* 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>
* Restore the connection line in the main window title
While online, the title showed the connection manager's diagnostic
"ON4KST session is authenticated and synchronized" instead of the line
the JavaFX window carried: categories, call, name, locator, users online,
stations shown and messages total. The title now builds that line from
the login preferences and the live station and message tables, so it
follows every new station, filter change and message. The diagnostic
stays where it belongs, in the status bar.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Title: typed categories, no empty category, review fixes
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Path analysis: own ground height ASL and highest point of the remote square
Two new settings in the path-analysis section of the Station tab:
- Own ground height ASL (stn_pathAnalysisOwnHeightAslMeters). Empty keeps
the elevation model value at the own locator centre; a value from -500 to
9000 m replaces the elevation of the first profile sample. The antenna
height AGL is still added on top. The field commits on Enter or focus
loss into the settings draft and reports a refused entry; it keeps the
operator's spelling while that names the drafted number, because the
draft is Compose state and would otherwise swap "25" for "25.0" under
the cursor right after Enter.
- Target station on the highest point of its locator square
(stn_pathAnalysisDxHighestPointInSquare, on by default). The remote square
is sampled on a 5 x 5 grid of cell centres in one batched request against
the Open-Meteo provider, the highest sample becomes the path end with its
position and elevation. LocatorHighestPointResolver caches the result per
locator for the session and backs off five minutes after a failure
(failure list pruned and bounded); the analysis then falls back to the
square centre and that result is not kept.
Both settings are drafted like the other path-analysis settings and take
effect with Apply/Close or Save; the reachability service reads the stored
values only.
The lookup runs on the reachability thread only. Building a request peeks
the resolver cache without blocking, so a cached square moves the endpoint
right away. The calculation key now includes the own height and the remote
ground elevation at full precision, and the switch, so changing either
recalculates.
Also assign the result of sanitizeEffectiveEarthRadiusFactor in the
PathAnalysisRequest constructor; it was called and dropped. No behaviour
change: the sanitizer is the identity for valid factors, and the analysis
service sanitizes again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Document own ground height and highest-point setting of the path analysis
Changelog v1.50 (New), configuration and features pages in English and
German, and the terrain section of PROJECT_CONTEXT.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Add a minimum QRB filter to the station filter bar (#21)
Next to "Show only QRB [km] <=" the bar gets "Show only QRB [km] >": it
hides every station up to the entered distance. With both switched on the
list keeps a range, min < QRB <= max. The minimum is exclusive, as the
issue asks, so the two filters split the list without overlap at the same
value. A station without a known QRB is hidden while either is on, as the
maximum already did. The value lives in the bar for the running session
only, like the maximum, and the station map follows through activeFilters.
The distance check moves from the controller predicate into
StationFilterState.acceptsQrb so it can be tested; the maximum keeps its
behaviour (null or > max is hidden).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Document the minimum QRB filter (#21)
Changelog v1.50 under New, the filter tables of the user interface pages,
the distance filter section of the feature pages, and the station filter
page of the website.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Let the first filter bar row wrap instead of clipping (#21)
With both QRB filters the first row of the station filter bar no longer
fits a narrow station pane, and as a plain Row it cut the bearing buttons
off at the right edge. It is a FlowRow now. Reset, the QRB pair, the QTF
controls and the bearings each move to the next line as one group; the
QRB pair and the bearings only break up inside their group when even the
group is wider than the pane. Separators in the wrapping row have a fixed
height, because a flow row line has no intrinsic height to fill.
A screen test at 300 dp checks that every bearing button, both QRB labels
and fields and the QTF label are displayed and end inside the bar.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Refuse NaN, Infinity and negatives in the QRB filter fields (#21)
toDoubleOrNull also reads "NaN" and "Infinity": a pasted "Infinity" as
the minimum hid every station, "NaN" let every one through, and the
maximum field had the same hole before. Both QRB fields now go through
parseQrbKm, which takes only finite, non-negative distances; anything
else leaves the field at its last valid value.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Let the station map follow the station filters (#21)
The Compose map built its markers from all station rows, so no filter
reached it, although its status line reported "filtered view active".
The JavaFX map, up to beta-v1.50.0-beta-001, took the sorted, filtered
list (StationMapBridge.refreshNow), so this is a regression of the
Compose port that never shipped. The markers now come from the rows the
list shows, and the filter is a key of the remembered markers so a
filter change redraws them. A selected station that a filter hides gets
no marker, as in the JavaFX map.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Filter again on redraw; no path analysis for a hidden station (#21)
DataTableState.visibleRows caches the filtered rows, and forceRedraw only
bumped contentRevision. The filter reads fields that change in place
(QRB, worked and QRV flags, last activity), so a redraw could hand back
rows filtered before the change -- in the list and, since the map draws
the same rows, on the map. forceRedraw now drops the cache, and
visibleRows reads contentRevision so a reader recomputes after it. In the
running application the user-list refresh also sets a new row filter,
which dropped the cache before; this makes the redraw itself sufficient.
A selected station without a marker -- hidden by a filter, or without a
locator -- no longer starts a path analysis: the map used to set
"Loading" with an empty target and send the request. The analysis is
cleared instead, and the effect re-runs when the station's marker
appears or disappears.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Add a settings draft that holds edits until they are applied
The settings window is to stage its edits and apply them with Apply/Close
or Save instead of writing each one at once. SettingsDraft holds one
pending change per setting, reads through to the stored value, and on
apply writes everything first and then runs each effect key once, in the
order the settings were first changed. Form.committed registers its typed
entry with the draft so a click on Apply cannot lose it.
The spec explains the order and the settings that stay immediate.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Stage the settings tabs' edits and apply them with Apply/Close or Save
Every tab state now reads through the window's draft and stages its edits
there instead of writing ChatPreferences: Station (all but the login
credentials), Log synch, TRX synch, Airscout, Notification, Shortcuts,
Beacon, Message handling and GUI (all but the language). The lists stage
their roster, the committed fields stage what their rule accepts.
Effects stay attached to their settings. The Win-Test listener and the DX
cluster server share one effect for switch and port that brings the running
service in line with the applied state, so applying both restarts it once,
on the new port. The own-QRG follower runs once for both sources.
Apply/Close applies and closes, Save applies and writes, Connect applies
before it connects, so the session uses what the tabs show. A field that
refuses its entry on the way stops the button with the reason shown.
Immediate as before: the login credentials, the language, the colours, the
file chooser, the test spot and the profile and worked-database actions.
Number fields keep the typed text while the staged value comes back as
Compose state; otherwise "1" became "1.0" before the next key.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Ask about unapplied settings when the window is closed
Closing the settings window with its X, or through the Windows menu, while
edits wait in the draft now asks: apply them, discard them, or go back to
the window. Without pending edits the window closes as before. Fields and
list rows still holding a typed entry are committed first, so an entry
typed just before the X counts.
The request only sets Compose state; the window decides on its next frame,
so nothing waits on the event thread. A close from outside -- a profile
switch, the shutdown -- discards the draft: the preferences it belongs to
are about to be replaced.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Document that settings take effect with Apply
PROJECT_CONTEXT records the draft and what stays immediate, the changelog
gets one entry under v1.50, and the configuration manual no longer says
that settings take effect while they are edited.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Repair stored settings once before the settings window is built
The fallback band of the DX cluster was repaired to 144 inside a getter
that runs during composition, and the Win-Test broadcast address was
auto-detected in a tab state's constructor. Both wrote ChatPreferences
past the draft at a moment nobody could see. repairStoredSettings now does
both once, before any tab is built; the tab states only read.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Ask before a profile switch or exit, and check fields before applying
A profile switch and the exit dropped pending settings edits silently. They
now go through SettingsWindow.settleThen: with edits pending, the window is
raised and asks Apply / Discard / Back; Apply applies them to the profile
still in force, Back cancels the switch or the exit. The question is Compose
state and the continuation is posted from the dialog's callback, so the
event thread never waits. Without pending edits both go on at once.
Fields are checked before anything is applied: a field still showing text
that does not parse, or that its rule refuses, refuses the apply and is
named in the notice, so the valid edits stay pending and Discard is still
possible. This covers number fields and the guarded external-interface
fields, which used to let Apply go ahead with the last valid value.
Every write and effect of an apply now runs on its own; failures are
collected and reported, a failed write stays pending and skips its effect,
and Save still writes the file when only an effect failed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Check every drafted setting with a real draft, and rename write-through tests
DraftedSettingsTableTest walks every drafted setter of every tab with a
real SettingsDraft: unchanged after the edit, changed after apply,
unchanged after discard. The tab tests run with a draft that applies at
once and cannot tell a drafted setting from one that bypasses the draft.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Document the question on switch and exit, the field checks and the repairs
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Decide the pending-settings question on the event thread, never trap
The question is now a SettingsQuestion decided in settle() on the calling
event thread instead of in the settings window's next frame, so an exit
from the main window does not depend on frames of a minimised settings
window; the window is raised and restored before the dialog is needed.
While a question is open, further requests are ignored: an X no longer
replaces the continuation of an exit or switch with a plain close. Each
continuation of settleThen runs at most once.
A field holding text it refuses no longer traps the operator: the X, the
switch and the exit show the question anyway, with the reason, Apply
disabled and Discard and Back available.
The apply-failure notice keeps the technical detail out of the sentence,
in parentheses after a translated one.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The release body was built from direct commits on main and GitHub's
generated PR list only; whatever was written into the tag was lost. An
annotated tag's subject and body now open the notes, ahead of "What's
Changed", and so reach the in-app update feed too. The tag object is
fetched explicitly because actions/checkout can leave the triggering tag
as a lightweight ref; contents:subject/body leave the SSH signature out.
Lightweight tags contribute nothing, as before.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Three nightly macOS DMG builds on the Intel runner failed in a row after a
successful notarization: `xcrun stapler validate` lost its connection to
api.apple-cloudkit.com ("The network connection was lost.", exit 68). The
ticket was fine; the runner's connection to Apple was not. staple and
validate now get up to five attempts with growing pauses (20, 40, 80,
160 s) before the build fails.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Translate the interface texts that stage 9 left in English
The GUI tab checkboxes, the file chooser and dropdown placeholders, the
profile dialog's Apply, the settings notice's OK, the chat clear button,
the selected-station map button and "all", six column titles, the monitor
window's title and heading, and the refusal and hint texts of the profiles,
beacon and notification tab states now come from the translation files.
Column titles go through CurrentStrings in a lambda like the existing ones,
so the id stays untranslated and the stored layout keeps its keys. The tab
states read CurrentStrings as StatusIndicatorState already does. The
English values are unchanged, so the screen tests still find every control
by its old label; a new test switches to German and checks one sample per
path: a composed label, a column title and a tab-state refusal.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Translate the settings notices, profile labels and alert button too
Wires the settings.spot* and settings.write* keys that existed but were
never used, translates the settings window title and its notice titles,
the profile tab's "unknown", "Copy of" and database-mode labels, and lets
the shared alert use alert.ok instead of a literal. The German for
messages.qsoOfTheOther becomes "Fremd-QSOs", the word the menu already uses.
