The 768m heap cap broke GitHub Actions (KSP OutOfMemoryError: Metaspace) because CI runners read the same gradle.properties. Restore -Xmx6g for CI; the 2GB local server should pass a one-off -Dorg.gradle.jvmargs instead.
Next release: versionCode 445 -> 446, versionName 4.4.5 -> 4.4.6. Adds the Roaming page (QTH定位器-style 3x3 Maidenhead grid) between Match and Settings, credits BG7NTA & the original author in the thanks title, and caps local Gradle heap for the 2GB build server.
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.
The settings outro title now reads 'BG7NTA & the original author would like to thank' (en) / 'BG7NTA 与原作者感谢' (zh) / 'BG7NTA ve orijinal yazar teşekkür eder' (tr), alongside the BA7OPF/BG7NTA entries added to the credits list.
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
Merge upstream commit 2298d8ee 'feat: add mutual radar overlay arrows'. Clean auto-merge: upstream changed RadarView.kt arrow drawing, our sweep optimization stayed intact. No conflicts.
First v4.4.5 was built from c3444aed; this release rebuilds from the merged tree.
Next release: versionCode 444 -> 445, versionName 4.4.4 -> 4.4.5, following the upstream scheme without suffix. Includes the station panel layout fix, 5-decimal coordinates, and credits additions already merged on main.
With 5-decimal coordinates the Lat/Lon/Qth row overflowed, pushing the QTH value to a wrapped line. Split into two rows: Lat + Lon on the first line, Qth on the second.
Also append BA7OPF (pass matching feature) and BG7NTA to the credits list in en/zh/tr string resources.
Conditionally applies the BG7NTA signingConfig when keystore.properties exists (gitignored). CI signs via apksigner with GitHub Secrets, so this only affects local builds.
App display name changed from Look4Sat to Look4Sat Pro (fork identity). Version follows upstream scheme without suffix: 4.4.3 -> 4.4.4, versionCode 443 -> 444.
Station lat/lon was stored rounded to 4 decimals (~11 m). Bump to 5 decimals (~1.1 m), slightly better than GPS hardware accuracy without showing noise.
- SettingsRepo.setStationPosition: round(4) -> round(5)
- QthConverter.qthToPosition: round(4) -> round(6) so 10-char locators fully roundtrip
- Update QthConverterTest expected values to 6-decimal precision
The official app signs com.rtbishop.look4sat with its own certificate. A fork sharing that applicationId cannot be installed over the official build (signature mismatch) and users saw overwrite/install failures.
- Add applicationId version catalog entry; namespace stays com.rtbishop.look4sat so source imports are untouched, applicationId becomes com.rtbishop.look4sat.bg7nta.
- Update PROPERTY_SATELLITE_DATA_OPTIMIZED meta-data to the fork id.
- Document the fork-applicationId requirement in the version catalog.
workflow_dispatch defaults TAG_NAME to the branch name (main), which breaks gh release create. Accept an optional tag_name input and fall back to github.ref_name for tag-triggered runs.
Remove Google Play upload (needs SERVICE_ACCOUNT_JSON we don't have) and AAB signing. Use built-in GITHUB_TOKEN instead of RELEASE_TOKEN secret. Pin actions to stable versions (checkout@v4, setup-java@v4, setup-gradle@v4). Builds assembleRelease, signs APK via apksigner with KEY_STORE secrets, creates GitHub release with APK. Triggered by v** tags.
Port the 8-char (4-pair) Maidenhead grid algorithm from the
"QTH定位器 2.0" app (com.us1pm.gridsquarelocator) into QthConverter
so locator precision matches common grid tools instead of being
truncated to 6 chars.
Previously positionToQth() emitted only 6-char locators and
qthToPosition() discarded everything past the 6th character via
take(6), losing the finer 30" x 15" resolution carried by 8-char
grid squares.
What changed:
- positionToQth(lat, lon, precision = 8) now emits 8-char locators
by default; precision = 6 / 10 available for backwards
compatibility and maximum resolution (1.25" x 0.625").
- qthToPosition() parses 6/8/10-char locators and returns the center
of the finest encoded cell (30" x 15" for 8-char, 1.25" x 0.625"
for 10-char) instead of the 6-char cell center.
- Locator validation regex tightened: 6/8/10 chars accepted,
4-char strings like "JN58" are now rejected as invalid.
- Boundary clamping added so lat = 90 / lon = 180 no longer overflow
the A-R / 0-9 / a-x alphabet (previously produced invalid chars).
Also fixed a pre-existing compile error in RadarView.kt: a delegated
property was assigned after declaration ("by" on an already declared
val). Converted the sweep angle to an if/else expression.
Verification:
- QthConverterTest extended to 5 test cases covering 6/8/10-char
roundtrips, invalid input, boundary coordinates and roundtrip
stability.
- Cross-checked against a Python reference model of the decompiled
APK algorithm: 20k random roundtrips at 8 and 10 chars, 0 failures.
- :core:domain:test green; :app:compileDebugKotlin passes.
- Radar pager now defaults to Transceivers tab (initialPage=0)
- Transceivers don't auto-expand (selectedUuid defaults to null)
- Mutual page default minElev changed from settings value to 0.0
- Display filters out portions below the minElev threshold
Elevation curve chart and mutual radar plot now only show the portion
where both stations are above their respective minimum elevation.
The curves are filtered at display time, the pass search still uses
the 0° horizon boundary for consistency with the Passes page.
