slot shows whether a wandering raw mean is being smoothed or a clock is genuinely drifting. This is the one change with no live-signal verification; the emulator decodes nothing and no rig was attached.🤖 Generated with Claude Code
https://claude.ai/code/session_01Lbd9brnZDMCzipgsu9n8Xg