RoamingScreen carried ~170 lines of decompiled range-lookup tables
(encodeLon/encodeLat) that re-implemented exactly what core:domain's
positionToQth already does. A probe calling both over 16,471 sampled
coordinates (every 2 degrees across the full globe) found byte-identical
8-char locators, so the tables were pure duplication - two implementations of
the same Maidenhead encoding that had to be kept in sync (the earlier boundary
fix had to be applied twice).
Replace them with positionToQth and split its standard-ordered output
(lonField latField lonSquare latSquare lonSub latSub lonSubsub latSubsub) back
into the per-axis segments the UI consumes: the 3x3 ring (first 4 chars),
markerLeft (lon subsquare), markerTop (lat subsquare). Out-of-range input
keeps the old blank-segment behaviour: positionToQth returns null, the locator
becomes spaces, qthNeighbors returns an empty ring, and the marker lookups
fall back to 0.
Net -160 lines. All existing RoamingState tests and the new equivalence probe
pass; :feature:roaming compiles.
The decompiled per-edge branches computing the surrounding nine squares had two
independent defects.
1. Field letters stepped past the alphabet
Every branch moved a field with raw character arithmetic (`str[0] - 1`,
`str5[0] + 1`) and Maidenhead fields only run A..R, so coordinates near the
edges of the world produced squares outside the alphabet:
(-89.9, -179.9) -> [@A91, AA01, AA11, @A90, AA00, @A10, @@99, A@09, A@19]
( 89.9, 179.9) -> [RS80, RS90, SS00, RR89, RR99, SR09, RR88, RR98, SR08]
2. Some moved cells kept the old field letter
The north-edge branch advanced the latitude field for the top-centre cell only,
leaving the two top corners in the previous field:
centre AA19 -> ported [AA00, AB10, AA20, ...]
correct [AB00, AB10, AB20, ...]
Cross-checked against the shared qthNeighbors helper, which is already covered
by QthConverterTest including the AA00 and RR99 wrap cases:
before: 64,800 sampled coordinates, 6,480 disagreed (all with centre square
digits 00 or x9, i.e. the north edge and the 00 corner)
after: 64,800 sampled coordinates, 0 disagree
The ring is plain Maidenhead arithmetic with no QTH-Locator-specific behaviour,
so call qthNeighbors instead of keeping a second, wrong implementation. The
now-unreferenced buildGrids branches are removed (grep confirmed the definition
was the only remaining occurrence). The existing OL42 reference grid and the
four ported edge-case tests still pass unchanged.
Reverting the fix fails both new regression tests; with it
:feature:roaming:testDebugUnitTest and :core:domain:test are green.
The QTH Locator port keeps the decompiled range tables, which close both
adjacent cells (`-20.0..0.0` then `0.0..20.0`). Kotlin's `when` takes the first
match, so any coordinate landing exactly on a field, square or subsquare
boundary was attributed to the previous cell:
(0, 0) II99xx99 should be JJ00aa00
(1, 1) JJ00lx99 should be JJ01ma00
(22, 108) OL31xx99 should be OL42aa00
(22.5, 108.5) OL42fl99 should be OL42gm00
(22.25, 108.25) OL42cf99 should be OL42dg00
At the field level the locator is wrong by a whole 20 deg x 10 deg field, and
the 3x3 neighbour grid plus the red position marker are derived from the same
characters, so the whole Roaming screen pointed at the wrong square.
Cross-checking the port against core/domain positionToQth over the grid:
before: 65,341 sampled points, 4 agreed
after: 65,341 sampled points, all agree
The independent converter was confirmed correct first: it reproduces the
user-verified reference sample OL42ih45, and hand-computing lon=-179.75
(0.25 deg into the field, x12 -> subsquare index 3 = 'd') and lat=-90
(subsquare 'a', extended digit 0) matches it rather than the port.
Rather than rewriting the faithful lookup tables, nudge the input by 1e-10 so
the closed ranges behave like the standard half-open [low, high) cells, keeping
+90/+180 inside the final R cell. Seven real-world city samples and all existing
ported-behaviour tests, including (90, 180) -> RR99xx99, are unchanged.
Regression tests added for the boundary cases and for cross-implementation
agreement. Reverting the fix fails both; with the fix
:feature:roaming:testDebugUnitTest is green.
All Chinese comments (//, /* */, KDoc) across core/app/feature/build-
logic translated to English (550 lines, 73 files after FT8 rollback).
Code logic untouched - comment text only. Verified: all modules
compileDebugKotlin BUILD SUCCESSFUL.