"Copy of" is only the suggested display name; the profile id is derived
from the entered name once at creation and stored, so a later language
switch does not touch it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Commit settings fields only on a real edit, not when a tab opens
Form.committed treated every onFocusChanged callback with "not focused"
as the operator leaving the field. Compose also fires that callback once
when the modifier is attached, so drawing a tab committed each of its
committed fields: opening Colours wrote all six roles as overrides of
themselves (shown as "changed", and after "Save settings" they hid the
profile's stylesheet file for good), opening Beacon validated both
templates and restarted the beacon timer, opening Notification
re-committed the port and the spotter callsign.
The field now commits only on a transition from focused to unfocused,
or on Enter, and only when the draft differs from the stored value.
A real edit still commits on leaving the field or on Enter.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Lock the Station tab's login fields while a session exists
StationTabState.loginFieldsEnabled was a plain field that nothing ever
wrote and that Compose could not observe, so callsign, names, locator,
password and both chat categories stayed editable while connected,
although the running session does not pick up a change.
The flag is now derived from the ON4KST connection state, read through
a lambda over the main window's Compose state (MainWindowSurroundings),
so the tab follows the session while it is open: the fields are locked
for every state but DISCONNECTED and unlocked again after disconnecting.
The screen-test helper fieldBeside now matches text fields by their
editable text, because a disabled field offers no set-text action.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Drop changelog entries for defects that never shipped
The committed-field and login-field defects exist only in the Compose
settings tabs, which are in neither v1.44.0 nor beta-v1.50.0-beta-001,
so there is nothing for an operator to read about.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Compare committed fields with their last commit, not a stale value
Form.committed skipped a commit when the draft equalled `stored`, the
value the caller passed at its last composition. Most callers read it
from plain preference getters, so nothing recomposes them after a
commit: after Enter, `stored` still named the old value, and leaving the
field committed the same entry again, restarting the beacon timer twice
for one edit. The field now keeps the value it last committed in its own
holder, compares the draft with that, and falls back to it on a refusal.
The comparison is pluggable. The Colours tab compares the way
PaletteOverrides normalises a colour (trimmed, case ignored), so retyping
the shown colour in lower case no longer becomes an override of itself.
Behaviour kept on purpose: Enter on an unchanged value commits nothing,
so it neither restarts the beacon timer nor reports a stored value the
rule refuses. A field removed while it holds the focus (tab switched,
window closed) still commits a typed entry once, because Compose reports
the cleared focus as a focus loss; a test now pins this, as it was before.
Also: a test drives the Station tab's lock from ConnectionStateFeed
through MainWindowSurroundings, the chain the application wires, and the
Javadoc of openSettingsWindow is back above its own method.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Judge a committed field against the live stored value
be6862c8 kept the value a field last committed in a holder keyed on
`stored`. That holder could outlive the truth: the field commits 9 over
5, nothing recomposes the tab, a profile switch puts 5 back and the tab
recomposes with 5 -- the same key -- so the holder still said 9. The
field showed 9, the preference held 5, and leaving the field committed
nothing, where main would have committed 9.
Form.committed now takes `current`, the value stored right now, read
when the field commits; the draft is judged against it and falls back to
it on a refusal. The holder is gone. Right after a commit the getter
already returns the new value, so Enter followed by leaving the field
still commits once. Every caller passes its preference getter: both
beacon templates, the beacon interval, the DX cluster port, the spotter
callsign and the colour roles.
`isUnchanged` must agree with what commit stores, which the KDoc now
says. The interval and the port compare as the number stored, the
spotter callsign ignoring case and surrounding blanks, the colours as
PaletteOverrides normalises them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Require the live value in committed fields and compare the spotter call as stored
Review follow-up. `current` has no default any more, so a future caller
cannot silently bring back the stale comparison. The spotter call is
compared after the same uppercase(ROOT) the commit applies, so a typed
"X" against a stored "x" still commits.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Keep tests out of the user profile
The no-argument ChatPreferences constructor seeds preferences.xml in
~/.praktiKST when it is missing, so 21 older test files could create a
profile file for whoever ran the build. They now use ChatPreferences(Path)
with a @TempDir. None of them reads the file: their assertions are on the
field initialisers, which both constructors share, so no assertion changes
meaning. ChatPreferencesStationMapVisibilityTest passes its file to the
constructor instead of redirecting it with the setter afterwards.
The two opt-in ON4KST practice tests keep the no-argument constructor: they
deliberately read the operator's real profile for live credentials.
As a net, the test tasks run with user.home pointed at build/test-user-home,
emptied before each run, and fail if a preferences.xml appears there. This
also keeps the sound files, the map tile cache and the Skiko native library
that tests unpacked into the real home directory out of it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Make DXClusterThreadPooledServer.serverSocket volatile
run() assigns the server socket on the server thread, and stop() closes it
from another thread. closeServerSocket() is synchronized, but the write in
run() is not, so nothing guaranteed that stop() saw the socket run() had
opened; it could find null and leave the socket listening.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Drop the ineffective SO_REUSEADDR from the Win-Test sked socket
WinTestSkedSender.sendUdp() set reuse address on a socket the constructor
had already bound, where it has no effect, so it was never in effect in any
shipped version. The socket binds an ephemeral port and only sends, which
gains nothing from it, and enabling it for real would change UDP multi-bind
semantics on Windows. Removing the call keeps the behaviour exactly as
shipped and stops the code from claiming an option it does not have.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Check the test user.home in a finalizer task
The stray-preferences check ran as a doLast of the test task, which Gradle
skips once a test fails, so a failing run could hide the file. It is now a
separate task, checkTestUserHome, that every test task is finalized by;
finalizers run when the test task fails too. It captures only a File, so it
stays configuration-cache safe.
Each test task now gets its own scratch home, build/test-user-home/<task
name>, so a second test task cannot wipe another's. Emptying it checks the
result of deleteRecursively(), which reports failure instead of throwing,
and stops the task, so a leftover file is never blamed on the current run.
The check stays limited to preferences.xml. Other files land in the scratch
.praktiKST by design, such as the sound files PlayAudioUtils seeds from the
bundled resources; absorbing those is the redirection's job, and a wider
check would need an exclusion list that grows with every seeded file.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Keep generateStrings configuration-cache safe
The argument provider of generateStrings referenced the script-level
generatedI18nDirectory, which captures the build script object, so storing
the configuration cache failed for every build that included app-desktop.
The task now works on a local copy of the provider.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Declare the test user.home as a test task output
Test tasks are cacheable, and a cache hit skips doFirst, so the scratch home
was neither emptied nor rewritten and checkTestUserHome judged whatever an
earlier run had left there: a false alarm after a violating run, or a hidden
violation when a passing result for violating inputs came from the cache.
The scratch home is now a declared output, restored and replaced together
with the cached result, so the check always sees the files belonging to the
inputs. The doFirst wipe stays for real executions.
Skiko unpacked its native library, about 30 MB, into the scratch home, which
would now land in every app-desktop cache entry. skiko.data.path points it
at build/skiko-data instead, outside the output and kept between runs.
user.home and skiko.data.path are passed through an argument provider: as
system properties they would be task inputs, and their absolute paths would
make the cache key differ between checkouts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Send the PSTRotator azimuth with one decimal instead of truncating it
setAzimuth cast the azimuth to int, so 145.7 degrees turned the antenna
to 145. It is now rounded to one decimal and formatted with Locale.ROOT,
so a German default locale cannot turn the dot into a comma; a value that
rounds up to 360.0 is sent as 0.0. The loopback test pinned the old
truncation and now expects the decimal.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Bring back the turn-antenna button lost in the Compose migration
The JavaFX station panel had "Turn ant1 to <call>", the only caller of
ChatController.rotateTo; removing JavaFX (fe4215bc) dropped it, so the
antenna could no longer be turned from KST4Contest at all. The Compose
panel gets the button again, enabled when a rotator callback is wired and
the selected station has a finite heading.
Review follow-ups for the decimal azimuth: formatAzimuth refuses NaN and
infinite values and wraps any finite value into [0, 360). The "reached"
check before the SPID retry uses a 1.0 degree tolerance, because the
target now carries a decimal while many controllers report whole degrees.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Keep rotateTo's Javadoc on the method and word the German label as a bearing
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* test: restore the frequency-context and Band.fromPrefix tests (#105)
The test file MessageBusManagementThreadFrequencyContextTest sat under
src/main, fully commented out, and was deleted with #62 without ever
having run. Its two concerns come back as real tests: the bare
three-digit context rule of hasExplicitBareFrequencyContext in the
controller test package, and Band.fromPrefix as BandFromPrefixTest in
the model package. The original expectations still match the current
behaviour; a few boundary cases (context window, invalid indices,
non-exact prefixes) are added.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test: characterise the autoanswer path of MessageBusManagementThread (#105)
Runs real CH frames through a started thread (receive queue plus poison
pill, real transmit queue, mocked controller collaborators, a real
ChatPreferences in a temporary directory) and pins the current
behaviour of the private-message branch before it is refactored:
- a QRG question addressed to the login callsign yields exactly one
reply with the fixed marker and the QRG of the incoming category;
- a second question from the same station and category within the
cooldown yields none, while another station or category is answered;
- a frame carrying the autoanswer marker, a public frame, a frame to
another station and a disabled setting yield none;
- an empty QRG and a blank generic text are rejected before the
cooldown is recorded, so a later valid reply still goes out;
- //VER, the generic reply and its precedence below the QRG reply.
One oddity is pinned, not fixed: a lower-case login callsign defeats
the "directed to me" comparison because the receiver callsign is
upper-cased by ChatMember.setCallSign but compared with equals.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor: extract the autoanswer decisions into pure functions (#105)
The QRG-question texts become a constant set; matchesQrgQuestion,
isAutoMessage and buildAutoAnswerMessageText are static package-private
functions with explicit parameters instead of reaching into the
preferences, and the cooldown reads an injectable millisecond clock
(default System::currentTimeMillis) through a new constructor overload.
Behaviour is unchanged: the characterisation test from the previous
commit passes untouched, the new decision and cooldown tests cover the
extracted functions without mocks.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor: extract frequency detection into a pure function (#105)
smartFrequencyExtraction now copies the sender's band timestamps, reads
the fallback prefix and the injectable clock, and hands the text to the
static package-private detectFrequencies, which returns the detections
in message order as DetectedChatFrequency records. The thread still
stores each one through applyDetectedFrequencyToActiveMembers and logs
it as before.
