Stage 1 is implemented and manually checked. What was worth keeping is in the commit messages, the spec and docs/PROJECT_CONTEXT.md.
Written-by: Claude Fable 5.2
rerunSetupWizard set the one-shot request before it tried to obtain the switch, and cleared it when the switch was refused. A second confirmed call during the first one's window teardown therefore removed the first call's request: the session ended and the profile restarted without the wizard. The request is now set only by the call that owns the switch.
Written-by: Claude Fable 5.2
Finish now requires every mandatory answer, since a summary without locator, band and category could not be saved. Adds an optional step for antenna and power values that feed the path analysis.
Written-by: Claude Fable 5.2
Describes a first-start assistant for unconfigured operator profiles: what it asks, how it covers multi-operator and multi-station setups, and why it writes preferences.xml before the runtime is built instead of going through the settings draft.
Written-by: Claude Fable 5.2
Written-by: Codex (GPT-6.2 Sol)
Written-by: Gemini 4 Hydrogen (Antigravity)
Written-by: DeepSeek V4 Pro
Apply beacon checkbox changes at runtime and keep the aggregate status chip synchronized.
Improve explicit station-map navigation, spatial filter overlays, overlay legend, selected-path contrast, and the documented map screenshot.
* test: print all threads when a test times out
The rare CI hang in SelectedStationMessageUiVisualRenderingTest blocks on a
monitor in SnapshotStateObserver.observeReads (line 643 is the monitorenter
of observedScopeMapsLock in Compose 1.8.2). The timed-out thread's stack
shows where it waits but not which thread holds the lock. With the thread
dump enabled, JUnit prints every thread to the test's standard output, which
ends up in the XML report of the test-results artifact.
Claude-Session: https://claude.ai/code/session_01FJBfPYm6rCH4zD6T7bcC7Y
* test: render every ImageComposeScene on the AWT event thread
The rare CI hang is a lock-order deadlock in Compose Multiplatform 1.8.2,
shown by the thread dump of run 40299943329:
- the test thread renders the scene and holds the FlushCoroutineDispatcher
runLock while LazyListState.scrollToItem measures and wants the
SnapshotStateObserver's observedScopeMapsLock;
- the AWT event thread sends the global snapshot notifications and holds
observedScopeMapsLock in drainChanges while a re-read derived state wakes
the scene's snapshotFlow collector, whose dispatch wants runLock.
The application renders on the event thread, where both monitors are simply
re-entered. The tests now do the same: onEventThread builds, renders and
closes each scene there and rethrows what the block throws.
Claude-Session: https://claude.ai/code/session_01FJBfPYm6rCH4zD6T7bcC7Y
Strip HTML tags and XML comments until the text no longer changes, so that
nested input cannot reassemble a tag or comment after a single pass
(js/incomplete-multi-character-sanitization in website/.eleventy.js and
website/scripts/validate-version-info.js).
Limit the GITHUB_TOKEN of the PR compile check and the test version-XML
workflow to read access (actions/missing-workflow-permissions).
Claude-Session: https://claude.ai/code/session_01FJBfPYm6rCH4zD6T7bcC7Y
The kst4contest and kst4contest-git packages built jpackage's launcher
with Arch's openjdk, which links libapplauncher.so against the shared
libstdc++. That build aborts now and then on exit with "pure virtual
method called" in a static destructor (SIGABRT, core dump), the failure
Debian describes for non-static libstdc++ builds. Temurin links
libstdc++ statically, like the CI-built packages.
Both PKGBUILDs now fetch a pinned Temurin 21.0.13+9 through
source_x86_64, so aur-publish.yml's rewrite of the sha256sums= line
leaves it alone, and point the Gradle toolchain at it alone.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Nightly #303 named the hanging test, SelectedStationMessageUiVisual-
RenderingTest at scene.render(), but the short exception format showed
only that line, not the call that blocked below it. The job timeout
cancels the run, and the test reports were kept only on failure().
