feat(cluster): feed the locator store from PSK Reporter, filtered at the broker

The store shipped without its main source. Locators came only from this
station's own WSJT-X decodes, which is what the whole MQTT discussion was
about.

PSK Reporter now feeds it through a new OnGrid callback, fired before any
geographic filtering: what the store wants is "which square is this callsign
in", and that is true whoever happened to hear the report.

The subscription filters on the RECEIVER's square, a level the v2 topic
exposes. Measured on the live feed: the four opening bands unfiltered are 83
messages a second, of which roughly one in a hundred survived the NearKm test
that already existed here — the rest was received, TLS-decrypted, JSON-parsed
and discarded. One ring of squares is 0.2 to 1.2 a second.

By square rather than by DXCC, which was the obvious alternative: one country
measured 1.2 messages a second (OH) against 72.5 (K) on a single band, because
a DXCC can be a continent. By square the same measurement is 0.2 to 1.2, so the
load follows distance — what the feed is actually about — and is the same for
every operator.

Grid chasing subscribes with the "+" band wildcard, so one subscription per
square covers every band instead of one per band per square.

The store gains a source column (decode | mqtt), migrated in place on an
existing file.

app.go

@@ -41,6 +41,7 @@ import (
 	"hamlog/internal/email"
 	"hamlog/internal/extsvc"
 	"hamlog/internal/geo"
+	"hamlog/internal/gridcache"
 	"hamlog/internal/integrations/udp"
 	"hamlog/internal/lookup"
 	"hamlog/internal/lotwusers"
@@ -235,6 +236,7 @@ const (
 	keyUltrabeamStep    = "ultrabeam.step_khz"   // re-tune hysteresis: 25 | 50 | 100 kHz
 	keyMotorTrackMode   = "motor.track_mode"     // "always" | "step" | "band"
 	keyMotorBandFreqs   = "motor.band_freqs"     // per-band tune frequency: "40m=7100,20m=14150"
+	keyChaseNewGrids    = "cluster.chase_grids"  // "1" → persist learnt locators across restarts
 	keyMotorType        = "ultrabeam.type"       // "ultrabeam" | "steppir" (default ultrabeam)
 	keyMotorTransport   = "ultrabeam.transport"  // "tcp" | "serial" (default tcp)
 	keyMotorCOM         = "ultrabeam.com"        // serial device name (COM3, /dev/ttyUSB0)
@@ -611,8 +613,23 @@ type App struct {
 	//
 	// In memory only, and bounded: it is a session-local view of who is on the
 	// air now, not a database.
-	decodeGrids   map[string]string
+	//
+	// Bounded in TWO generations. The cap used to drop the whole map, which was
+	// survivable while the only source was this station's own decodes — it never
+	// reached the ceiling. It is a cliff for any larger feed: every locator in the
+	// list would vanish at once, periodically. Keeping the previous generation
+	// means a rotation costs the older half and nothing more, and it needs no
+	// insertion order, no per-entry timestamp and no bookkeeping on the write
+	// path — which "evict the oldest thousand" would all require.
+	decodeGrids    map[string]string // current generation, written to
+	decodeGridsOld map[string]string // previous generation, still readable
 	decodeGridsMu  sync.RWMutex
+	// gridStore persists the map across restarts when grid chasing is on. With
+	// it, rotation is switched OFF: rotating would drop callsigns the database
+	// still holds, and a lookup would then miss something we know. The store
+	// bounds itself by age instead, so memory follows how many distinct stations
+	// have actually been heard in two years rather than a made-up ceiling.
+	gridStore *gridcache.Store
 	// pskr is the PSK Reporter MQTT feed, up only while the opening watch is on.
 	// It is the source that makes VHF detection work at all: the cluster and RBN
 	// carry a handful of 6 m spots where PSK Reporter carries hundreds.
@@ -1375,8 +1392,11 @@ func (a *App) startup(ctx context.Context) {
 	go a.sendTelemetryHeartbeat()
 	go a.liveStatusLoop() // multi-op: heartbeat current activity to shared MySQL
 	go a.chatLoop()       // multi-op: poll the shared chat + heartbeat presence
-	// PSK Reporter, when the opening watch is on. After the operator's grid is
-	// known: without it there is no distance to measure and the feed stays down.
+	// Locator store BEFORE the feed: the feed asks whether it exists to decide
+	// which bands to subscribe to.
+	a.startGridCache()
+	// PSK Reporter. After the operator's grid is known: without it there is no
+	// distance to measure and no receiver squares to filter on, so it stays down.
 	a.startBandOpenFeed()
 	// One-time tidy-up of a field nothing used to record. Background, once.
 	a.backfillDistancesOnce()
@@ -1625,6 +1645,14 @@ func (a *App) shutdown(ctx context.Context) {
 	if a.qsoRec != nil {
 		a.qsoRec.Stop()
 	}
+	// Before the databases: Close flushes what the last minute learnt, and a
+	// restart is exactly when the grid cache is worth the most.
+	if a.gridStore != nil {
+		if err := a.gridStore.Close(); err != nil {
+			applog.Printf("gridcache: close: %v", err)
+		}
+		a.gridStore = nil
+	}
 	if a.logDb != nil && a.logDb != a.db {
 		_ = a.logDb.Close() // shared MySQL logbook (separate from the local config DB)
 	}
@@ -11937,15 +11965,7 @@ func (a *App) consumeUDPEvents() {
 			// Remember the grid before anything else: a CQ is the one message that
 			// carries it, and the station may never send another.
 			if ev.DecodeGrid != "" {
-				a.decodeGridsMu.Lock()
-				if a.decodeGrids == nil {
-					a.decodeGrids = make(map[string]string, 512)
-				}
-				if len(a.decodeGrids) > 20000 {
-					a.decodeGrids = make(map[string]string, 512) // bound a long session
-				}
-				a.decodeGrids[strings.ToUpper(ev.DecodeCall)] = ev.DecodeGrid
-				a.decodeGridsMu.Unlock()
+				a.rememberDecodeGrid(ev.DecodeCall, ev.DecodeGrid, gridcache.SourceDecode)
 			}
 			// A WSJT-X decode (heard station). Render it on the FlexRadio
 			// panadapter when the option is on; green + SNR comment, auto-expiring
@@ -17014,6 +17034,151 @@ func (a *App) clusterStatusMaps() *clusterStatusCache {
 // was ambiguous and the frontend couldn't infer) we degrade gracefully
 // to band-only — saying "worked" rather than wrongly flagging "new-slot"
 // just because we don't know the mode.
+// decodeGridsCap is how many callsigns one generation holds before it rotates.
+//
+// 100 000 measured at 82 bytes an entry — 8 MB a generation, 16 MB for both,