Behaviour is unchanged. In production the stored detection lands in
the same map the next match of the message reads, so detectFrequencies
feeds every detection back as context with the current time; a test
pins that "10368.100 then .150" yields 10368.150. The unreachable
NumberFormatException catches around parseExplicitFrequency are gone,
since that parser already returns null instead of throwing. Also pinned:
a plain decimal between 50 and 54 (e.g. "qrb 52,3 km") is read as a
6 m frequency.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* refactor: extract the directional-opportunity and spot decisions (#105)
classifyDirectionalOpportunity returns MISSING_LOCATOR, IN_ANGLE_AND_RANGE
or OUT_OF_ANGLE_OR_RANGE from the locators, the maximum QRB and the beam
width, and replaces the identical locator check plus isInAngleAndRange
call in both the live (CH) and the history (CR) branch.
shouldPlayDirectionalTick, decideDxClusterSpot (NO_FREQUENCY,
ALREADY_WORKED, SPOT) and buildDxClusterSpotEntry take over the tick
condition and the automatic DX Cluster spot of the CH branch. All are
static package-private and covered without mocks.
Behaviour is unchanged. Pinned: the spot decision only checks that the
frequency text is non-blank, so a value outside every supported band is
spotted even for a station flagged as worked.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* refactor: remove dead code from MessageBusManagementThread (#105)
processRXMessage23000 belonged to the long-retired raw port-23000
interface and had no caller. With it go the two helpers only it
reached, checkIfItsUserListEntry and checkIfMessageInhibitsFrequency,
plus the caller-less checkIfItsUserListEntry23001 and
checkListForChatMemberIndexByCallSign, the patterns PTRN_USERLISTENTRY,
PTRN_QRG_CAT2 and PTRN_QRG_CAT3, the unused chatMemberTable field and
the serverReady flag with its never-called accessors. A grep across core
and app-desktop finds no remaining reference.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* refactor: keep the sender-category fallback in the QRG auto-answer warning (#105)
Review follow-up. The extracted buildAutoAnswerMessageText logged only
the message category, while the old warning used autoAnswerCooldownKey,
which falls back to the sender's category. The caller now passes the
same fallback, so the log line is unchanged from main.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* test: draw and click the in-window menu row (#107)
Pins what MainMenuModelTest cannot see: every item picked through the
drawn row runs its own action, the enabled states and the away and
window-toggle labels follow their state while the menu is down, and a
palette or language change reaches the row on screen.
Named regressions: fe4215bc (menu actions hung on deleted JavaFX
MenuItems) and 83e23f55 (unthemed Swing menu bar), both in main as
2a5c89b9.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: draw the station list with its real row key (#107)
The station row key moves from an inline Java lambda into
StationColumns.rowKey, unchanged in its text, so the list can be tested
with the key the application really uses.
The screen test pins three regressions on the drawn table: the same
callsign in two categories collided under one key and sorting dropped
the filters (both 21251377), and the AP column sorted by its text
(ab282162, #87). It also pins that rows replaced after the first draw
reach the screen and that the selection follows a station by key.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: draw the monitor window's live tables (#107)
MonitorContent becomes internal so its body can be drawn headless.
The screen test pins the defects the port shipped in this window
(cc8c9020, in main as 2a5c89b9): rows arriving while the window is
open, also after a redraw that was not about the rows, must reach both
tables, and two messages that read alike stay two rows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: draw and click the station map window's body (#107)
StationMapWindow.show now hands its body to an internal Content
composable that takes the tile fetcher, the member roster and the map
state as parameters, with the old inline values as defaults, and the
map canvas carries a test tag.
The screen test pins the map selection fixed in 21251377: a marker
click and Reset view go through the central station selector, a click
keeps the category variant already selected and prepares "/cq CALL",
and neither changes the zoom. The header switches write the profile.
Tiles never arrive: an empty temporary cache and a failing source.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: draw the Station, Log synch, TRX synch and Airscout tabs (#107)
Each tab is composed headless and checked for its sections and key
controls, and editing a control through the drawn form must write
through to ChatPreferences and redraw. Refused entries stay on screen
as typed and leave the stored value alone. A small helper finds a
control by the label beside it, the way the operator finds it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: draw the Message handling, GUI, Beacon and Notification tabs (#107)
Each tab is composed headless with its sections and key controls, and
edits through the drawn controls reach ChatPreferences and the
application callbacks: the auto answer for both categories, the
debug-to-file switch c8ce0e98 restored, the startup design and the
language, the beacon templates and interval, the cluster server, the
fallback band, the test spot and the monitored callsign dialog.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: draw the Shortcuts and Worked database tabs (#107)
The shortcut and snippet lists show what the profile holds, and an edit
or an added entry made in the drawn list reaches the roster and the
refresh hook the button row depends on (the hook that was called into
thin air before fe4215bc). The worked database lists its rows, re-reads
them on refresh, asks before a reset and leaves the lists alone when
the database reset fails.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: draw the Profiles and Colours tabs (#107)
Profiles: the default profile and every action are on screen, deleting
it is refused after the confirmation, a cancelled New dialog creates
nothing, and switching hands the selected profile to the application.
Only a temporary registry is touched.
Colours: the six roles show the resolved colour, a typed colour is
committed into the store, a mistyped one is refused and put back, the
shipped reset undoes a change, and the stylesheet export writes once
into a temporary profile directory and never over an existing file.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* refactor: give the map's path analysis a seam and tag the monitoring field (#107)
StationMapWindow.Content takes a MapPathAnalysis supplier. The default
looks the running application's reachability service up on every
selection, exactly as the inline chat-controller lookup did, and null
still means: no pan, no analysis. This lets a test run the zoom rule
on selection that cc8c9020 fixed.
The callsign field of the add-to-monitoring dialog in NotificationTab
carries a test tag, so a test types into it by name rather than by
position.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: isolate and harden the screen tests after review (#107)
- No screen test builds ChatPreferences() any more: that constructor
resolves the real user profile and seeds preferences.xml there. Each
class uses ChatPreferences(Path) under a @TempDir.
- Compose state changed from the test thread (roster refresh, language
reset) goes through runOnIdle.
- Dialogs and dropdowns are waited for with waitUntil and found inside
their overlay (isDialog / isPopup) instead of by tree order; the
monitored-callsign field is found by its test tag.
- The menu test now asserts Disconnect when online and Connect when
offline instead of skipping one.
- The map test runs the zoom-on-selection rule through the new seam,
closes its tile fetcher after each test and keeps its cache in the
@TempDir.
- StationColumns.rowKey is pinned directly: two categories, two keys.
- The menu colour is sampled further inside the row, and the station
list click picks its row by position.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: prove the map's selection effect ran when no analysis is available (#107)
Review follow-up. The no-analysis test only asserted that the map stayed
put, which would also hold if the selection effect never ran; it now
counts the analysis lookup. The effect reads the lookup through
rememberUpdatedState, so a caller passing a new lambda is not served a
stale one.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* test: pin the Win-Test sked frames over a loopback socket (#106)
WinTestSkedSender is pointed at the loopback address instead of a
broadcast address, and a socket on an ephemeral port receives the
LOCKSKED/ADDSKED/UNLOCKSKED sequence. The frames and their checksum
byte are written out literally, including the \260 degree escape and
the SSB fallback for an unexpected mode ID.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: cover the PSTRotator UDP client over loopback (#106)
A socket on an ephemeral loopback port plays PSTRotator. The tests pin
the poll queries, the <PST> command frames, the AZ:/EL:/MODE: replies
reaching the listener, an unparsable reply leaving the loop alive, and
the explicit-port constructor reporting its status after start() (#115).
The reply format is derived from the parser and the PSTRotator UDP
documentation, not from a capture.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: cover the AirScout UDP listener over loopback (#106)
ReadUDPbyAirScoutMessageThread gains a package-private constructor that
takes an already bound socket, as ReadUDPbyUCXMessageThread has; the
public constructor still binds the configured port in run(). The tests
pin applying an ASNEAREST response to every active variant of the
receiver, the case-sensitive server/client match with a UTF-8 client
name, the disabled listener, the poison pill, recovery from a malformed
response, and the date losing its space because the parser strips all
spaces.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: cover UCXLog/DXLog QSO packets over loopback (#106)
The UCX loopback test now sends contactinfo and contactreplace packets:
transport bytes before the XML are dropped, the QSO reaches the
controller with band, locator and source, the gross field is
registered, an unknown station is stored in the database and the UDP
backup file, a known one is only updated, and nothing happens while log
sync is disabled. A RadioInfo packet sent afterwards serves as an
ordering marker instead of a sleep.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: cover the local DX Cluster server over loopback (#106)
DXClusterThreadPooledServer gains a package-private constructor that
takes an already bound server socket; the public constructor still binds
the configured port in run(). Plain sockets on the loopback address act
as logging programs. The tests pin the listening status, the "login: "
prompt, one DXSpider-style line with BEL BEL CR LF per spot to every
client, the rejection of spots that cannot be formatted, and stop()
closing the clients and ending the server thread.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix: keep the UCXLog/DXLog listener alive on bad input (#106)
An unwritable or empty backup path left the FileWriter null, and the
following BufferedWriter threw a NullPointerException; a RadioInfo
without <Freq> or <Mode> threw one as well. Neither was caught, so the
listener thread ended for good. The backup line is now written with
try-with-resources and skipped with a log line when the file cannot be
opened, an incomplete RadioInfo element is skipped, and run() drops a
packet whose processing throws any RuntimeException, reports a
non-critical status and keeps receiving.
The loopback tests cover all three paths. The interrupt test now waits
for a processed marker packet instead of sleeping and no longer claims
more than it can observe, and the marker helper states its in-order
assumption for UDP on loopback.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix: let a second DX Cluster server run() bind again (#106)
The pre-bound-socket constructor made run() skip binding whenever a
server socket existed. A run() after the first one ended would then
accept on a closed socket. Bind whenever the socket is missing or
closed, which restores the behaviour of the public constructor.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: make the ignored AirScout responses observable (#106)
The tests for responses to another client or server and for the
disabled listener only checked that nothing happened before the poison
pill, which a listener that never read the datagrams would also pass.
Each now sends a valid sentinel afterwards and waits for it, and the
ignored responses name a different receiver, so a wrongly processed
copy is detectable; the disabled case also shows that the switch was
asked once per response. The date test is renamed and documented as a
pinned known defect, so a parser fix knows to flip it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: wait for the PSTRotator listener thread after stop() (#106)
The client keeps its listener thread private, so the test finds it by
the name start() gives it and joins it after stop(), asserting that
exactly one such thread existed and that it ended.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix: keep the UCX listener alive when the status callback throws (#106)
Review follow-up. The safety net reported a dropped packet through the
status callback without guarding it, so a throwing callback still ended
the listener. The test for an unwritable backup file is renamed to what
it proves, and the changelog mentions that any unexpected error now
drops only the affected packet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix: let ReadUDPbyUCXMessageThread own its DatagramSocket so interrupt() closes it (#115)
The field `socket` was a java.net.Socket that was never assigned; the real
DatagramSocket lived in a local variable of run(). interrupt() therefore
closed nothing and terminateConnection() would have thrown a
NullPointerException. The receive loop kept running in two-second timeout
steps until the user-disconnect flag was set.