Failed tests now print their full stack, causes included, and the
reports are uploaded on cancelled() as well.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Nightly #302 and the PR check for #147 hung in :app-desktop:test for
over an hour, and the log named only the task. Thirteen reruns of the
same commit passed, so the hang is rare and still unidentified.
Every test now gets a two-minute JUnit timeout in a separate thread, so
even an uninterruptible hang fails with the test's name and the stack
where it stood. The test jobs are capped at 20 minutes for a hang
outside any test.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The cluster and QSO monitor split its two tables at half the height, fixed:
the Compose port had left the JavaFX SplitPane out until a splitter existed.
SplitterPane exists now, so the monitor uses it and reads and writes the same
profile value the SplitPane did (GUIpnl_directedMSGWin_dividerpositionDefault,
default 0.8), so a profile gets its old position back. Each heading now sits
in the pane of its table, and the "DX cluster" heading is translated
(monitor.dxCluster).
Fixes#144.
Claude-Session: https://claude.ai/code/session_01AaorTymXfkXPm8yb4Btva7
Co-authored-by: Claude <noreply@anthropic.com>
Material3's DropdownMenu clamps every popup to 48dp below the window's
top edge (MenuVerticalMargin in its position provider); the margin is
applied after the offset parameter and the provider is not replaceable,
so the in-window menu bar's menus all opened with a visible gap hanging
below the bar. Replace them with a plain focusable Popup in MenuDefaults
styling and an own position provider that puts the menu directly under
its anchor, flipping above only when there is no room below. The anchor
box takes the full bar height so the menu starts at the bar's bottom
edge, not at the centred title text.
Claude-Session: https://claude.ai/code/session_01BegdsGSNNxKR6VPNPNvfoc
Co-authored-by: Claude <noreply@anthropic.com>
* Move remaining references to the HamRadioOnline organisation
The repository moved from praktimarc/kst4contest to
HamRadioOnline/kst4contest. Document the move in the German and English
changelog, point manuals, README, website, AUR packaging and the
version-feed tooling at the new address, and keep accepting pre-move
release URLs in the update-feed validator because published feeds still
carry the old owner. Personal identities (Ko-Fi account, contact
e-mail) stay unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WFGQ7wkSPo21itLMHPi8Yw
* Lowercase the owner in the generated Flatpak Pages URL
GITHUB_REPOSITORY_OWNER is now the mixed-case HamRadioOnline, and the
next build would write that spelling into the published flatpakrepo and
flatpakref files. Hostnames are case-insensitive, but the published
metadata should match the documented lowercase address.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WFGQ7wkSPo21itLMHPi8Yw
* Build the Flatpak against org.freedesktop.Platform 25.08
Runtime branch 24.08 is end-of-life and no longer receives security
updates, so every flatpak update now prints an EOL warning. 25.08 is
the current supported branch on Flathub. The app bundles its own Java
runtime, so the platform bump only changes the base libraries.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WFGQ7wkSPo21itLMHPi8Yw
---------
Co-authored-by: Claude <noreply@anthropic.com>
Second slice of making MessageBusManagementThread testable. The CH (live)
and CR (history) chat-frame branches of processRXMessage23001 were near
copies; they are now one processChatFrame(fields, live), with behaviour
unchanged.
- InboundChatFrame, ChatRoute and classifyRoute: where a message belongs
(public, to me, from me, between others), pure.
- ChatMessageEffects and decideChatFrameEffects: what a message triggers
(sounds, //BELL, //VER reply, auto-answer, directional tick, in-angle
flag, DX cluster spot), pure. History frames trigger nothing but the flag.
- The four live/history differences are kept on purpose: frequency
detection before routing (live) or after publishing (history), effects
for messages to me, tick and spot between others, AirScout info on the
SERVER sender.
- MessageBusManagementThreadChatFrameTest pins both branches through the
real thread, including the frequency order on directed frames and the
spot entry; ChatFrameDecisionTest covers the pure functions.