+// which is nothing next to what it buys: a locator on a spot instead of an empty
+// column. The old 20 000 was chosen when the only feed was this station's own
+// decodes and the ceiling was never reached anyway.
+const decodeGridsCap = 100000
+
+// GetChaseNewGrids reports whether learnt locators are kept across restarts.
+func (a *App) GetChaseNewGrids() bool { return a.settingOr(keyChaseNewGrids, "") == "1" }
+
+// GridCacheStatus is the live count under the option. A store that is on but
+// has learnt nothing looks exactly like a broken one until a number moves.
+type GridCacheStatus struct {
+	Enabled bool `json:"enabled"`
+	Known   int  `json:"known"`   // locators in memory, stored + learnt this session
+	Pending int  `json:"pending"` // waiting for the next batch write
+}
+
+// GetGridCacheStatus reports what the locator store holds.
+func (a *App) GetGridCacheStatus() GridCacheStatus {
+	out := GridCacheStatus{Enabled: a.gridStore != nil}
+	a.decodeGridsMu.RLock()
+	out.Known = len(a.decodeGrids) + len(a.decodeGridsOld)
+	a.decodeGridsMu.RUnlock()
+	if a.gridStore != nil {
+		out.Pending = a.gridStore.Pending()
+	}
+	return out
+}
+
+// SetChaseNewGrids turns grid chasing on or off and applies it immediately.
+func (a *App) SetChaseNewGrids(on bool) error {
+	a.setSetting(keyChaseNewGrids, boolStr(on))
+	a.startGridCache()
+	// Resubscribe: with grid chasing the feed covers every band, without it only
+	// the opening bands — and with neither consumer it comes down entirely.
+	a.startBandOpenFeed()
+	return nil
+}
+
+// startGridCache brings the locator store up or down to match the setting.
+//
+// Switching it ON seeds the in-memory map from the database, which is the whole
+// point: the cluster's locator column is populated in the first second instead
+// of after an hour of listening. Switching it OFF closes the file and leaves the
+// map alone — what has already been learnt this session stays usable, it simply
+// stops being remembered. Nothing is deleted: turning the option back on picks
+// up where it left off.
+func (a *App) startGridCache() {
+	if a.gridStore != nil {
+		if err := a.gridStore.Close(); err != nil {
+			applog.Printf("gridcache: close: %v", err)
+		}
+		a.gridStore = nil
+	}
+	if !a.GetChaseNewGrids() {
+		return
+	}
+	path := filepath.Join(a.dataDir, "grids.db")
+	st, err := gridcache.Open(path, applog.Printf)
+	if err != nil {
+		applog.Printf("gridcache: disabled — %v", err)
+		return
+	}
+	seed, err := st.LoadAll()
+	if err != nil {
+		applog.Printf("gridcache: cannot read %s (%v) — starting empty", path, err)
+		seed = map[string]string{}
+	}
+	a.decodeGridsMu.Lock()
+	if a.decodeGrids == nil {
+		a.decodeGrids = make(map[string]string, len(seed)+512)
+	}
+	// Seed UNDER what this session already heard: a locator decoded a minute ago
+	// is newer than one stored days back, and the newest report is the one that
+	// counts when a station has moved.
+	for call, grid := range seed {
+		if _, live := a.decodeGrids[call]; !live {
+			a.decodeGrids[call] = grid
+		}
+	}
+	n := len(a.decodeGrids)
+	a.decodeGridsMu.Unlock()
+
+	a.gridStore = st
+	st.Start(a.ctx)
+	applog.Printf("gridcache: grid chasing on — %d locators loaded from %s (%d known in total)", len(seed), path, n)
+}
+
+// rememberDecodeGrid records the grid a station announced.
+//
+// Rotation, not eviction: when the current generation fills it becomes the
+// previous one and a fresh map takes over. Nothing is scanned, nothing is
+// timestamped, and the write stays a single map assignment — which matters
+// because this runs once per decode.
+func (a *App) rememberDecodeGrid(call, grid, source string) {
+	call = strings.ToUpper(strings.TrimSpace(call))
+	grid = strings.TrimSpace(grid)
+	if call == "" || grid == "" {
+		return
+	}
+	store := a.gridStore
+
+	a.decodeGridsMu.Lock()
+	if a.decodeGrids == nil {
+		a.decodeGrids = make(map[string]string, 512)
+	}
+	// Rotate only while nothing is persisting the map. With a store the cap would
+	// discard callsigns the database still holds, so the lookup would miss what
+	// we know; age in the store is the bound instead.
+	if store == nil && len(a.decodeGrids) >= decodeGridsCap {
+		a.decodeGridsOld = a.decodeGrids
+		a.decodeGrids = make(map[string]string, 512)
+	}
+	// Only a CHANGE is worth writing. The feeds repeat themselves — the same
+	// station is reported by dozens of receivers a minute — so queueing every
+	// report would put the whole stream in the batch instead of the news in it.
+	changed := a.decodeGrids[call] != grid
+	if changed && a.decodeGridsOld != nil && a.decodeGridsOld[call] == grid {
+		// Known already, just in the older generation: promote it without calling
+		// it news.
+		changed = false
+	}
+	a.decodeGrids[call] = grid
+	a.decodeGridsMu.Unlock()
+
+	if changed && store != nil {
+		store.Put(call, grid, source)
+	}
+}
+
+// lookupDecodeGrid returns the last grid heard for a callsign, "" if unknown.
+// The previous generation is consulted second, so a station that has gone quiet
+// keeps its locator across one rotation instead of losing it at the cliff.
+func (a *App) lookupDecodeGrid(call string) string {
+	call = strings.ToUpper(call)
+	a.decodeGridsMu.RLock()
+	defer a.decodeGridsMu.RUnlock()
+	if g := a.decodeGrids[call]; g != "" {
+		return g
+	}
+	return a.decodeGridsOld[call]
+}
+
 func (a *App) ClusterSpotStatuses(spots []SpotQuery) []SpotStatus {
 	out := make([]SpotStatus, len(spots))
 	if a.qso == nil {
@@ -17102,12 +17267,10 @@ func (a *App) ClusterSpotStatuses(spots []SpotQuery) []SpotStatus {
 		// is part of the key, so the "group digital modes" option decides whether a
 		// grid worked on FT8 still counts as new on FT4 — one rule, no branch here.
 		{
-			// The length check has to be INSIDE the lock: the decode goroutine
-			// replaces this map wholesale when it grows too large, so reading len()
-			// unguarded is a race on the map header, not a cheap fast path.
-			a.decodeGridsMu.RLock()
-			g := a.decodeGrids[strings.ToUpper(q.Call)]
-			a.decodeGridsMu.RUnlock()
+			// Read through the accessor: the decode goroutine swaps these maps on
+			// rotation, so touching them unguarded is a race on the map header,
+			// not a cheap fast path.
+			g := a.lookupDecodeGrid(q.Call)
 			if g != "" {
 				out[i].Grid = g
 				cm := out[i].Mode