The field is now the DatagramSocket, run() binds it when no pre-bound socket
was handed in, the loop ends on the interrupt flag or a closed socket, and
both interrupt() and terminateConnection() are null-safe. A package-private
constructor accepts a pre-bound socket so loopback tests can listen on an
ephemeral port; the public constructor used by ChatController is unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix: stop ReadUDPbyUCXMessageThread truncating datagrams after a short packet (#115)
A DatagramPacket keeps the length of the last datagram it received, so a
long RadioInfo packet following a short one was cut to the short one's
length and failed to parse. The length is reset to the buffer size before
every receive, the same way ReadUDPByWintestThread does it. The loopback
test sends a short and then a long packet on port 0 and checks that both
frequencies arrive.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix: give PstRotatorClient's port constructor a callback, decode AirScout as UTF-8 (#115)
The explicit-port constructor left the thread-status callback null, so
start() and the first received packet threw an NPE; the default
constructor now delegates to it. The AirScout listener decoded replies
with the platform charset although the inquirer sends UTF-8.
PROJECT_CONTEXT listed PSTRotator as TCP; the client speaks UDP.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix: decode UCX datagrams as UTF-8 and tighten the interrupt test (#115)
Review follow-up. UCXLog sends UTF-8 XML, but the listener decoded with
the platform charset before handing a String to the parser, so the XML
header's encoding was ignored. The port-in-use fallback now sets its
receive timeout before the first receive(). The interrupt test waits
1 s, below the 2 s receive timeout, so it fails if interrupt() stops
closing the socket; the QRG latches now count one value each.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Both source PKGBUILDs ran "./gradlew --offline" against the builder's
personal ~/.gradle. That only worked as long as the local cache happened to
hold every dependency; the SpotBugs plugin 6.5.12 added in #114 broke
"yay -S kst4contest-git" with "could not resolve plugin artifact".
- drop --offline and point GRADLE_USER_HOME into $srcdir, so the package
builds the same on any machine
- run with --no-daemon so no Gradle process outlives makepkg
- drop gradle from makedepends: the wrapper fetches its own distribution
Verified with makepkg -f on an empty Gradle home for kst4contest-git.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test: pin the Maidenhead and great-circle maths of Location against definitions (#108)
Subsquare centres from the Maidenhead definition, round trip through
toMaidenhead (lower-case subsquare pinned as is), distance and bearing
between locators against an independent spherical computation on
R = 6371.009 km, consistency with PathGeometryUtils, and the NaN bearing
for identical points pinned as current behaviour.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test: pin PathGeometryUtils against ITU-R P.525/P.526 and textbook values (#108)
Great circle against the Veness worked example, Fresnel radius against
P.526-15 eq. (2) and (3), earth bulge against the 8500 km effective
radius, radio horizon against 4.12*sqrt(h), knife-edge loss J(v) against
eq. (31), the v parameter against eq. (26), free-space loss pinned at the
project's 32.44 constant with the P.525 and exact values documented,
plus wattsToDbm, the sample-count rule and the k-factor sanitiser.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test: pin the path-analysis pipeline on a flat profile (#108)
A flat 20 km profile with 10 m antennas at 144.3 MHz through an
in-memory terrain provider: bulge, line-of-sight clearance, first
Fresnel radius and intrusion, no knife edge and 0 dB diffraction. Pins
that a partial Fresnel intrusion without a blocked line of sight
contributes 0 dB to the link budget; P.526 would give 0 to 6 dB there,
which is tracked separately.
Fixes#108
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* test: add bounded awaitUntil helper for timing-sensitive tests (#109)
One small class per module (core and app-desktop share no test sources):
polls a condition every 10 ms and fails with a message when the timeout
elapses, so tests no longer guess with fixed Thread.sleep pauses.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(map): let TileFetcher take its nano clock as a test seam (#109)
A package-private constructor accepts a LongSupplier for the monotonic
clock the retry pause is measured against; the public constructors
delegate with System::nanoTime, so production behaviour is unchanged.
TileFetcherTest advances a fake clock instead of sleeping 20 ms three
times and hoping the 1 ms pause has elapsed.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(view): put CoalescingTrigger on a DelayTimer seam (#109)
DelayTimer is a package-private one-shot timer (restart/stop) with a
Factory; DelayTimer.SWING wraps javax.swing.Timer and is what the public
constructor uses, so production behaviour is unchanged. A package-private
constructor takes the factory. CoalescingTriggerTest fires a FakeDelayTimer
itself: the negative checks (no second run, cancel prevents the run) are
now deterministic and the four Thread.sleep calls are gone. One test keeps
the real Swing timer end to end, waiting with awaitUntil.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(view): put LayoutAutosave on the DelayTimer seam (#109)
The public constructor still builds a javax.swing.Timer via DelayTimer.SWING
with the same 750 ms delay; a package-private constructor takes the factory.
LayoutAutosaveTest fires a FakeDelayTimer on the event thread itself, so
"cancelled means not written" no longer rests on a 1.2 s sleep and the
burst test no longer sleeps 300 ms to prove the absence of a second write.
The foreign-thread test keeps the real Swing timer path with awaitUntil.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* refactor(controller): inject scheduler and clock into On4KstConnectionManager (#109)
A package-private constructor takes the ScheduledExecutorService and a
LongSupplier for wall-clock milliseconds; the existing constructor delegates
with the same single-thread supervisor and System::currentTimeMillis, so
production behaviour is unchanged. Session's idle clocks start from the
injected clock instead of a field initializer. The static
parseMessageTimestamp keeps the wall clock: it sanity-checks server
timestamps and is not part of the supervisor's timing.
On4KstConnectionProbeTest drives unansweredProbeUsesExistingReconnectFlow
through a ManualScheduler (immediate work on a real thread, delays and the
five-second tick released on demand) and a fake clock: 90 s of silence
sends the CK probe, which the test reads off the socket, 210 s fails the
session, and the released backoff opens the second connection. The
reflection-based state setup is gone because the state is now reached the
production way; the test takes well under a second instead of seven. The
two ignored await(2 s) results in the server stubs are asserted now.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test: close the profile picker as soon as it shows instead of after 1.5 s (#109)
The two display-bound picker tests disposed the modal dialog from a
Swing timer after a fixed 1.5 s, which was long enough on a fast machine
only and cost 3 s per run. The closer now waits with awaitUntil for the
dialog to appear and disposes it on the event thread; the suite drops
from 3.0 s to 0.26 s.
Remaining fixed sleep: TopPriorityCandidatesWindowTest waits 400 ms for
Compose's double-tap timeout; whether ImageComposeScene's time can
advance it deterministically is still to be checked.
Fixes#109
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* build: write SpotBugs XML and HTML reports (#110)
The SpotBugs tasks had no report configured, so findings went to stdout
only and, with ignoreFailures, scrolled past unread. Every SpotBugsTask
now writes XML (for CI to count) and HTML (for people) to the plugin's
default location, <module>/build/reports/spotbugs/main.{xml,html}.
ignoreFailures stays; the build still passes with findings present.
Fixes#110
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* ci: run PMD and SpotBugs on pull requests and publish the reports (#110)
The compile job runs pmdMain and spotbugsMain after compiling (report
only: both tools have ignoreFailures, a crash of a tool still fails),
counts <violation> and <BugInstance> elements per module from the report
XML into the job summary, and uploads */build/reports/pmd and
*/build/reports/spotbugs as the static-analysis artifact for 14 days.
A missing report prints "no report" without failing the step.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs: name the PMD and SpotBugs report files and the CI artifact (#110)
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* build: give beta DEB and RPM packages a pre-release number (#113)
composePackageVersion is the package version of every Compose target
format, so an RPM or DEB built from a beta tag carried exactly the
version of the following stable one. dnf, rpm -U and dpkg refuse to
replace a package by one of equal version, so a beta user could not
move to the stable.
Keep the version and tell the packages apart by the package release
number, following the Fedora pre-release convention: the new Gradle
property linuxAppRelease is passed to jpackage as --linux-app-release
(Compose LinuxPlatformSettings.appRelease) and defaults to 1.
beta-v1.50.0-beta-001 -> 1.50.0-0.beta.001
beta-v1.50.0-rc02 -> 1.50.0-0.rc02
v1.50.0 -> 1.50.0-1
Everything starting with "0." sorts below "1" for both rpm and dpkg.
Nightlies and local builds do not set the property and keep release 1.
Fixes#113
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* ci: derive the Linux release number from the tag and verify the tag version (#113)
The DEB and RPM jobs now compute the package release from the tag
before calling Gradle and pass it as -PlinuxAppRelease: a stable
vX.Y.Z tag gives 1, a beta-vX.Y.Z-<suffix> tag gives 0.<suffix> with
"-" turned into ".". The result must match [A-Za-z0-9.]+, otherwise
the step fails. A workflow_dispatch run from a branch keeps release 1
like the nightlies; nightly-artifacts.yml is unchanged.
The test job, which every packaging job depends on, now checks right
after the checkout that the X.Y.Z in the tag equals
composePackageVersion in gradle.properties and fails with a clear
message otherwise, so a mistagged commit fails before an hour of
packaging rather than shipping packages whose version contradicts
their file name and upgrade order.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs: describe the pre-release number of beta DEB and RPM packages (#113)
The German and English installation manuals say in the Debian/Ubuntu
and Fedora sections that beta packages carry a pre-release number
below the stable one, so the stable installs as an upgrade over a
beta. The v1.50 changelogs gain the matching "Fixed" entry, and
PROJECT_CONTEXT.md records the rule next to the MSI decision.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* build: configure the Windows MSI installer in the Compose windows block (#61)
Add the jpackage/WiX options for a per-user MSI next to the existing
icon: a fixed upgradeUuid so Windows Installer replaces an installed
KST4Contest instead of installing side by side, a Start menu group and
desktop shortcut, no UAC prompt, and a directory chooser. The portable
ZIP (createDistributable) is unaffected; the MSI is produced only by
the tagged-release workflow.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* ci: build and attach a Windows MSI installer for tagged releases (#61)
The Windows job of tagged-release.yml now runs packageMsi after the
portable ZIP, renames the result to KST4Contest-<tag>-windows-x64.msi
(consistent with the other platforms; the ZIP keeps its praktiKST name)
and uploads it as the windows-msi artifact, which the release job
downloads and attaches next to the ZIP. The nightly workflow stays
ZIP-only because a nightly MSI would always carry the same version.