Three detail fixes plus branding, per user review:
1. Night mode visibility (roaming page): hardcoded reference blues
(#01DDFF/#0BACF1) collapse to black under the red ColorMatrix
filter (R channel only), making the page unreadable. Replace ALL
colors with MaterialTheme.colorScheme (background/surface/
surfaceVariant/onSurface/onSurfaceVariant/primary/error). GPS
status dots become themed circles (primary/error) so they survive
the red filter; red marker keeps the original pnt drawable.
2. Settings cards unclamped:
- OtherCard: fixed height(268.dp) squeezed the last toggle row
(spacing 42/42/42/15px on device); drop the fixed height.
- CardCredits: same 268dp + SpaceBetween overflowed, gluing the
last thanks entry to the warranty line; drop fixed height, use
spacedBy(8.dp) and insert 24dp before the warranty text.
3. Branding:
- Settings top title: "Look4Sat v%s" -> "Look4Sat Pro v%s" (en/tr);
title Text no longer marquees and wraps instead (heightIn(min=48)).
- APK asset name: look4sat-<ver>.apk -> Look4Sat-Pro-<ver>.apk.
- What's-new dialog (pass_whatsnew_message) rewritten in all three
locales (en/zh/tr) with this release's changes; zh gets its own
localized title "Look4Sat Pro 更新内容".
Verified: roaming/settings/passes/app compile, 11 unit tests pass.
The user spotted two white "beams" (20dp gaps) between the center
cell and the side cells. Root cause: the reference app's RelativeLayout
ignores the fixed 60dp width when a child has BOTH a left rule
(alignParentLeft) and a right rule (toLeftOf=center cell) - the width
is stretched to right-rule minus left-rule, i.e. 80dp on a 360dp
screen. The three cells therefore sit flush with only 2dp margins
between them, no gaps.
Port that behavior: side cells 60dp -> 80dp (6 places), center cell
stays 200dp centered. Verified against the reference screenshot pixel
measurements (side cells ~76.5dp incl. margins, center ~198dp).
Verified: feature + app compile, 11 unit tests pass. Marker lookup,
grid math and GPS logic untouched.
Two remaining proportion issues on the user's device:
1. Right-side gap: the three grid rows used a continuous Row
(60+200+60dp), leaving ~84px of blank space on the right of the
screen. The reference app uses a RelativeLayout where the right
column is pinned to the screen edge and the 20dp gaps sit on both
sides of the centered middle column. Convert each row to a Box:
left cell align(CenterStart), middle cell align(Center), right
cell align(CenterEnd) - pixel-identical to the reference.
2. Footer legibility: the credit line ("制作：US1PM 汉化：BA7LCE")
had a fixed 15dp height with no bottom margin, so it sat directly
against the navigation bar and part of the text was hard to read.
Drop the fixed height and add 10dp bottom padding so the text
renders fully with breathing room above the nav bar.
Verified: feature + app compile, 11 unit tests pass. Marker lookup,
grid math and GPS logic untouched.
The ported page rendered edge-to-edge: the GPS bar started right at
the top of the screen and the footer sat against the navigation bar,
so system UI overlapped the content (user: "顶头"). The reference app
is not edge-to-edge and keeps its content inside the safe area.
Add windowInsetsPadding(WindowInsets.systemBars) on the page root
column, shrinking the top and bottom by the status/navigation bar
height exactly as the user requested ("上方和下方往里面缩一点点").
Grid columns (60/200/60dp), rows, marker lookup and all logic are
untouched - verified pixel-identical column ratios vs the reference
screenshot (147:395:147 vs 146:395:148).
Verified: feature + app compile, 11 unit tests pass.
The roaming page was repeatedly rebuilt by hand and the red marker
still rendered at the wrong spot on the user's device. Per the user's
explicit instruction the whole page is now a faithful, line-by-line
port of the reference app (QTH定位器 2.0, com.us1pm.gridsquarelocator)
with zero UI or logic changes.