bandopen_sources.go

@@ -20,6 +20,8 @@ import (
 
 	"hamlog/internal/applog"
 	"hamlog/internal/bandopen"
+	"hamlog/internal/geo"
+	"hamlog/internal/gridcache"
 	"hamlog/internal/cluster"
 	"hamlog/internal/pskr"
 )
@@ -133,17 +135,45 @@ func (a *App) startBandOpenFeed() {
 		// on a timer fed by spots this path no longer looks at — so they would
 		// hang there until the app restarted.
 		a.clearBandOpenings()
+	}
+	chaseGrids := a.gridStore != nil
+	if !s.Enabled && !chaseGrids {
 		return
 	}
 	// Every spot is measured from the operator's position. Without one there is
 	// nothing to measure, and a detector fed unmeasurable spots reports nothing
 	// while looking like it is working.
 	if !a.opSet {
-		applog.Printf("bandopen: no station grid set — the opening watch needs one to measure a path")
+		applog.Printf("pskr: no station grid set — the feed needs one to measure a path")
 		return
 	}
+
+	// Grid chasing wants every band; the opening watch wants its four. "+" is the
+	// MQTT single-level wildcard, so one subscription per square covers the lot.
+	bands := s.Bands
+	if chaseGrids {
+		bands = []string{"+"}
+	}
+	// Filter at the BROKER on the receiver's square rather than receiving the
+	// world and discarding it here. Measured on the live feed: the four opening
+	// bands unfiltered are 83 messages a second, of which roughly one in a
+	// hundred survived the NearKm test below. One ring of squares — about the
+	// same 300 km — is 0.2 to 1.2 a second, and the same for every operator,
+	// where filtering by DXCC ranged from 1.2 (OH) to 72.5 (K).
+	rxGrids := geo.NeighbourGrids(a.opLat, a.opLon, 1)
+
+	var onGrid func(call, grid string)
+	if chaseGrids {
+		onGrid = func(call, grid string) { a.rememberDecodeGrid(call, grid, gridcache.SourceMQTT) }
+	}
+	var onSpot func(pskr.Spot)
+	if s.Enabled {
+		onSpot = a.feedBandOpen
+	}
+
 	a.pskr = pskr.New(pskr.Config{
-		Bands: s.Bands,
+		Bands:   bands,
+		RxGrids: rxGrids,
 		OpLat:   a.opLat, OpLon: a.opLon,
 		Geo: func(grid string) (int, int, bool) {
 			lat, lon, ok := gridToLatLon(grid)
@@ -156,12 +186,16 @@ func (a *App) startBandOpenFeed() {
 			b := int(initialBearingDeg(a.opLat, a.opLon, lat, lon) + 0.5)
 			return d, b, true
 		},
-		OnSpot: a.feedBandOpen,
+		OnSpot: onSpot,
+		OnGrid: onGrid,
 		Logf:   applog.Printf,
 	})
 	if err := a.pskr.Start(); err != nil {
-		applog.Printf("bandopen: PSK Reporter feed did not start: %v", err)
+		applog.Printf("pskr: feed did not start: %v", err)
+		return
 	}
+	applog.Printf("pskr: feed up — bands %v, %d receiver squares (openings=%v, grids=%v)",
+		bands, len(rxGrids), s.Enabled, chaseGrids)
 }
 
 // feedBandOpen hands one PSK Reporter decode to the detector.

changelog.json

@@ -1,4 +1,18 @@
 [
+  {
+    "version": "0.24.8",
+    "date": "",
+    "en": [
+      "Cluster: the grid cache now holds 100,000 callsigns and rotates instead of emptying itself, so locators stop vanishing from the list.",
+      "New option \"Chase new grids\": locators learnt from your decodes and from PSK Reporter are kept in their own database, with their source.",
+      "Band openings: the PSK Reporter feed is now filtered at the broker, which cuts it from about 83 messages a second to under two."
+    ],
+    "fr": [
+      "Cluster : le cache de locators garde 100 000 indicatifs et tourne au lieu de se vider, les locators ne disparaissent donc plus de la liste.",
+      "Nouvelle option « Chasse aux nouveaux carrés » : les locators appris de tes décodes et de PSK Reporter sont conservés dans leur propre base, avec leur source.",
+      "Ouvertures de bande : le flux PSK Reporter est désormais filtré chez le broker, ce qui le fait passer d environ 83 messages par seconde à moins de deux."
+    ]
+  },
   {
     "version": "0.24.7",
     "date": "",