The website download data (downloads.js, beta.js) gets an MSI card that
is shown only when the release actually has that asset, so the page
never links to a file that does not exist for an older release.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs: describe the Windows MSI installer next to the portable ZIP (#61)
The German and English installation manuals list the MSI in the package
table, describe both Windows forms (installer: per user, Start menu
entry, desktop shortcut, chosen directory, in-place upgrade, removal via
the Windows apps list; ZIP: portable) and the shared data directory
%USERPROFILE%\.praktiKST, and add the MSI path to the Windows update
section. Changelog entries for v1.50 and a note in PROJECT_CONTEXT.md
record the fixed upgrade UUID and the tags-only decision.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* ci: build the Windows MSI for stable tags only (#61)
A beta MSI would carry the same composePackageVersion as the following
stable release. jpackage's MSI upgrade table only accepts strictly higher
versions, so Windows Installer would refuse the stable over an installed
beta and the tester would have to uninstall first. Betas therefore stay
ZIP-only like nightlies; the build, upload and download steps are skipped
for beta- tags, and the release action tolerates the missing asset path.
Manual, changelog and project context say so.
Raised by the Codex review on #111.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* docs: add the test-coverage design spec (#62)
Records the state found for issue #62 (JUnit classes under src/main that
never run, test libraries in the shipped app image, dead 2023/2024 scratch
tests, no coverage figure), the decisions taken (delete, move to src/test,
JaCoCo report-only without a threshold, tests before packaging), what is
deliberately left to separate issues, and the definition of done.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test: move the JUnit classes out of src/main and drop the dead 2023 tests (#62)
Gradle's test task scans src/test only, so the six JUnit classes under
src/main compiled on every build and were never executed. They move to
src/test of the module that owns the class under test, keeping their
package names: On4KstProtocolTest, On4KstSocketThreadTest and
ChatMemberDirectionOpportunityTest within core; Utils4KSTFrequencyTest,
FrequencyTextParserTest and ChatPreferencesStationMapVisibilityTest from
app-desktop into core, next to the sibling tests that already live there,
because they test core classes only. All 26 of them run and pass.
The fully commented-out MessageBusManagementThreadFrequencyContextTest is
deleted. Eight files under core/src/test/java/kst4contest/test from
2023/2024 go as well: they carry no assertion, five of them annotate a
static main() with @Test, which Jupiter never runs, and some write the real
preferences file, send real UDP broadcasts or open the real user database
when started by hand. The five main() scratch programs under src/main
(ReadUDPByWintestThreadTest, PatternMatcherTest, mapTest, RegexTester,
locatorUtils/TestLocatorUtils) and MockKstServer, which only an unused
import in ChatController still referenced, are removed too.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* build: make JUnit and Mockito test-only dependencies (#62)
With no test class left under src/main, the implementation declarations of
junit-jupiter-api and mockito-core in core and app-desktop lose their only
justification. They put junit-jupiter-api, junit-platform-commons,
opentest4j, mockito-core, byte-buddy, byte-buddy-agent and objenesis on the
runtime classpath and into lib/app of every app image, Flatpak and
installer. The duplicate catalog entries junit-jupiter-api,
mockito-core-compile and the mockitoCompile version go with them.
Verified: `:app-desktop:dependencies --configuration runtimeClasspath` no
longer lists any of the seven artifacts.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs: state that test failures fail the build and tests live under src/test (#62)
AGENTS.md and docs/PROJECT_CONTEXT.md still said the build configuration
had allowed test failures to pass. That was true of the Maven build; under
Gradle a test failure fails the build, and only PMD and SpotBugs run with
ignoreFailures. Both documents now say so, note that test classes live under
src/test only and that JUnit and Mockito are test-only dependencies, and
keep the rule to read the test-result XML rather than the exit code.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* build: add a JaCoCo coverage report (#62)
The jacoco plugin is applied to every module from the root build with the
tool version held in the catalog. jacocoTestReport now depends on test and
writes XML and HTML per module; a root task jacocoAggregatedReport merges
core and app-desktop into build/reports/jacoco/aggregated/ (XML plus HTML).
i18n-generator is excluded: it is a build tool and its coverage says nothing
about the shipped application. Kotlin classes are counted, because the
Kotlin plugin adds its output to the main source set the report reads.
Report only, on purpose. There is no jacocoTestCoverageVerification and no
threshold: a minimum set at today's number would trip on the next unrelated
change, and one set at a target would fail every build until tests exist for
the number's sake. The figure is published so it can be read when deciding
what to test next; the reasoning is in
docs/superpowers/specs/2026-10-11-test-coverage-design.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* ci: run the tests before packaging and publish coverage (#62)
The nightly and tagged-release workflows built seven package formats from
commits nobody had tested on the way: pr-compile-check.yml runs the suite
for pull requests only, and main can be pushed to directly. Both workflows
now carry a copy of that test job, including the libgl1 step and the
XML-based summary, and every packaging job declares `needs: test`. Copied
rather than factored into a reusable workflow on purpose: the comments in
the job explain decisions the packaging workflows should carry with them,
and three copies read better than one indirection. build-docs-pdf runs
pandoc only and is left alone; publish-flatpak-repo and release-tag are
gated through the jobs they already need.
The PR test job additionally runs jacocoTestReport and
jacocoAggregatedReport once the tests have run (not `if: always()`: a
report from an incomplete run would mislead), prints a line/branch coverage
table per module and aggregated into the job summary from the JaCoCo XML,
and uploads the aggregated HTML report as the coverage-report artifact for
14 days whenever it exists.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs: describe the coverage report in the project context (#62)
A short paragraph on JaCoCo in docs/PROJECT_CONTEXT.md: report only and
why, where the per-module and aggregated reports land, which Gradle tasks
write them, that i18n-generator is excluded, how the PR job summary and the
coverage-report artifact show the figure, and that the nightly and
tagged-release workflows run the tests before any packaging job.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test: round-trip every ChatPreferences field through the XML file (#62)
Drive all public single-argument setters by reflection to non-default
values, write preferences.xml, read it back and compare every getter.
Fields that are not persisted on purpose are listed with their evidence;
two real reader/writer gaps (second-category locator, sniffer words) are
documented with a BUG marker instead of being fixed here. Two further
tests load the bundled template and a legacy file without configVersion.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* test: cover the DBController schema migration from older user_version values (#62)
Open the seeded template database and hand-built release 1.1 and 1.3
schema fixtures with a DBController and verify that PRAGMA user_version
reaches the current level, every column of the current CREATE TABLE
statements exists, legacy callsign keys are merged to base callsigns and
existing rows and flags survive. A further test proves that reopening an
up-to-date database changes neither schema nor rows.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* chore: drop the committed JVM crash report and ignore future ones (#62)
hs_err_pid26344.log in the repository root was a HotSpot crash dump that
had been committed by accident. It carries no information about the
project and must not travel with the sources.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix: drop the dead QSO sniffer word list that corrupted the callsign list on load (#103)
ChatPreferences carried a second sniffer list, lstNotify_QSOSniffer_sniffedWordsList,
that no UI, no manual and no consumer outside ChatPreferences ever touched. It was
still written to preferences.xml as <snifferWords><w>, and the matching reader block
was a copy of the <snifferCallSigns> block: it cleared the word list and then added
every word, normalised as a callsign, to lstNotify_QSOSniffer_sniffedCallSignList.
Any entry that ever reached the word list therefore leaked into QSO monitoring on
every load.
Remove the field, its accessors, the write block and the read block. A <snifferWords>
element in an older preferences file is now ignored like any unknown element; the
callsign list and its persistence are unchanged.
ChatPreferencesSnifferListTest loads a file that carries both lists and asserts that
only the <snifferCallSigns> entry is monitored. The round-trip test loses its
exclusion entry for the removed setter.
Fixes#103
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* fix: remove the second-category locator that could never be set (#104)
ChatPreferences declared stn_loginLocatorSecondCat with the compiled-in default
"jn49fk". Nothing wrote it to or read it from preferences.xml, no settings tab
exposed it and no caller ever invoked the setter, so the field always held its
default. On4KstConnectionManager.configureMainChat nevertheless compared that
default with the station's locator and, with the second chat enabled, raised the
warning "ON4KST uses one locator per TCP session. The second-chat locator ... is
ignored" whenever the station's locator was not jn49fk, which is practically every
installation.
ON4KST uses one locator per TCP session by design, so a second-category locator can
never become meaningful. Remove the field, its accessors, the comparison and the
warning; the rest of configureMainChat is unchanged. The round-trip test loses its
exclusion entry for the removed setter.
No automated test covers configureMainChat: it is private, works on the manager's
private Session and needs a mocked ChatController plus a live socket handshake, so a
harness would far exceed the scope of this fix. The compile of core and app-desktop
and the shorter exclusion list are the evidence that nothing references the member.
Fixes#104
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Add the new spot condition to the DX Cluster Server manual pages and the
forwarding section of the Features page in German and English, record
the fix in the v1.50 changelog, and note the rule in PROJECT_CONTEXT.
Fixes#91
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Automatic spots from directional opportunities were generated without
looking at the worked state, so stations already in the log kept being
spotted to the connected loggers.
The spot block in MessageBusManagementThread now asks
isAlreadyWorkedForDxClusterSpot(sender) before broadcasting. The rule:
- the band is taken from the sender's known frequency via
FrequencyTextParser; an unknown band keeps the previous behaviour;
- the per-band worked flag of that band suppresses the spot
(ChatMember.isWorkedOnBand maps B_1296/B_2320/B_5760 to the
historically named worked1240/worked2300/worked5600 flags);
- the global worked flag suppresses the spot only when no per-band
flag exists at all (Simplelogfile source without band information);
- the global flag together with per-band flags for other bands does
not suppress, because the per-band information is authoritative.
Suppressed spots are logged at FINE. The direction check, the
inAngleAndRange marker, the tick sound and manual spots from the map
window and the settings test spot are unchanged.
Fixes#91
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The "AP [minutes / pot%]" column of the station list and of the directed
messages had no comparator, so DataTableState fell back to the rendered
text and "13 (75%)" sorted above "3 (50%)".
AirplaneCell now owns a shared comparator, AirplaneCell.order, that both
column definitions use. The sort key is the first aeroplane of the cell,
which is the one the operator reads and the one the colour marks:
ascending by its arriving minutes, then by the better potential first for
equal minutes. A null info (nothing computed) and an empty list ("nil")
carry Int.MAX_VALUE, so they sort after every real aeroplane in ascending
order, consistent with the qrb and qtf sentinels. The second aeroplane of
a cell does not influence the order. Cell text, colours, column ids,
titles and weights are unchanged, as are DataTableState and the sort
cycle.
Fixes#87
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Make DBController work on one database file per instance
The controller was a static singleton: an eagerly created static instance
opened the root database during class initialization, and both the connection
and the path were static fields. A second operator profile in the same process
was therefore impossible, and ChatController's own "new DBController()" never
opened anything - it silently adopted the eagerly opened root connection.
- drop the eager static instance in favour of a lazily created default instance
- turn connection and path into instance state
- add a constructor taking a database file name relative to the application
directory plus a flag whether a missing file is seeded from the bundled
template
- create additional profile databases empty instead of seeding them: the
bundled template carries 3452 foreign callsigns and user_version 0, which
would show a new operator foreign data and trigger the full callsign
normalization rebuild. The schema is created by the existing table setup.