- Spec: docs/superpowers/specs/2026-10-16-message-bus-second-slice-design.md
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Spec: SO2R with two radios in one instance (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Plan: SO2R implementation (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Move the rotator command rules into RotatorControl (#46)
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* Store the SO2R switch, the second Win-Test station and rotator 2 (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Route frequency reports to radio 1 or 2 (#46)
Add the RadioSlot enum (FIRST, SECOND) with its QRG and QTF targets, and
the pure RadioSlotRouter that decides per RadioInfo packet and per
Win-Test station which radio a report belongs to.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Send RadioInfo of radio 2 to SECONDQRG when SO2R is on (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Send Win-Test STATUS of the second station to SECONDQRG when SO2R is on (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Run a second PSTRotator for radio 2 with SO2R (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Start rotator 2 only while an ON4KST session runs (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Settings for SO2R, the second Win-Test station and rotator 2 (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Turn antenna 2 and show QTF2 with SO2R (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Make the SO2R switches observable so Apply shows them (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Document SO2R (#46)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Refuse clashing PSTRotator ports in both directions (#46)
Writing either port now checks same-or-adjacent against the drafted value of
the other; rotor 1 is only checked while SO2R and rotor 2 are on. Also fixes
a misplaced Javadoc, stale WinTest parseStatus Javadoc, docs wording and
removes an unused import and dead comment.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Remove the executed SO2R plan (#46)
The plan's content lives on in the commits, the spec and PROJECT_CONTEXT.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Open secondary windows on the main window's screen (#59)
Every Compose window but the main one went to the platform's default
position, which on a multi-monitor desk is the primary screen - the
station map could open on a monitor that was switched off. They now open
centred on the screen the main window is on, with their size clamped to
that screen's usable area (insets considered). A window the operator
moved during the session reopens there, as long as its title bar can
still be grabbed on a connected screen; otherwise it falls back to the
main window's screen. The position lives only in the window's host and
is never persisted. ComposeAlert dialogs and the profile picker are
centred the same way once the main window exists.
The rule is a pure function (WindowPlacement) with 18 unit tests; the
AWT side (WindowPlacementOnScreen) only reads the screens and the main
window's bounds on the event thread. The clamped opening size of a
secondary window is not written back, so the stored size survives a
smaller screen.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Track window moves reliably and keep remembered windows on screen (#59)
Review fixes for the window placement:
- Once a window has left the tolerance around its opening position,
every later position is remembered. Before, the anchor was never
re-set, so dragging away and back to near the opening spot left the
stale far position remembered.
- A window the host did not place (opened before the main window
exists) only trusts its anchor once two consecutive reports lie
within the tolerance, so a provisional position followed by the
window manager's placement jump is no longer taken as the operator's
move.
- A remembered position is shifted the least distance that puts the
(clamped) window wholly inside the usable area of the screen holding
its title bar, as the KDoc promised.
The move rule moves into a pure MoveTracker with 5 tests; WindowPlacement
gains keptWithin and 2 more tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Keep window positions across restarts as an opt-in; fit windows to small screens (#59)
A new GUI tab option, off by default, keeps the position the operator last
moved each window to, the main window included. It is stored per window
with the bounds of the screen it was on, written by the layout autosave,
and restored only when a screen with exactly those bounds is connected
again; otherwise the session rule applies. Only operator moves are
stored, never window manager nudges, so a restored window cannot drift.
An open window that rests on a screen too small for it is now cut down
to that screen's usable area, and the station map's 900x600 minimum is
capped to the smallest connected screen, so 640x480 monitors stay usable.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Never store a size or position the host fitted a window to (#59)
The cut-down to a small screen was written back as the stored size
(main window via onResized, others via the tracker) and counted as an
operator move. The host now records its own fit: that size is not
written back, and that position re-anchors the move tracker instead of
being stored. A main window opened at a stored position skips its
opening size like the other windows, since it may be clamped too.
The opt-in gate moves into PreferencesWindowPositionMemory, with tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Match the fit echo with tolerance and consume it once (#59)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>