decodegrids_test.go

@@ -0,0 +1,160 @@
+package main
+
+import (
+	"fmt"
+	"path/filepath"
+	"sync"
+	"testing"
+
+	"hamlog/internal/gridcache"
+)
+
+// The grid cache used to drop EVERYTHING at its ceiling. That was survivable
+// while the only feed was this station's own decodes, which never reached it.
+// Fed by anything larger it is a cliff: every locator in the cluster list
+// disappears at once, periodically, and the operator sees the column empty
+// itself for no reason.
+//
+// Rotation keeps the previous generation, so a full cache costs the older half
+// and nothing more.
+func TestDecodeGridRotationKeepsThePreviousGeneration(t *testing.T) {
+	a := &App{}
+
+	// Fill exactly one generation.
+	for i := 0; i < decodeGridsCap; i++ {
+		a.rememberDecodeGrid(fmt.Sprintf("CALL%06d", i), "JN36", gridcache.SourceDecode)
+	}
+	if got := a.lookupDecodeGrid("CALL000000"); got != "JN36" {
+		t.Fatalf("first entry lost before any rotation: %q", got)
+	}
+	if a.decodeGridsOld != nil {
+		t.Fatal("rotated early — the cap is the ceiling of ONE generation")
+	}
+
+	// One more entry rotates.
+	a.rememberDecodeGrid("NEWCALL", "IO91", gridcache.SourceDecode)
+	if a.decodeGridsOld == nil {
+		t.Fatal("did not rotate at the cap")
+	}
+	if got := a.lookupDecodeGrid("NEWCALL"); got != "IO91" {
+		t.Errorf("the entry that caused the rotation was lost: %q", got)
+	}
+	// The whole previous generation is still readable — this is the point.
+	if got := a.lookupDecodeGrid("CALL000000"); got != "JN36" {
+		t.Errorf("a locator from the previous generation was dropped: %q — that is the cliff again", got)
+	}
+	if got := a.lookupDecodeGrid("CALL099999"); got != "JN36" {
+		t.Errorf("previous generation incomplete: %q", got)
+	}
+
+	// A second rotation is what finally retires the oldest half.
+	for i := 0; i < decodeGridsCap; i++ {
+		a.rememberDecodeGrid(fmt.Sprintf("SECOND%06d", i), "KP20", gridcache.SourceDecode)
+	}
+	if got := a.lookupDecodeGrid("CALL000000"); got != "" {
+		t.Errorf("the cache is unbounded: %q survived two rotations", got)
+	}
+	if got := a.lookupDecodeGrid("NEWCALL"); got != "IO91" {
+		t.Errorf("an entry one generation old was retired too early: %q", got)
+	}
+}
+
+// Callsigns are normalised on the way in AND on the way out, or a spot for
+// "f4bpo" would miss a grid learnt as "F4BPO".
+func TestDecodeGridCaseAndBlanks(t *testing.T) {
+	a := &App{}
+	a.rememberDecodeGrid(" f4bpo ", "JN36", gridcache.SourceDecode)
+	if got := a.lookupDecodeGrid("F4BPO"); got != "JN36" {
+		t.Errorf("lookup of the upper-case form failed: %q", got)
+	}
+	a.rememberDecodeGrid("", "JN36", gridcache.SourceDecode)
+	a.rememberDecodeGrid("K1ABC", "", gridcache.SourceDecode)
+	if got := a.lookupDecodeGrid("K1ABC"); got != "" {
+		t.Errorf("stored an empty grid: %q", got)
+	}
+}
+
+// The write path runs on the decode goroutine while the cluster status builder
+// reads. Rotation swaps the map headers, so an unguarded read is a data race
+// rather than a stale value.
+//
+// This build is CGO-free, so -race is not available here; the test exercises the
+// interleaving and would fault on a concurrent map access even without it.
+func TestDecodeGridConcurrentAccess(t *testing.T) {
+	a := &App{}
+	var wg sync.WaitGroup
+	wg.Add(2)
+	go func() {
+		defer wg.Done()
+		for i := 0; i < 20000; i++ {
+			a.rememberDecodeGrid(fmt.Sprintf("W%05d", i), "FN31", gridcache.SourceDecode)
+		}
+	}()
+	go func() {
+		defer wg.Done()
+		for i := 0; i < 20000; i++ {
+			_ = a.lookupDecodeGrid(fmt.Sprintf("W%05d", i))
+		}
+	}()
+	wg.Wait()
+}
+
+// Only a CHANGE may reach the write batch.
+//
+// The feeds repeat themselves — the same station is reported by dozens of
+// receivers a minute — so queueing every report would put the whole stream in
+// the batch instead of the news in it, and turn a cache into a write amplifier.
+func TestOnlyChangesAreQueued(t *testing.T) {
+	st, err := gridcache.Open(filepath.Join(t.TempDir(), "grids.db"), nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer st.Close()
+	a := &App{gridStore: st}
+
+	a.rememberDecodeGrid("F4BPO", "JN36", gridcache.SourceDecode)
+	if n := st.Pending(); n != 1 {
+		t.Fatalf("a new locator queued %d writes, want 1", n)
+	}
+	if err := st.Flush(); err != nil {
+		t.Fatal(err)
+	}
+
+	// The same report, a hundred times over, is not news.
+	for i := 0; i < 100; i++ {
+		a.rememberDecodeGrid("F4BPO", "JN36", gridcache.SourceDecode)
+	}
+	if n := st.Pending(); n != 0 {
+		t.Errorf("unchanged reports queued %d writes — the batch would carry the whole feed", n)
+	}
+
+	// A station that moved is.
+	a.rememberDecodeGrid("F4BPO", "KP30", gridcache.SourceDecode)
+	if n := st.Pending(); n != 1 {
+		t.Errorf("a changed locator queued %d writes, want 1", n)
+	}
+	if got := a.lookupDecodeGrid("F4BPO"); got != "KP30" {
+		t.Errorf("the map kept the old locator: %q", got)
+	}
+}
+
+// Rotation must be OFF while a store is attached: it would discard callsigns the
+// database still holds, and the lookup would then miss something we know.
+func TestNoRotationWhilePersisting(t *testing.T) {
+	st, err := gridcache.Open(filepath.Join(t.TempDir(), "grids.db"), nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer st.Close()
+	a := &App{gridStore: st}
+
+	for i := 0; i < decodeGridsCap+10; i++ {
+		a.rememberDecodeGrid(fmt.Sprintf("CALL%06d", i), "JN36", gridcache.SourceDecode)
+	}
+	if a.decodeGridsOld != nil {
+		t.Error("rotated while a store was attached — locators the database holds would go missing")
+	}
+	if got := a.lookupDecodeGrid("CALL000000"); got != "JN36" {
+		t.Errorf("the first entry was dropped: %q", got)
+	}
+}

frontend/src/components/SettingsModal.tsx

@@ -52,7 +52,7 @@ import {
   GetADIFMonitor, SaveADIFMonitor, PickADIFMonitorFile,
   GetRelayAuto, SaveRelayAuto, GetStationDevices,
   GetAwardDefs, GetTrackedAwards, SaveTrackedAwards,
-  GetBandOpenSettings, SaveBandOpenSettings, GetPSKReporterStatus,
+  GetBandOpenSettings, SaveBandOpenSettings, GetPSKReporterStatus, GetChaseNewGrids, SetChaseNewGrids, GetGridCacheStatus,
 } from '../../wailsjs/go/main/App';
 import type { profile as profileModels } from '../../wailsjs/go/models';
 import type { LookupSettingsForm, StationSettingsForm, ListsSettingsForm, ModePresetForm } from '@/types';
@@ -1552,6 +1552,8 @@ export function SettingsModal({ onClose, onSaved, initialSection, onMainPaneChan
   // has side effects there — adding the RBN nodes, bringing the PSK Reporter
   // feed up or down — so the write has to go where those live.
   const [bandOpen, setBandOpen] = useState<any>({ enabled: false, bands: [], available: [] });
+  const [chaseGrids, setChaseGrids] = useState(false);
+  const [gridStat, setGridStat] = useState<any>(null);
   const [pskrStatus, setPskrStatus] = useState<any>(null);
   const saveBandOpen = async (next: any) => {
     setBandOpen(next);
@@ -1560,11 +1562,13 @@ export function SettingsModal({ onClose, onSaved, initialSection, onMainPaneChan
   useEffect(() => {
     (async () => {
       try { setBandOpen(await GetBandOpenSettings()); } catch { /* defaults stand */ }
+      try { setChaseGrids(await GetChaseNewGrids()); } catch { /* defaults stand */ }
     })();
     // Poll the feed while the panel is open: a live count is the only thing that
     // distinguishes "connected" from "connected and receiving nothing".
     const t = window.setInterval(async () => {
       try { setPskrStatus(await GetPSKReporterStatus()); } catch { /* ignore */ }
+      try { setGridStat(await GetGridCacheStatus()); } catch { /* ignore */ }
     }, 3000);
     return () => window.clearInterval(t);
   }, []);
@@ -4227,6 +4231,23 @@ export function SettingsModal({ onClose, onSaved, initialSection, onMainPaneChan
               things set up once. A preferences dialog you reopen every ten minutes
               is a filter in the wrong place. */}
 