- remember the shutdown hook so closeDBConnection can deregister it; otherwise
every profile switch would leave another hook holding a dead connection
All SQL statements keep referencing the plain field and are unchanged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Allow preferences and chat controller to be bound to a file set
ChatPreferences gains a constructor taking a preferences file name relative to
the application directory, so "profiles/OP2/preferences.xml" is as valid as the
historic flat "preferences.xml". A missing file is still seeded from the
bundled template, which gives an additional operator the same clean defaults a
first installation gets.
ChatController gains a constructor that passes both relative file names and the
seed flag through to ChatPreferences and DBController. The existing
constructors delegate to the historic file names, so nothing changes yet.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Introduce the operator profile model and its registry
Adds the descriptor, the path derivation and the registry persistence for
operator profiles. Nothing calls them yet, so behaviour is unchanged.
The descriptor stores only a shared/own flag, never a path. All file names are
derived in OperatorProfilePaths, so a stored path can never drift apart from
the flag that produced it. A profile identifier is a stable, file system safe
slug assigned once, so renaming a profile never moves a directory.
The registry is created lazily. An installation that only has the historic flat
layout gets no registry file and no profiles directory; the root profile is
synthesised in memory instead. That keeps a single operator installation
byte for byte the one it was before, and it keeps a downgrade to an older
release a no-op. A missing, unreadable or malformed registry is logged and
treated like an installation without additional profiles, never as an error.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Select the operator profile at startup
Resolves the active operator profile before the chat controller is built and
passes its two file names on, so preferences, layout and worked data follow the
profile.
The resolution is deliberately quiet for existing installations. With no
registry or exactly one profile nothing is asked and nothing is written, so a
single operator start is unchanged. Only from two profiles on does a small
picker appear with the last used profile preselected, where Enter or a double
click starts immediately. A "--profile" argument, or the equivalent system
property, skips the picker; an unknown name warns and falls back to the normal
selection instead of refusing to start.
The startup decision itself lives in OperatorProfileBootstrap and contains no
user interface code, so it is covered by headless tests. The window title gains
the profile name only when a second profile exists.
Command line parsing happens in init() and is kept in a process wide holder,
because JavaFX only knows the parameters of the instance it launched itself.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Release the background resources a discarded runtime owns
Several resources outlived a disconnect on purpose, which was harmless while a
process ran exactly one session for its whole life. They are now released when
the controller itself is closed:
- the ON4KST connection supervisor thread, which stopByUser did not touch
- the sked reminder scheduler, which had no shutdown at all
- the reachability executor, whose shutdown method existed but was never called
- the PSTRotator retry scheduler and its pending retry
- the map tile proxy, whose stop method existed but was never called, leaving a
server socket and a twelve thread pool behind
- the station map bridge listeners and its coalescing animation, which would
otherwise keep firing into a dead user interface
These are real leaks today; they only become visible when a second runtime is
built in the same process.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Switch the operator profile while the application is running
Splits the teardown out of stop() into a reusable, idempotent shutdownRuntime()
and adds File - Switch operator profile..., which tears the current runtime
down and builds a fresh one for the selected profile.
The new runtime is a new application instance rather than a second start() on
the existing one. Many controls are instance fields created once, so reusing
the instance would re-parent mounted nodes and register every listener twice.
A fresh instance is safe because the class keeps no mutable static state.
Closing every window during a switch would end the process under the JavaFX
default, so the application takes over the exit decision: implicit exit is
turned off, the main window gets an explicit close handler, and every exit path
runs through the launcher. That also fixes losing the window layout on exit
after a switch, because JavaFX only calls stop() on the instance it launched.
The two view timers are now cancelled null-safe; stop() used to dereference
them unguarded, which would fail if shutdown happened before they were created.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Manage operator profiles from the settings window
Adds a "Profiles" tab, appended last so no established tab position shifts,
offering create, duplicate, rename, delete, a switch between shared and own
worked stations, and activation of another profile.
Duplicating copies the whole configuration except callsign and password. The
antenna, locator, layout and integration settings are exactly the work nobody
wants to enter twice, while the credentials belong to one operator only. The
root profile can neither be deleted nor moved off the common station database,
because its files are the installation itself.
The login callsign and its raw form now default to empty instead of a real
callsign. The XML reader treats an empty element as "not set" and falls back to
the field default, so without this change a profile created without credentials
would come up carrying the callsign compiled into the defaults - and an
operator could transmit under someone else's call.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Document operator profiles
Adds a manual section in both languages covering where profile files live, the
choice between shared and own worked stations, managing profiles, the startup
selection and the --profile argument, switching while running, and the fact
that passwords stay in clear text so profiles are not an access boundary.
Records the architecture in PROJECT_CONTEXT: lazy registry, derived paths, why
an additional profile database is created empty, why the login callsign default
is empty, and the constraints of rebuilding the runtime for a switch.
Adds a v1.50 changelog entry in both languages including the upgrade notes: no
file is moved, and going back to an older release stays possible.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Make the terrain profile cache multi operator safe
The cache stored a single owner identity in a meta table and dropped the whole
TerrainProfileCache table whenever the configured callsign or locator differed
from it. With several operator profiles that turns the cache into a permanent
miss: every switch between two operators with different locators would discard
every computed profile.
Entries are separated by owner identity through the primary key already, so the
wipe is replaced by an owner table that simply records which identities are in
use. A different owner now misses the cache instead of clearing everybody's.
The cache also moves out of the worked station database into its own global
terrainprofilecache.db. Terrain profiles are pure geometry derived from two
locators and a sample count; at a multi operator station both operators share
one location, so a per profile copy would only double the traffic against the
terrain service. No migration is needed, the new file refills itself, and the
old tables stay readable for older releases.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Fail cleanly when the registry path has no directory
SpotBugs flagged a null passed to Files.createTempFile: the null check guarded
only createDirectories, while the temporary file creation would still have
dereferenced it. The path is resolved absolute so this is practically
unreachable, but bailing out with a log line is cheaper than the latent NPE.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Hpa6bjie5qkeNG62y6FmXm
* Code Version 1.50
* Add the in-app update concept for 1.50
Captures the agreed design so it can be handed over later: extended
update feed with per-artifact metadata, a build-time package-type
marker, a testable check service replacing the blocking call in the
ChatController constructor, a non-modal hint suppressed while connected
to ON4KST, verified downloads and platform-specific execution.
Concept only, nothing implemented yet. Line anchors were verified
against this branch.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HT8R7zTMoQnmJj6dGbaqqf
* Use one base font size and a per-profile startup design
JavaFX picks a 13px default font on macOS and Linux but 12px on
Windows, so labels in fixed-size controls were cut off outside
Windows. Both stylesheets now set 12px on the root.
Each operator profile now chooses light or dark mode for startup on
the GUI settings tab. The Windows menu entries remain a quick switch
for the running session. All windows, including the update window
and the top candidates list, follow the design, and the main window
no longer starts light while the station map starts dark.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Show a drag-to-install layout in the macOS DMG
The DMG window now has a background with an arrow from the app to
the Applications folder, in the KST4Contest colours. Finder applies
the layout through AppleScript; if that fails, e.g. on a headless
runner, the script warns and builds the plain DMG as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Use the macOS system menu bar in all windows
On macOS the main menus move into the system menu bar and are shared
with the secondary windows, so they stay available whichever window
has the focus. The connection state indicator becomes a read-only
LINK menu there instead of the status bar button. Other platforms
are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* List direct commits on main in the release notes
GitHub's generated release notes only cover merged pull requests, so
commits pushed straight to main were missing. The release job now
builds the notes itself: "What's Changed" holds "Commits on main"
(first-parent, without bot and squash-merge commits) and GitHub's PR
list as "Merged PRs".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Build with Gradle in two modules, and free core of JavaFX
Etappen 1 and 2 of the Compose migration (43 commits).
The Maven build, the single source tree and the JPMS module-info went together.
core now holds the domain and app-desktop the user interface, both built by
Gradle, packaged through jpackage from Gradle, with SpotBugs and PMD wired in.
JavaFX moves to the classpath, which is why the launcher is Main and not the
Application subclass. The PR check, the nightly artifacts, the release and the
AUR packages all build from Gradle; the macOS DMG points at its output.
Verified on Linux, macOS and Windows.
core then loses JavaFX entirely. It observes through kst4contest.observe --
ObservableValue, ObservableRoster, and UiDispatcher as the one boundary a
worker thread crosses to touch the interface. The controller collections become
rosters and the visible station list a pure derivation rather than a
FilteredList/SortedList chain, which is also what fixed two focus defects only
a load test had exposed: the keyboard focus stays where the operator put it,
and the send field keeps it after a row click.
The rule that outlives this commit: ObservableRoster is the store and the list
a view is bound to is a mirror of it, never the other way round.
* Draw the whole interface in Compose
Etappen 3 to 6a of the migration (129 commits). Every window the operator sees
is Compose; the JavaFX ones still exist beside them so the two could be
compared against a live session, which is the only way the differences came to
light.
The toolchain first, then the settings window with its eleven tabs and 72 real
preferences; then the multi-column table as a component, and with it the rules
it needs -- stable row identity, a selection that survives a refresh, column
order the operator can drag, and a repaint when the rows change rather than
when the list instance does. Then the update and monitor windows, and the main
window: 22 station columns, the filter bar, the selected-station panel, the
message tabs, the priority bar, the sked controls, the status indicators and
the timeline, assembled behind a splitter because Compose has none.
The map stops being a WebView. A Compose canvas projects WebMercator tiles
itself, with the clustering in core and an asynchronous fetcher in place of the
local tile proxy; derive() becomes a pure Kotlin reimplementation instead of a
reflection call into com.sun.javafx.
Most of the volume here is parity the operator would notice rather than new
behaviour: desktop density instead of Material's phone minimums, colours taken
from the JavaFX stylesheets instead of Material's palette, row dividers, band
columns, grid-square labels and their colours, the home marker and antenna
beam, pinch-to-zoom under Wayland, and layouts that survive a window narrower
than the chart they stack.
Two lessons are worth more than the diff. A control bound to a value Compose
cannot observe writes through and never redraws -- that one shipped repeatedly
with a green build, because no test drew anything. And a layout that clips
instead of wrapping is worse than one that is merely too narrow: half a word
is the worst available outcome.
* Remove JavaFX
The second half of Etappe 6 (61 commits). grep -rl "javafx\." over both main
source trees is empty, nothing imports javafx, neither classpath resolves it,
and org.openjfx is out of the build and the version catalog.
The finding that shaped the work: the JavaFX main window was built on every
start and never shown -- primaryStage's setScene and show were both commented
out. Most of those 11,655 lines were dead weight rather than a port. But not
all of them, and that is the trap this commit is really about: dead-looking
methods still fed live Compose sinks. They were cut at their seams instead of
deleted, and a sweep script that follows calls was written to find them,
because reading the code does not reveal them.