UI (res/layout/main.xml, byte-verified via aapt2 dump):
- 25dp holo-blue GPS bar: "GPS" 14sp, green/red status dot 15dp
(original mipmaps copied as drawables), centered date, right time,
translucent-yellow "设置启用GPS" button that opens location settings
- Latitude/longitude rows: 16sp black labels, right-aligned decimal
values, DMS label format "纬度 22° 18' 50" N" exactly as reference
- 43sp bold black locator, centered, with progress spinner + notice
- 3x3 continuous grid: 60/200/60dp columns, middle row fixed 205dp,
edge cells #0BACF1, center cell 200x200dp holo-blue (#01DDFF as
shown on the user's device), center label 100x80dp 30sp bold white
(textColorHighlight) centered, bottom "制作：US1PM 汉化：BA7LCE"
- Red marker: original pnt.png (red square with white outline),
10x10dp, absolutely positioned by the reference lookup tables
(lon 3rd pair a..x -> leftMargin -2..190dp, lat 3rd pair a..x ->
topMargin 190..-2dp, screen-Y inverted)
Logic (MainActivity.java showLocation/checkEnabled/onResume):
- 8-char locator via the reference range-lookup tables
- 3x3 neighbor grid via parseInt3 five-branch logic incl. all four
corner-carry tables (00/09/99/90)
- Live GPS + network updates 10s/10m while the page is shown,
provider filtered to gps/network, checkEnabled three-state
(green dot / red dot + settings button) exactly like the reference
- Time = cached hour prefix + fix minutes; date "dd MMM yyyy"
Dropped only the Play-store ad banner (conflicts with GPL project).
Verified: 11 unit tests pass (locator, marker lookup, grid, DMS,
time, all edge branches); feature + app modules compile.
RoamingState.kt removed - state and math now live in RoamingScreen.kt.
The marker's vertical position was hardcoded (10dp below the label) — markerY never participated, so the dot could only move horizontally and could not reflect where the GPS fix sits inside the 4-char square.
Now the marker is positioned in both axes from the cell's top-left corner:
- x = markerX * cellWidth, y = markerY * cellHeight (screen Y)
- markerX/markerY come from the 3rd character pair via the reference lookup tables (lon a=-2..x=190 as leftMargin; lat a=190..x=-2 as topMargin, i.e. latitude inverted on screen)
- Verified numerically against the decompiled tables: a=0/1.0, i=0.333/0.667, x=0.958/0.042 — matching within 3% (the reference table is slightly non-uniform)
Label stays at upper-middle (28sp bold); marker is only drawn when a valid locator exists.
Two bugs from the last release:
1. The header clock was frozen: Date() was only evaluated during recomposition, and with no state changes the time never moved. Added a 1s LaunchedEffect ticker that updates a now-state, so the date/time text re-renders and actually advances — matching the reference app, which refreshes the clock on every location callback.
2. GPS indicator: replaced the 10-minute freshness check with a direct 'has a real fix' check (timestamp > 0 and coords non-zero). The page mirrors the station position (站位) from the shared StateFlow, so the GPS dot is green whenever the station has a fix — GPS shows exactly what the station GPS says, nothing more.
Reworks the Roaming grid to structurally match the reference app instead of a card-style panel:
- Removed the outer ElevatedCard, rounded cells, cell gaps and inner padding: the 3x3 grid is now one continuous table that fills the panel, cells connected edge-to-edge.
- Cells separated by 2dp divider lines that run the full width/height of each row/column, crossing at right angles like a real coordinate grid (the reference app's continuous separator lines).
- Cells are square-cornered (no rounded corners), background fills each cell fully.
- Column widths 21.4% : 56.2% : 21.4%, row heights 31.5% : 35.9% : 31.7% retained.
- Center cell: OL42 is larger (28sp bold) and placed at upper-middle; the red marker sits BELOW the text with a 10dp gap, horizontally offset by the 3rd-pair fraction — never overlapping the label, matching the reference 'text above, marker below' layout.
- Surrounding labels bumped to 16sp Medium (larger/stronger than before).
- Info header also switched from an ElevatedCard to a flat continuous block so the page reads as one continuous surface, like the reference.
The grid proportions, locator algorithm, marker mapping and boundary logic were already faithful; this change makes the visual structure faithful too.
Reverts the auto-update machinery after review — the page now simply mirrors the station position (站位) from the shared settingsRepo.stationPosition StateFlow, exactly what the Settings page shows:
- RoamingScreen: removed the LocationManager listeners, the 30s re-request loop, the provider filter and the location-disabled hint. No polling, no auto-updates; coordinates are whatever the station GPS says.
- Settings: removed the '漫游位置实时更新' toggle (stateOfRoamingLive, key, action, strings en/zh/tr) that caused the 'Other' card to overflow — the sixth unlabeled switch clipped past the card's rounded bottom edge was that row overflowing a fixed-height card. Card height back to 268.dp, five rows fit again.
Verified: core:domain tests, roaming/settings/app compile clean; zero references to RoamingLive remain.
Completes the port of the QTH定位器 app's location logic (read from the decompiled MainActivity):
1. Live location updates: a LocationListener registers GPS+NETWORK providers (10s / 10m, matching the reference onResume) while the Roaming page is visible and removes itself on dispose. Every fix is pushed through settingsRepo.setStationPosition, so the shared stationPosition StateFlow updates the Settings page and the map in lockstep — the page now refreshes in real time instead of only showing stale cached coordinates.