+          {/* Grid chasing. Here rather than in the filter panel because it is set
+              up once: it decides whether locators learnt from decodes are KEPT
+              across restarts, not what the list shows right now. */}
+          <div className="border-t border-border/60 pt-3 space-y-2">
+            <label className="flex items-start gap-2 text-sm cursor-pointer">
+              <Checkbox checked={chaseGrids} className="mt-0.5"
+                onCheckedChange={(c) => { setChaseGrids(!!c); SetChaseNewGrids(!!c).catch(() => {}); }} />
+              <span>{t('clu.chaseGrids')} <span className="text-xs text-muted-foreground">{t('clu.chaseGridsHint')}</span></span>
+            </label>
+            {chaseGrids && (
+              <p className="pl-6 text-xs text-muted-foreground">
+                {t('clu.chaseGridsStat', { n: gridStat?.known ?? 0, p: gridStat?.pending ?? 0 })}
+                {pskrStatus?.running ? ` · ${t('bo.feedUp', { n: pskrStatus.received ?? 0 })}` : ''}
+              </p>
+            )}
+          </div>
+
           {/* Band-opening watch. It lives HERE, with the cluster nodes, because
               switching it on adds two of them — the operator should see that
               happen where it happens rather than find nodes they did not add. */}

frontend/src/lib/i18n.tsx

@@ -270,7 +270,7 @@ const en: Dict = {
   'clu.muteWorkedHint': '(they stay in the list, just quiet — leaves the colour for what is left to do)',
   'clu.slotHighlight': 'Colour the stations not worked on this band and mode',
   'clu.slotHighlightHint': '(by callsign, whatever the entity status says)',
-  'bo.open': 'open', 'bo.liveTip': '{band} is open — {n} stations, ~{km} km, {sector}{season}. Click for the band map.', 'bo.enable': 'Watch for band openings', 'bo.enableHint': '(10, 12, 6, 4 and 2 m. Switching this on adds the two RBN nodes and subscribes to the PSK Reporter feed — the detection needs far more ears than a cluster can give it.)', 'bo.feedUp': 'PSK Reporter feed up — {n} decodes seen', 'bo.feedDown': 'PSK Reporter feed down — needs your station grid, and a moment to connect', 'clu.workedSameSlot': 'Already worked only on the same slot',
+  'bo.open': 'open', 'bo.liveTip': '{band} is open — {n} stations, ~{km} km, {sector}{season}. Click for the band map.', 'bo.enable': 'Watch for band openings', 'bo.enableHint': '(10, 12, 6, 4 and 2 m. Switching this on adds the two RBN nodes and subscribes to the PSK Reporter feed — the detection needs far more ears than a cluster can give it.)', 'bo.feedUp': 'PSK Reporter feed up — {n} decodes seen', 'bo.feedDown': 'PSK Reporter feed down — needs your station grid, and a moment to connect', 'clu.chaseGrids': 'Chase new grids', 'clu.chaseGridsHint': '(learns locators from your own WSJT-X decodes AND from PSK Reporter, and keeps them in their own database so the cluster shows them from the first second)', 'clu.chaseGridsStat': '{n} locators known — {p} waiting to be written', 'clu.workedSameSlot': 'Already worked only on the same slot',
   'clu.workedSameSlotHint': '— a spot shows "worked" only if you worked that call on the SAME band and mode, not just anywhere. Combines with digital-mode grouping (Settings → General): with it on, a call worked on 20m FT8 also counts as worked for a 20m FT4 spot; with it off, FT8 and FT4 are separate slots.',
   // Backup panel
   'bk.hintMysql': 'On close (once/day) OpsLog snapshots the local SQLite (config) AND exports the shared MySQL log to ADIF — opslog-log-<date>.adi — so your contacts are protected even though they live on the server. Rotation keeps the last N of each.',

frontend/wailsjs/go/main/App.d.ts

@@ -404,6 +404,8 @@ export function GetCatalogCodes():Promise<Array<string>>;
 
 export function GetChangelog():Promise<Array<main.ChangelogEntry>>;
 
+export function GetChaseNewGrids():Promise<boolean>;
+
 export function GetChatHistory(arg1:number):Promise<Array<main.ChatMessage>>;
 
 export function GetClublogCtyInfo():Promise<main.ClublogCtyInfo>;
@@ -440,6 +442,8 @@ export function GetFlexBandPower():Promise<Record<string, main.FlexBandPower>>;
 
 export function GetFlexState():Promise<cat.FlexTXState>;
 
+export function GetGridCacheStatus():Promise<main.GridCacheStatus>;
+
 export function GetIcomState():Promise<cat.IcomTXState>;
 
 export function GetListsSettings():Promise<main.ListsSettings>;
@@ -970,6 +974,8 @@ export function SetCIVTrace(arg1:boolean):Promise<void>;
 
 export function SetCWDecoderPitch(arg1:number):Promise<void>;
 
+export function SetChaseNewGrids(arg1:boolean):Promise<void>;
+
 export function SetClublogCtyEnabled(arg1:boolean):Promise<void>;
 
 export function SetClublogMostWantedEnabled(arg1:boolean):Promise<void>;

frontend/wailsjs/go/main/App.js

@@ -750,6 +750,10 @@ export function GetChangelog() {
   return window['go']['main']['App']['GetChangelog']();
 }
 
+export function GetChaseNewGrids() {
+  return window['go']['main']['App']['GetChaseNewGrids']();
+}
+
 export function GetChatHistory(arg1) {
   return window['go']['main']['App']['GetChatHistory'](arg1);
 }
@@ -822,6 +826,10 @@ export function GetFlexState() {
   return window['go']['main']['App']['GetFlexState']();
 }
 