In order: the colour derivation frozen as checked-in test data, so the pin
survives JavaFX being removed instead of silently comparing against itself;
the live feeds lifted out of the never-shown window and given names and tests;
the dead scaffolding deleted in slices; then the lifecycle. UiDispatcher now
delivers to the AWT event dispatch thread, the ten Alerts are modal Compose
dialogs, HostServices gives way to java.awt.Desktop with the mailto/http split
JavaFX did not need, and extends Application, start(Stage), stop and launch are
gone -- main builds the runtime and a shutdown hook replaces stop().
An independent review found two deadlocks in that move, both the same failure
class, both invisible to the build: a blocking wait on the event thread for
work that can only advance on the event thread. The profile picker waited on a
latch and froze the client for good; ComposeWindowHost.close waited on
Thread.join and returned with the window still marked open, which made a
profile switch raise the dead runtime's window and leave the rebuilt profile
without its Connect button. A third came out of fixing them: a ComposeDialog
built off the event thread cannot create its GL context and never paints.
Three defects only a GUI run showed, never the build: an NPE on a deleted
field, a Compose menu still hanging on deleted MenuItems, and a category
bookkeeping field that looked dead but is read by the outgoing-message category
resolution -- deleting it would have sent messages in the wrong category.
The binding docs and the manual no longer describe a JavaFX client.
* Draw the menu bar in Compose, and give the drawing layer a net
What the removal left open (4 commits).
Compose's MenuBar is a javax.swing.JMenuBar set on the window, so Swing draws
it and no theme reaches it; with no look-and-feel configured that meant the
default Metal menu above a themed window. The bar is drawn in the window now,
except on macOS, where the system bar is where that platform expects it. Two
renderers read one description -- mainMenuModel is a plain function of the
state, which also gives the menus their first tests. Closing the main window
quits the application again, and the window writes its size back to the
profile; both had quietly stopped working when the last JavaFX stage went.
Then Etappe 7. Four faults shipped in the settings-window branch alongside a
green suite of 382 tests, because not one of those tests drew anything: a
control that never redrew, a tab strip that clipped mid-word, a form taller
than its window with no way down, and colours that came from Material rather
than the stylesheet. One regression test per class, and the spec's acceptance
criterion is met rather than assumed -- each fault was put back deliberately
and the matching test watched to fail.
The CI had never run the suite at all; it does now, headless. What that took
was not Xvfb: skiko's native library links against libGL and only has to be
loadable, since the drawing tests render in software. Without the package the
whole Compose half of the suite dies with a message that reads like a missing
display. Found by running the job under act rather than on GitHub.
The six tests that open a real window skip headless by their own guard. The
summary step counts from the result XML and fails on failures there, because
this project's rule is never to infer a green suite from an exit code.
* Make the colours adjustable per operator profile
Six colour roles per design, per profile: surface, window surface, field interior,
text, accent and separator line. A change applies at once in every open window and
survives a restart once the settings are saved, like any other preference.
Three sources, later wins role by role: the shipped stylesheet from the classpath,
then the profile's own stylesheet file if there is one, then the overrides stored in
the profile's preferences. resolvePalette is a pure function and reports where every
role's colour came from, which the tab shows per field -- with two sources able to
decide one colour, that is the only thing keeping "why is this colour what it is"
answerable, and the spec calls it mandatory rather than decorative.
Four things this had to get right, each learned the hard way and each pinned:
A role is a selector and a property, not a property alone. Three of the six are
stated outside .root -- text on .label, the separator on .separator *.line -- and
-fx-background-color alone appears a dozen times in the evening sheet. Measured: the
shipped daylight sheet states no colour at all, the evening sheet states five and
implies the sixth.
Kst4ContestTheme reads the palette from the store's state and not from remember.
That read is what makes a change repaint every open window; remember(darkMode) held
the palette until somebody toggled day/evening, so a change reached nothing.
A derived colour follows its source's value, never its provenance. The root
profile's own stylesheet is the one copyResourceIfRequired writes at every startup --
the shipped sheet itself -- so layer 2 declares the accent on every existing
installation with nobody having chosen anything. Keying textAccent on "does this come
from a file" repainted the evening accent from green to blue on upgrade and made
"reset to shipped" disagree with "discard my changes".
The contrast guard measures where colours are actually drawn, read off Theme.kt's
Material mapping rather than off the role names: background and surface are both the
window surface, and primary is textAccent, not accent. It reports a pair only when it
is both under WCAG AA and worse than the shipped sheet -- the shipped sheets have
three pairs under 4.5 themselves, so an absolute floor would warn on an untouched
client and teach the operator to ignore warnings.
The way back out of an unreadable palette is deliberately overbuilt, because it is
the one failure that locks itself in. "Reset to the shipped colours" takes the
shipped palette as a required parameter and reads no composition local, so there is
no path by which it could be drawn in the colours it exists to undo; it looks out of
place under a customised palette, and that is what makes it findable. Ctrl+Z, Ctrl+D
and Ctrl+R reach the three ways back even with the caret in a colour field. A
mistyped colour is refused and the field snaps back.
The profile's stylesheet is the hand-editing route the copy under ~/.praktiKST has
been promising since it was first written and never read. It is read now. A button
writes the shipped template as a starting point and never replaces a file that is
already there. An unreadable, truncated or non-CSS file does not prevent startup --
the role falls back to the shipped design. A sheet stating -fx-base and no
-fx-background states the derived window surface too, the way JavaFX does, or the
most natural hand edit would leave every window the shipped light grey.
101 new tests. All of it runs headless.
* Document the adjustable colours, and record the design and the plan
Both manuals get the Colours tab in full: the six roles, that a change applies at
once but survives a restart only after Save settings, the three ways back with their
keyboard shortcuts, what each provenance label means, and the readability notice that
warns without refusing.
Two things an operator would otherwise have to discover. The reset button is always
drawn in the shipped colours and works when nothing else can be read -- the risk
section of the design says plainly that a keyboard path only helps whoever knows it,
so the manual is where this has to be written down. And the root profile already has
both stylesheet files, because KST4Contest has been writing them at startup for
years, so most of its roles report "from file" rather than "shipped" with identical
values.
PROJECT_CONTEXT.md records the decisions that outlive this stage: the three sources
and their precedence, that a role is a selector and a property, that the theme must
read the store's state rather than remember, where the contrast pairs come from, why
the root profile's file layer always exists, and why the lifeline refuses the active
palette. Also corrects its claim that CSS stays global -- a profile now owns its
stylesheets.
The design document and the implementation plan are kept as they were written,
including the plan's own corrections where measuring the code contradicted it.
* Speak the operator's language
The interface is German or English, chosen per operator profile on the GUI tab, in force at
once in every open window, and remembered across restarts. 376 texts in both languages.
The texts are ordinary .properties files that nobody has to compile: a contributor copies the
English one, translates the values, and opens a pull request. A third module -- build tooling
that never ships -- turns them into Kotlin before anything compiles, which is what makes a
mistyped key and a wrong argument count compile errors rather than things an operator
discovers. It lives in the main build rather than buildSrc because buildSrc's tests do not run
on `./gradlew build`, measured, and a generator whose tests never run is the exact failure it
exists to prevent.
Incompleteness is allowed and incorrectness is not. A missing key warns and falls back to
English, so a partial contributed language can be submitted at all; an orphan key, a
placeholder the translation drops, a duplicate key or an unreadable file fail the build with
the generator's own message naming the file. The argument count always comes from the English
base. German is held complete by a test because it ships; a contributed language is not.
What is deliberately not translated is the harder half. Protocol frames and chat commands
would break the connection or go unread; the beacon texts, auto answers, snippets and
shortcuts go to other radio amateurs and sit in every operator's stored profile; the 273
println calls in core are diagnostics, and a German stack trace is worse for whoever reads the
bug report; and QRA, QRG, QTF, LOS, Fresnel and wkd read the same in either language. Three
tests guard the boundary by measuring behaviour rather than text: the protocol frames built
under three locales, every preference default compared by reflection, and no translation value
being a protocol command. The reflective one is the only one that covers what nobody thought
of.
One insight turned up three times. Anything built outside a composition freezes the language it
was built in: the window title, the settings tab strip and every column header are all
constructed once at startup -- the columns from Java -- so each now holds a function read when
it is drawn. That also fixed something older and unrelated to language: the main window's title
was a fixed "KST4Contest (Compose)" for the whole session, where the JavaFX window had carried
the live connection line. It carries it again.
A column's id is never translated. It is what the remembered widths and order are keyed on.
* Document the two languages, and how to contribute a third
Both manuals get the language picker in full: where it sits, that it applies per profile and
takes effect at once in every open window, and -- at more length -- what is deliberately left
alone and why. An operator who sees an English beacon text in a German client should be able
to read that it is meant that way.
The contribution route is written down in both, because plain text files are only useful if
somebody knows they are there: copy, translate, submit; a partial file is welcome; the build
reports mistakes by file name and writes a worksheet of what is still missing.
PROJECT_CONTEXT records the decisions that outlive this stage -- the four kinds of text and
the rule that the addressee decides, that core produces no display text, that incompleteness
is allowed and incorrectness is not, that anything built outside a composition needs its text
read inside one, and that a column id is never translated.
CLAUDE.md said two modules; there are three, and the third never ships. It also states the
German orthography now, which this session got wrong across a spec and two plans: ss after a
short vowel, so muss and dass.
* Remove the implementation plans now that every stage is executed
Thirteen plans, all of them step-by-step instructions for work that is committed. A plan left
behind reads to the next person like an open intention, and these describe a build that no
longer exists -- the first still explains how to introduce Gradle.
The specifications stay. They say why the architecture is what it is, which the code cannot,
and three of them are still argued from. The live-findings note stays too: CLAUDE.md sends
people to it, and its hardest lesson -- that a blocking wait on the AWT event thread deadlocks
the interface while the build stays green -- is written down nowhere else.
Its sweep script stays as well. The note argues explicitly that the script belongs beside the
finding so the next reader can measure rather than believe, and removing it would contradict
the reasoning it illustrates.
Two references in the note and one in the JavaFX-removal spec pointed at plans and now say
where the reasoning went instead.
* Point the JavaFX-removal spec at where its plan's reasoning went
* Remove the stray release-metadata patch from the working tree
A 416-line patch file that was applied long ago and has sat in the repository root since. It
predates this branch and stays in the shared history, where rewriting it would mean rewriting
commits that are already on main and on origin.
* Check the translation files in CI and say what they hold
classes already runs the generator, so an orphan key, a dropped placeholder
or a duplicate fails the compile job with the generator's own message naming
the file -- verified by breaking strings_de.properties on purpose.