2. Provider filtering: only gps/network fixes are accepted, mirroring the reference showLocation() guard that rejects passive fixes.
3. Location-disabled hint: when GPS is off or permission is missing, the header shows a tappable '定位未开启,点击前往系统设置' row that opens ACTION_LOCATION_SOURCE_SETTINGS — the port of the reference btnLocationSettings button. The GPS status dot now has three states: fresh fix (primary), provider on but stale (error), provider off (outline).
4. Periodic recovery: a 30s re-request loop (honoring the 漫游位置实时更新 toggle) re-arms the location request after the chip falls idle.
Not ported (conflict, deliberate): the reference app's Play-store ad banner, 'New! Grid Square with map' promo and GP_IN preferences — commercial advertising does not belong in a GPL satellite tracker.
Addresses three review findings:
1. Coordinates now come straight from settingsRepo.stationPosition in the screen (collectAsStateWithLifecycle) — the exact same StateFlow the Settings page shows. Previously a separate ViewModel re-derived them, and it could lag behind the Settings page (user: '设置页更新了站位但漫游页死活不更新'). With the shared source the two pages can never disagree. RoamingViewModel removed; state derivation moved to RoamingState.fromPosition().
2. Red marker placement ported faithfully from the QTH定位器 app: it is driven by the 3rd character pair of the 8-char locator (the 'ih' in OL42ih45), mapped to a 0..1 fraction (lon a=west..x=east, lat inverted a=south..x=north), then scaled to the actual center-cell size. The grid now uses the reference proportions (columns 21.4/56.2/21.4, rows 31.5/35.9/31.7) and fills the screen, so the marker lands accurately on any device.
3. Workflow now uploads a versioned APK (look4sat-<version>.apk instead of look4sat.apk).
Also: Settings 'Other' card rows got vertical spacing (Arrangement.spacedBy) so the new roaming toggle is not glued to the night-mode row.
Rework the Roaming page after user review. It is now a faithful port of the QTH定位器 location panel, not a loose re-skin:
UI (matching the reference layout):
- GPS status dot (real: green when a fresh fix exists, outline when stale/missing) + date + time header
- Lat/Lon rows with DMS and 5-decimal display, big 8-char locator centered below
- 3x3 grid of neighboring 4-char squares with the reference proportions: center column ~2.6x wider (21.4% : 56.2% : 21.4%) and center row the tallest (31.5% : 35.9% : 31.7%); red position marker now placed at the fractional position of the fix inside the center cell (was fixed center)
- No oversized GPS button: live updating is now a Settings toggle '漫游位置实时更新' (stateOfRoamingLive, default on) that drives periodic location refresh
Style & night-mode safety:
- All colors come from MaterialTheme.colorScheme (surfaceVariant/secondaryContainer/error) — no hardcoded cyan/blue from the original app. The red night filter (ColorMatrix keeping only the R channel) blanked the old hardcoded palette; theme colors survive it.
- Chinese nav label '漫游' added to values-zh (was missing, showing English 'Roaming')
Settings:
- OtherSettings.stateOfRoamingLive persisted (default true), toggle row in Other card, height adjusted
Verified: feature:roaming, feature:settings, app compile clean.
Include feature:roaming/build.gradle.kts (missed from the module commit) and keep the reduced Gradle heap (-Xmx768m) so local light builds don't freeze the 2GB server.
Ports the QTH定位器 (com.us1pm.gridsquarelocator) location panel into Look4Sat as a new 'Roaming' page, restyled with the app's own look.
UI (top to bottom):
- Info header: GPS status dot, big 8-char locator, Lat/Lon rows with DMS + 5-decimal display, and a GPS 定位 button
- 3x3 grid panel: the current 4-char Maidenhead square (e.g. OL42) centered, surrounded by its 8 neighbors (OL33..OL51), with a red position marker in the center cell
Logic:
- QthConverter gains qthNeighbors(square) building the 3x3 grid with field/square carry at boundaries (AA00 wraps to RR99, IO91 crosses into J field), and qthToSquare(locator) extracting the 4-char square
- Verified against the decompiled app algorithm and the reference screenshot (OL42 grid matches exactly); 9 unit tests cover normal, boundary and field-wrap cases
Navigation:
- New bottom-nav item 'Roaming' between Match and Settings, with a crosshair icon
- New feature:roaming module (ViewModel + Compose screen) registered in the app
Build config:
- Lowered Gradle JVM heap from -Xmx6g to 768m: the 2GB build server froze on the old value; heavy release builds stay on GitHub Actions
mckero