+export function GetGridCacheStatus() {
+  return window['go']['main']['App']['GetGridCacheStatus']();
+}
+
 export function GetIcomState() {
   return window['go']['main']['App']['GetIcomState']();
 }
@@ -1882,6 +1890,10 @@ export function SetCWDecoderPitch(arg1) {
   return window['go']['main']['App']['SetCWDecoderPitch'](arg1);
 }
 
+export function SetChaseNewGrids(arg1) {
+  return window['go']['main']['App']['SetChaseNewGrids'](arg1);
+}
+
 export function SetClublogCtyEnabled(arg1) {
   return window['go']['main']['App']['SetClublogCtyEnabled'](arg1);
 }

frontend/wailsjs/go/models.ts

@@ -2419,6 +2419,22 @@ export namespace main {
 	        this.body = source["body"];
 	    }
 	}
+	export class GridCacheStatus {
+	    enabled: boolean;
+	    known: number;
+	    pending: number;
+	
+	    static createFrom(source: any = {}) {
+	        return new GridCacheStatus(source);
+	    }
+	
+	    constructor(source: any = {}) {
+	        if ('string' === typeof source) source = JSON.parse(source);
+	        this.enabled = source["enabled"];
+	        this.known = source["known"];
+	        this.pending = source["pending"];
+	    }
+	}
 	export class ModePreset {
 	    name: string;
 	    default_rst_sent?: string;

internal/geo/geo.go

@@ -67,3 +67,54 @@ func DistanceBetweenGrids(a, b string) (km float64, ok bool) {
 	}
 	return HaversineKm(lat1, lon1, lat2, lon2), true
 }
+
+// LatLonToGrid returns the 4-character Maidenhead square for a position.
+func LatLonToGrid(lat, lon float64) string {
+	lon = math.Mod(lon+180, 360)
+	if lon < 0 {
+		lon += 360
+	}
+	lat = lat + 90
+	if lat < 0 {
+		lat = 0
+	} else if lat > 180 {
+		lat = 180
+	}
+	return string([]byte{
+		byte('A' + int(lon/20)),
+		byte('A' + int(lat/10)),
+		byte('0' + int(math.Mod(lon, 20)/2)),
+		byte('0' + int(math.Mod(lat, 10)/1)),
+	})
+}
+
+// NeighbourGrids returns the square holding (lat, lon) and the ring of squares
+// around it — 9 squares for ring 1, 25 for ring 2.
+//
+// Used to filter the PSK Reporter feed at the BROKER rather than in OpsLog. A
+// square is about 111 km tall and 150 km wide at mid latitudes, so one ring is
+// roughly the 300 km "around here" the feed already meant, and the traffic that
+// used to be received and discarded is never sent.
+func NeighbourGrids(lat, lon float64, ring int) []string {
+	if ring < 0 {
+		ring = 0
+	}
+	seen := map[string]bool{}
+	out := []string{}
+	for dLat := -ring; dLat <= ring; dLat++ {
+		for dLon := -ring; dLon <= ring; dLon++ {
+			// One square step: 1° of latitude, 2° of longitude.
+			la := lat + float64(dLat)
+			lo := lon + float64(dLon)*2
+			if la > 90 || la < -90 {
+				continue // past a pole there is no square, not a wrapped one
+			}
+			g := LatLonToGrid(la, lo)
+			if !seen[g] {
+				seen[g] = true
+				out = append(out, g)
+			}
+		}
+	}
+	return out
+}

internal/geo/neighbour_test.go

@@ -0,0 +1,49 @@
+package geo
+
+import "testing"
+
+// The squares the PSK Reporter feed is filtered on. A wrong ring means either
+// receiving the world again or hearing nothing.
+func TestNeighbourGrids(t *testing.T) {
+	// JN36 is around 46.5N 5.5E.
+	lat, lon, ok := GridToLatLon("JN36")
+	if !ok {
+		t.Fatal("JN36 did not resolve")
+	}
+	if got := LatLonToGrid(lat, lon); got != "JN36" {
+		t.Fatalf("round trip gave %q, want JN36", got)
+	}
+
+	ring := NeighbourGrids(lat, lon, 1)
+	if len(ring) != 9 {
+		t.Errorf("ring 1 has %d squares, want 9: %v", len(ring), ring)
+	}
+	found := false
+	for _, g := range ring {
+		if g == "JN36" {
+			found = true
+		}
+		if len(g) != 4 {
+			t.Errorf("not a 4-character square: %q", g)
+		}
+	}
+	if !found {
+		t.Errorf("the operator's own square is missing from %v", ring)
+	}
+	if n := len(NeighbourGrids(lat, lon, 0)); n != 1 {
+		t.Errorf("ring 0 has %d squares, want just the operator's", n)
+	}
+	if n := len(NeighbourGrids(lat, lon, 2)); n != 25 {
+		t.Errorf("ring 2 has %d squares, want 25", n)
+	}
+}
+
+// Near a pole a step north has nowhere to go; it must be dropped, not wrapped
+// onto a square on the far side of the world.
+func TestNeighbourGridsNearThePole(t *testing.T) {
+	for _, g := range NeighbourGrids(89.5, 25, 1) {
+		if len(g) != 4 {
+			t.Errorf("bad square near the pole: %q", g)
+		}
+	}
+}