A step summary counts each language against the base, so the author of a
translation PR sees what they covered without reading the build log.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Refuse a translation key that would not form a Kotlin name
The generator escaped every translation value but emitted the key verbatim as
the member identifier, and identifierFor strips only . _ - -- so a key carrying
a space, a brace, a quote or a newline from a \uXXXX escape broke out of
val <name>: String get() = lookup(...) into an init block or property
initializer that runs when Strings is instantiated. Translation files are the
contribution surface that arrives by pull request, so this was a path from a
merged .properties change to code in every client.
The generated identifier is now held to letters-then-letters-and-digits before
any source is written, alongside the existing orphan and collision refusals.
Proven by replaying a file-writing payload against the built generator: it now
throws and writes nothing.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Port the rebased diagnostics and session logging onto the modules
origin/main landed replayable session logging (MessageHistoryRecorder), file
logging (ApplicationFileLogging) and a user-list refresh split on the old flat
layout. The rebase carried the new files, but three things needed the new
architecture:
- ChatController.fireUserListUpdate() used javafx Platform; core is toolkit-free
now, so it dispatches through uiDispatcher like every other fire method.
- Two new tests were left in the old src/test/java; moved into core.
- ScoreServiceRefreshTest mocked javafx Platform statically; it now injects the
DirectUiDispatcher the no-arg constructor installs.
The Leaflet map performance work was dropped with the HTML map it optimised.
classes testClasses and test are green: 1069 tests, 0 failures, 3 skipped.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Give the debug-to-file recording a home in the Compose settings
origin/main added a checkbox to the old JavaFX settings that recorded the whole
ON4KST session to file; the Compose redraw dropped the old settings UI, so the
toggle had no control left even though the preference and its wiring survived
the rebase.
It returns to the message-handling tab: the state writes the preference and asks
the application, through a new SettingsHost callback, to open or close the
error-log handler and the history recorder in the order the old checkbox kept.
The explanatory line and its German translation say that passwords are redacted.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Fix Codex review findings on PR 89: station row key, map selection, sort after filter
- Station row key now includes the chat category, so the same callsign in two
categories no longer collides.
- Map marker click and Reset view go through the central station selection.
- Sorting now operates on the filtered rows, so filters stay effective.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Keep the map drawing test off the network and the user's tile cache
The test assumed no tile ever arrives, but it used the default TileFetcher,
which reads the real disk cache and the OSM tile server. On CI a tile now and
then landed before the frame was rendered and covered the background the
'unreachable tile source' test asserts on. An empty temporary cache and a
source that always fails make the assumption hold by construction.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
A logging program started as a Flatpak, or through a Wine environment
such as Bottles, uses the network permissions of that sandbox. Without
shared=network its 127.0.0.1 is not the loopback KST4Contest listens on,
so the connection is refused although KST4Contest reports the port
correctly.
Adds the check and the override command to the DX Cluster troubleshooting
section in both language versions, and notes that the published
KST4Contest Flatpak manifest already carries --share=network.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HT8R7zTMoQnmJj6dGbaqqf
* Sign and notarize macOS builds
jpackage cannot produce a distributable macOS bundle on its own. It ad-hoc
signs the embedded runtime and then re-runs codesign on the same files without
--force, which codesign rejects; and "--type dmg --app-image" re-signs the app
it is handed, replacing a Developer ID signature with an ad-hoc one. So the
build now creates an unsigned app-image, signs it from the inside out, and
wraps it with hdiutil.
Apple's notary service also unpacks JARs and checks the native libraries
inside them, which sqlite-jdbc ships for both architectures. Those are signed
before the bundle is sealed, since rewriting a JAR afterwards would invalidate
the seal. A preflight check verifies Apple's two criteria locally, so a missed
binary costs seconds rather than a round trip to the notary service.
Two long-standing defects surfaced while testing and are fixed here: the
bundle identifier defaulted to the main class's package name (kst4contest.view
instead of de.x08.KST4Contest), and every release reported version 1.0 in
Finder because --app-version was never passed. Neither affects existing users:
the app keeps its settings in ~/.praktiKST, independent of the bundle ID.
Both workflows call the same script the local Mac uses, so the two cannot
drift apart. Signing needs a keychain that can answer a UI prompt, which a
runner cannot, so ci-import-cert.sh creates a throwaway keychain whose
password is generated per job and discarded with it. Notarization goes through
an App Store Connect API key and needs no keychain at all.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* Sign bundle contents serially
Signing the app image's Mach-O files with "xargs -P 8" passed locally but
failed on a runner: codesign reported "replacing existing signature" and then
"No such file or directory" for that same path. The two libjli.dylib copies are
separate inodes, so this is not hard links being signed twice -- concurrent
codesign runs over one bundle are simply not reliable. Serially costs about a
minute, since each call waits on Apple's timestamp server.
Also stop the matrix from cancelling the other architecture on a failure; that
throws away half the diagnostic information from a failed run.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* Document macOS signing from 1.42
The installation guides described the right-click workaround as the normal
first launch. That stays, but as the path for 1.41.1 and older; from 1.42 a
double-click works. Both guides also show how to verify a download with spctl,
so the claim is checkable rather than something to take on faith.
The per-channel download notes distinguish where the channels actually stand:
Nightly is built from main and is signed as of now, while Stable still points
at 1.41.1, so those notes name the version instead of claiming it outright.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
The jdk.net incident was caused by duplication rather than by a single
oversight: the module list existed in module-info.java, in the jpackage Maven
plugin and in fourteen hardcoded --add-modules arguments across the workflows
and AUR PKGBUILDs. Only the path that CI does not use was kept up to date, so
every packaged build shipped a runtime image without jdk.net.
Add packaging/AddModules.java, a single file source program that reads the
requires clauses and prints the platform modules. It runs identically on the
Linux, macOS and Windows runners without a build step, and skips third party
requires such as jlayer, which is an automatic module and cannot be linked
into a runtime image at all, as well as test only requires and requires
static. All sixteen packaging call sites now resolve the list through it, so
they can no longer drift from the descriptor.
The jpackage Maven plugin takes its modules as individual XML elements and
cannot consume a generated value, so it remains a second copy. To keep it
honest the helper has a pom verification mode, bound to the validate phase
via exec-maven-plugin. Binding it to the build rather than to a workflow
trigger means it also fires on direct pushes to main, on tagged releases, in
both AUR PKGBUILDs and on local builds, none of which run the pull request
check.
The released AUR PKGBUILD builds from a tag tarball that may predate the
helper, and aur-publish.yml rewrites pkgver to the latest release, so it
falls back to the list carried in that tarball's own pom.xml.
Verified that the generated list produces a byte identical runtime image to
the previous hardcoded one, that removing a requires fails the build with a
precise diff, and that the pull request check and the push triggered nightly
AppImage job both succeed under act.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Generated-By: Claude Code (Claude Opus 5)
The ON4KST connection manager introduced in f8c04e7 uses
jdk.net.ExtendedSocketOptions to configure kernel side TCP keepalives. The
new dependency was added to module-info.java and to the jpackage Maven
plugin, but the CI and AUR builds do not use that plugin: they call jpackage
directly with a hardcoded --add-modules list that still lacked jdk.net.
Because jdk.net is not pulled in transitively, every packaged runtime image
shipped without the class while development runs against the full JDK and
kept working.
The resulting NoClassDefFoundError is an Error, so neither the catch in
configureSocket nor the surrounding catch (Exception) in openConnection
handled it. Running on a ScheduledExecutorService, the throwable was stored
in the task future and never surfaced, leaving the state machine stuck in
CONNECTING with no reconnect attempt and no user visible failure.
Add jdk.net to all 16 module lists, covering nightly artifacts, tagged
releases and both AUR PKGBUILDs, so releases are affected as well as
nightlies. Additionally catch LinkageError in configureSocket so a runtime
image without jdk.net degrades to application level heartbeats, and catch
Throwable in openConnection so an Error can no longer be swallowed by the
scheduler.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Issue #68
Adds Band.B_50/B_70 and wires them through the same band-opportunity
machinery as the other bands: X/a/B+/o table columns and filter button
in the User table and the Workedstn database table, NOT-QRV checkboxes
and propagation across callsign variants, "My station uses 6m/4m band"
toggles, station-name detection ("50", "6M", "70MHZ", "4M" - without
stealing the existing bare "70"/"6" cm-band shorthand), Win-Test sked
band IDs (10/50MHz, 11/70MHz), and UCX-logger worked-band recognition.
Persists worked50/70 and notQRV50/70 via an additive SQLite migration
(same ensureColumnExists pattern as the earlier v1.1->v1.2 migration),
verified against a real database file.
ReachabilityService.resolveAutoBand() now also falls back to 50 MHz
(then 70 MHz) for stations in the "50/70 MHz" chat category, mirroring
the existing Microwave-category fallback to 23cm.
Assisted by Claude Sonnet 5 <noreply@anthropic.com>
Issue #70, #66, #65
Introduces a shared BandOpportunityResolver so the station table's band
columns, the New bands filter, the band-upgrade hint, the priority
score, the map markers and the automatic reachability band selection
all derive band availability the same way: recent QRG detections (30
min window) plus station-name hints, evaluated across every active
callsignRaw variant, with manual NOT-QRV always taking precedence.
The per-band table cells now distinguish:
- X: worked on this band
- a: band available, call not worked on any band yet
- B+: band available, call already worked on another band
- o: grid square already worked on this band (any station) - combines
with the others, e.g. "ao" or "B+o"
Both "a" and "o" can be toggled off in the GUI settings tab.
Assisted by Claude Sonnet 5 <noreply@anthropic.com>
GitHub never clears the prerelease flag once a beta ships as stable,
so the latest prerelease can be older than the latest stable release.
Compare publish dates and fall back to the empty state when stable is
newer.
Assisted by Claude Sonnet 5 <noreply@anthropic.com>
* Added a map to show where other stn are // refactored message adding to tables for performance, max 30.000 msg now
* debugging map failure
* debugging map failure
* debugging map failure
* fix Pipeline Modules n map Linux
* changed UI design to save some space. Made the UI more reactive for smaller screens. Maximized resolution at startup to the possible maximum of the current display of the user
* Docs added for new Features
* Added filters and sorters for Tropo reachability, AS availability and new big fields / locators. Added reachability band selector for tropo calculation based on a choosen band
* Fixed worked-23cm-tag when working with dxlog/n1mm logger
* delete not needed files from repo
* Reduced opentopo api calls // added 4 char locator new-filter // added new band and tropo filter and priority sorter // decreased map load
---------
Co-authored-by: Marc Froehlich <praktimarc@gmail.com>
* Added a map to show where other stn are // refactored message adding to tables for performance, max 30.000 msg now
* debugging map failure
* debugging map failure
* debugging map failure
* fix Pipeline Modules n map Linux
---------
Co-authored-by: Marc Froehlich <praktimarc@gmail.com>
Fix Pipelines hopefully
again
nochmal ... :)
AAAA
BBBB
fix Flatpack from main
Flatpak again ....
flatpak github packages test
Flatpak?
flatpakref fork-ready
Install docs n fix
Desktop Entries
Rsclub2_2andClaude Opus 5.5