Removed the automatic posA=stationPos override. Added a button
'当前精确位置' below the grid input that fills in the exact station
position from settings (lat/lon + grid). Users can now freely edit
the position fields and use the button when they want the exact
position.
Root cause: gridToLatLon used (lon+180)%360-180 and (lat+90)%180-90
for normalization, treating grid values as from prime meridian/equator
when they are actually from IDL/South Pole. A 6-char grid like OL62AA
(112°E, 22°N) was returning -67.96°, -67.98° (Atlantic Ocean).
Also added bidirectional sync: entering a grid auto-fills lat/lon,
entering lat/lon auto-fills the 6-char grid, so users can verify.
getFullPosition and getElevation both call calculateObs internally,
but the elevation values reported by the user (-60°) suggest they may
differ. Now elevation is computed via getElevation (same as getLeoPass),
while getFullPosition is only used for azimuth.
The passes list AOS/LOS times can be stale or computed for a different
station position, causing elevation curves to show -60° at pass start.
The independent search (refineEdge + sampleMutualPass) always computes
elevation for the actual positions. With posA now set to the exact
station position, the pass times should match the main page.
The refineEdge function re-computed boundaries using the grid-center
position (posA), which differs from the exact station position that
the passes list was computed with. This caused all passes to be
filtered out. Now we trust the pass list's AOS/LOS times directly
and just sample the elevation/azimuth curves.
If the main pass list (satelliteRepo.passes) is empty or the common
window check filters everything out, fall back to the independent
search algorithm so the user always gets results.
findMutualPasses was computing passes independently with its own
search algorithm, which produced different results from the main
page's getLeoPass. Now it directly reuses satelliteRepo.passes,
which is the same list shown on the main passes page. The only
additional computation is refining the common window for station B
and sampling elevation/azimuth curves.
Root cause: mutual search used the min-elevation threshold (default 10°)
as the AOS/LOS boundary, while the main radar uses the 0° horizon
(getLeoPass). With identical stations the windows therefore never
matched. Also, an in-progress pass was counted as a new one.
- AOS/LOS boundaries now at the 0° horizon (refined to 1s)
- Skip in-progress mutual windows at search start (like getLeoPass)
- Filter requires BOTH stations to reach their min elevation
- Default min elevation follows the main radar passes setting
- Elevation ring labels were inverted (90/60/30 from outer to inner);
now 30/60/90 correctly from outer ring to center
- Split track paths at the 0/360° azimuth wrap in both MutualRadarView
and main RadarView, so passes crossing due north no longer draw a
line straight across the plot
Track line is no longer time-limited (always shows the whole mutual
arc). The station-B live position is drawn separately on the line
using the same pulsing-dot mode as the local station.
- ElevationCurveChart is now controlled (progress + onProgressChange)
- MutualRadarView shows shared time-cursor positions for both stations
and is draggable/tappable to move the cursor
- MutualPassCard owns a single dragProgress feeding both charts, so
dragging either chart moves the other in sync
- MutualViewModel is Activity-scoped, so returning from Radar keeps query results
- TrackSampleData gains time field for live cut-off
- Removed standalone mutual card from radar page
- RadarViewCompose draws optional dashed station-B track + current dot
- RadarScreen builds trackB from mutual data up to current time
- MutualPass now carries TrackSample (azimuth/elevation for both stations)
- New MutualRadarView: polar plot with elevation rings (30/60/90),
cardinal spokes, solid A-track vs dashed B-track, AOS/LOS markers
- Shown in expanded mutual pass card and radar page mutual card
Absolute epoch millis (~1.7e12) exceeds Float precision (ULP ~131s
at that magnitude), so 5s-spaced samples collapsed onto identical
x coordinates, producing a stepped/jagged curve. Use Long relative
time deltas (sample - start) before converting to Float.
- Fix cubic Bezier control points (proper Catmull-Rom to Bezier conversion)
- Refine AOS/LOS more robustly by walking from edge into the pass window
- Still 5s sampling for smooth curves
- Reduce sample interval from 10s to 5s for smoother curves
- Add refineEdge() to find exact AOS/LOS at 1s resolution
- Curve now starts/ends at the correct horizon-crossing points
- Pre-fill station A grid from settingsRepo.stationPosition
- Find ALL mutual passes per satellite (not just first)
- Fix Color.hashCode() -> Color.toArgb() for native canvas paint
- Use 2min gap between pass searches to avoid duplicates
Bug: OrbitalObject.getElevation() returns radians, but
MutualViewModel was comparing it directly with degree values
from the slider (10-90), causing no matches to be found.
Fix: add elevationDeg() helper that converts to degrees.
Also add Maidenhead grid square input support (4/6/8 chars)
alongside existing lat/lon fields. Grid takes priority when
filled.
Port satlover.de dual-station pass matching feature:
- New feature/mutual module with mutual pass data model
- MutualViewModel: compute overlapping passes for two stations
- ElevationCurveChart: Canvas-based dual elevation curve with drag
- MutualScreen: input form + results with expandable cards
- Navigation: add Mutual tab to bottom navigation bar
- i18n: add Chinese/English strings for new feature
Extract dateFormat() helper to ensure computeSunTimes and
groupPasses use identical date format, fixing sunrise/sunset
display bug.
For Chinese locale: "2026年8月1日 星期六"
For English locale: "Sat, 01 Aug 2026"
DateFormat.FULL may cause inconsistent date labels between
computeSunTimes and groupPasses, leading to missing or
wrong sunrise/sunset times. Revert to the original pattern
"EEE, dd MMM yyyy" but with Locale.getDefault() so day
names follow the system language.
mckero
atsunatsu