internal/gridcache/gridcache.go

@@ -0,0 +1,222 @@
+// Package gridcache is the long-term callsign→grid store behind grid chasing.
+//
+// Its own SQLite file, not a table in the settings database: that one sits
+// wherever the operator put it, often a synchronised folder, and this rewrites
+// itself every minute. Deleting the file costs a few days of listening.
+//
+// A callsign has one grid and the newest report wins — a stale locator reads as
+// a square already worked, which is worse than none.
+package gridcache
+
+import (
+	"context"
+	"database/sql"
+	"fmt"
+	"strings"
+	"sync"
+	"time"
+
+	_ "modernc.org/sqlite"
+)
+
+// Retention bounds a store that would otherwise only ever grow. Two years is
+// chosen to be far longer than any propagation interest and short enough that a
+// reassigned callsign eventually stops carrying its previous holder's square —
+// the one way this cache can be actively wrong rather than merely empty.
+const Retention = 2 * 365 * 24 * time.Hour
+
+// FlushEvery is the batch interval. Long enough that a burst of reports costs
+// one transaction, short enough that a crash loses a minute of learning.
+const FlushEvery = 60 * time.Second
+
+// Sources a locator can come from.
+const (
+	SourceDecode = "decode" // a CQ this station's own receiver decoded
+	SourceMQTT   = "mqtt"   // a PSK Reporter report
+)
+
+// Entry is one locator and where it came from.
+type Entry struct {
+	Grid   string
+	Source string
+}
+
+type Store struct {
+	db *sql.DB
+
+	mu    sync.Mutex
+	dirty map[string]Entry // call → what to write
+
+	stop     chan struct{}
+	stopOnce sync.Once
+	wg       sync.WaitGroup
+
+	logf func(string, ...any)
+}
+
+// Open creates or opens the store at path and prunes what has aged out.
+func Open(path string, logf func(string, ...any)) (*Store, error) {
+	if logf == nil {
+		logf = func(string, ...any) {}
+	}
+	// WAL so a flush never blocks a read, and a busy timeout because the flush
+	// goroutine and the startup load can overlap on a slow disk.
+	db, err := sql.Open("sqlite", path+"?_pragma=journal_mode(WAL)&_pragma=busy_timeout(5000)")
+	if err != nil {
+		return nil, fmt.Errorf("gridcache: open %s: %w", path, err)
+	}
+	if _, err := db.Exec("CREATE TABLE IF NOT EXISTS grids (" +
+		"call TEXT PRIMARY KEY, grid TEXT NOT NULL, updated_at INTEGER NOT NULL, " +
+		"source TEXT NOT NULL DEFAULT '')"); err != nil {
+		db.Close()
+		return nil, fmt.Errorf("gridcache: schema: %w", err)
+	}
+	// A file written before the column existed keeps its rows; this fails
+	// harmlessly when the column is already there.
+	db.Exec("ALTER TABLE grids ADD COLUMN source TEXT NOT NULL DEFAULT ''")
+	s := &Store{db: db, dirty: map[string]Entry{}, stop: make(chan struct{}), logf: logf}
+	if n, err := s.prune(); err != nil {
+		s.logf("gridcache: prune failed: %v", err)
+	} else if n > 0 {
+		s.logf("gridcache: pruned %d locators not heard in %d days", n, int(Retention.Hours()/24))
+	}
+	return s, nil
+}
+
+func (s *Store) prune() (int64, error) {
+	cut := time.Now().Add(-Retention).Unix()
+	res, err := s.db.Exec(`DELETE FROM grids WHERE updated_at < ?`, cut)
+	if err != nil {
+		return 0, err
+	}
+	return res.RowsAffected()
+}
+
+// LoadAll returns every stored locator, for seeding the in-memory map at
+// startup. One query and one map build — the point of the whole package is that
+// nothing afterwards has to ask the database anything.
+func (s *Store) LoadAll() (map[string]string, error) {
+	rows, err := s.db.Query(`SELECT call, grid FROM grids`)
+	if err != nil {
+		return nil, fmt.Errorf("gridcache: load: %w", err)
+	}
+	defer rows.Close()
+	out := make(map[string]string, 4096)
+	for rows.Next() {
+		var call, grid string
+		if err := rows.Scan(&call, &grid); err != nil {
+			return nil, err
+		}
+		out[call] = grid
+	}
+	return out, rows.Err()
+}
+
+// Put queues a locator for writing. Callers pass only what CHANGED — an
+// unchanged report is the common case by a wide margin and must not reach here,
+// or the batch would carry the whole feed instead of the news in it.
+func (s *Store) Put(call, grid, source string) {
+	call = strings.ToUpper(strings.TrimSpace(call))
+	grid = strings.TrimSpace(grid)
+	if call == "" || grid == "" {
+		return
+	}
+	s.mu.Lock()
+	s.dirty[call] = Entry{Grid: grid, Source: source}
+	s.mu.Unlock()
+}
+
+// Start runs the flush loop until Close.
+func (s *Store) Start(ctx context.Context) {
+	s.wg.Add(1)
+	go func() {
+		defer s.wg.Done()
+		t := time.NewTicker(FlushEvery)
+		defer t.Stop()
+		for {
+			select {
+			case <-s.stop:
+				return
+			case <-ctx.Done():
+				return
+			case <-t.C:
+				if err := s.Flush(); err != nil {
+					s.logf("gridcache: flush failed: %v", err)
+				}
+			}
+		}
+	}()
+}
+
+// Flush writes the pending batch in one transaction. Safe to call with nothing
+// pending, which is most of the time on a quiet band.
+func (s *Store) Flush() error {
+	s.mu.Lock()
+	if len(s.dirty) == 0 {
+		s.mu.Unlock()
+		return nil
+	}
+	batch := s.dirty
+	s.dirty = map[string]Entry{}
+	s.mu.Unlock()
+
+	tx, err := s.db.Begin()
+	if err != nil {
+		s.requeue(batch)
+		return err
+	}
+	st, err := tx.Prepare(`INSERT INTO grids (call, grid, updated_at, source) VALUES (?, ?, ?, ?)
+		ON CONFLICT(call) DO UPDATE SET grid = excluded.grid, updated_at = excluded.updated_at,
+		source = excluded.source`)
+	if err != nil {
+		tx.Rollback()
+		s.requeue(batch)
+		return err
+	}
+	defer st.Close()
+	now := time.Now().Unix()
+	for call, e := range batch {
+		if _, err := st.Exec(call, e.Grid, now, e.Source); err != nil {
+			tx.Rollback()
+			s.requeue(batch)
+			return err
+		}
+	}
+	if err := tx.Commit(); err != nil {
+		s.requeue(batch)
+		return err
+	}
+	return nil
+}
+
+// requeue puts a failed batch back, without overwriting anything learnt while it
+// was in flight — the newer value is the right one.
+func (s *Store) requeue(batch map[string]Entry) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	for call, e := range batch {
+		if _, newer := s.dirty[call]; !newer {
+			s.dirty[call] = e
+		}
+	}
+}
+
+// Pending reports how many locators are waiting to be written.
+func (s *Store) Pending() int {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return len(s.dirty)
+}
+
+// Close stops the loop and writes whatever is pending. A restart is the moment
+// the cache is most valuable, so losing the last minute of learning to a clean
+// shutdown would be a poor trade.
+func (s *Store) Close() error {
+	s.stopOnce.Do(func() { close(s.stop) })
+	s.wg.Wait()
+	err := s.Flush()
+	if cerr := s.db.Close(); err == nil {
+		err = cerr
+	}
+	return err
+}

internal/gridcache/gridcache_test.go

@@ -0,0 +1,153 @@
+package gridcache
+
+import (
+	"path/filepath"
+	"testing"
+	"time"
+)
+
+func open(t *testing.T) (*Store, string) {
+	t.Helper()
+	path := filepath.Join(t.TempDir(), "grids.db")
+	s, err := Open(path, nil)
+	if err != nil {
+		t.Fatalf("open: %v", err)
+	}
+	return s, path
+}
+
+// The whole reason the store exists: what was learnt is still there after a
+// restart, so the cluster's locator column is full in the first second instead
+// of after an hour of listening.
+func TestSurvivesRestart(t *testing.T) {
+	s, path := open(t)
+	s.Put("F4BPO", "JN36", SourceDecode)
+	s.Put("OH5CX", "KP30", SourceDecode)
+	if err := s.Flush(); err != nil {
+		t.Fatalf("flush: %v", err)
+	}
+	if err := s.Close(); err != nil {
+		t.Fatalf("close: %v", err)
+	}
+
+	again, err := Open(path, nil)
+	if err != nil {
+		t.Fatalf("reopen: %v", err)
+	}
+	defer again.Close()
+	got, err := again.LoadAll()
+	if err != nil {
+		t.Fatalf("load: %v", err)
+	}
+	if got["F4BPO"] != "JN36" || got["OH5CX"] != "KP30" {
+		t.Errorf("locators lost across a restart: %v", got)
+	}
+}
+
+// A callsign has ONE grid and the newest report wins. Operators move, go
+// portable, go on expedition — and a stale locator is worse than none for grid
+// chasing, because it reads as a square already worked.
+func TestNewestReportWins(t *testing.T) {
+	s, _ := open(t)
+	defer s.Close()
+
+	s.Put("F4BPO", "JN36", SourceDecode)
+	if err := s.Flush(); err != nil {
+		t.Fatal(err)
+	}
+	s.Put("F4BPO", "KP30", SourceDecode) // moved
+	if err := s.Flush(); err != nil {
+		t.Fatal(err)
+	}
+
+	got, err := s.LoadAll()
+	if err != nil {
+		t.Fatal(err)
+	}
+	if got["F4BPO"] != "KP30" {
+		t.Errorf("grid = %q, want the newer KP30", got["F4BPO"])
+	}
+	if len(got) != 1 {
+		t.Errorf("a callsign must hold one row, got %d: %v", len(got), got)
+	}
+}
+
+// Nothing reaches the disk until a flush, and a flush with nothing pending is
+// not an error — that is most minutes on a quiet band.
+func TestBatching(t *testing.T) {
+	s, _ := open(t)
+	defer s.Close()
+
+	for _, c := range []string{"A1AA", "B2BB", "C3CC"} {
+		s.Put(c, "JN36", SourceDecode)
+	}
+	if n := s.Pending(); n != 3 {
+		t.Errorf("pending = %d, want 3 queued and unwritten", n)
+	}
+	if got, _ := s.LoadAll(); len(got) != 0 {
+		t.Errorf("wrote before the flush: %v", got)
+	}
+	if err := s.Flush(); err != nil {
+		t.Fatal(err)
+	}
+	if n := s.Pending(); n != 0 {
+		t.Errorf("pending = %d after a flush, want 0", n)
+	}
+	if got, _ := s.LoadAll(); len(got) != 3 {
+		t.Errorf("flush wrote %d rows, want 3", len(got))
+	}
+	if err := s.Flush(); err != nil {
+		t.Errorf("empty flush must be a no-op, got %v", err)
+	}
+}
+
+// Age is what bounds a store that would otherwise only grow. A callsign not
+// heard in two years is likely reassigned, and carrying its previous holder's
+// square is the one way this cache can be actively wrong rather than empty.
+func TestPruneOnOpen(t *testing.T) {
+	s, path := open(t)
+	s.Put("FRESH", "JN36", SourceDecode)
+	s.Put("STALE", "IO91", SourceDecode)
+	if err := s.Flush(); err != nil {
+		t.Fatal(err)
+	}
+	// Backdate one row past the retention window.
+	old := time.Now().Add(-Retention - 24*time.Hour).Unix()
+	if _, err := s.db.Exec(`UPDATE grids SET updated_at = ? WHERE call = 'STALE'`, old); err != nil {
+		t.Fatal(err)
+	}
+	s.Close()
+
+	again, err := Open(path, nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer again.Close()
+	got, _ := again.LoadAll()
+	if _, ok := got["STALE"]; ok {
+		t.Error("an entry past the retention window survived — the store is unbounded")
+	}
+	if got["FRESH"] != "JN36" {
+		t.Error("pruning took a live entry with it")
+	}
+}
+
+// Close has to write what the last minute learnt: a restart is exactly when the
+// cache is worth the most, so losing it to a clean shutdown would be a poor
+// trade.
+func TestCloseFlushes(t *testing.T) {
+	s, path := open(t)
+	s.Put("LATE", "JN36", SourceDecode)
+	if err := s.Close(); err != nil {
+		t.Fatalf("close: %v", err)
+	}
+	again, err := Open(path, nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer again.Close()
+	got, _ := again.LoadAll()
+	if got["LATE"] != "JN36" {
+		t.Error("what was pending at shutdown was dropped")
+	}
+}

internal/pskr/pskr.go

@@ -81,6 +81,15 @@ type Config struct {
 	// OnSpot receives every accepted decode. Called from the MQTT goroutine, so
 	// it must not block: the broker's buffer is what pays for it if it does.
 	OnSpot func(Spot)
+	// OnGrid receives the transmitter of EVERY message, before any geographic
+	// filtering, for the callsign-to-locator store. Same goroutine as OnSpot and
+	// the same rule: do not block.
+	OnGrid func(call, grid string)
+	// RxGrids filters at the BROKER: only reports collected by a receiver in one
+	// of these squares are sent at all. Empty keeps the old behaviour, which was
+	// to receive the world and discard it here — measured at 83 messages a second
+	// for the four opening bands, of which about one in a hundred survived.
+	RxGrids []string
 	Logf    func(string, ...any)
 }
 
@@ -115,6 +124,32 @@ func New(cfg Config) *Watcher {
 	return &Watcher{cfg: cfg}
 }
 
+// topics builds the subscription list.
+//
+// The v2 topic is
+//
+//	pskr/filter/v2/<band>/<mode>/<tx call>/<rx call>/<tx grid>/<rx grid>/<tx dxcc>/<rx dxcc>
+//
+// so the receiver's square is a level the broker can filter on, and a band of
+// "+" means every band. Filtering by RECEIVER square rather than by receiver
+// DXCC is deliberate: measured on 20 m, one country ranged from 1.2 messages a
+// second (OH) to 72.5 (K), because a DXCC can be a continent. By square the
+// same measurement is 0.2 to 1.2 — the load follows distance, which is what the
+// feed is actually about, and it is the same for every operator.
+func (w *Watcher) topics() []string {
+	out := []string{}
+	for _, b := range w.cfg.Bands {
+		if len(w.cfg.RxGrids) == 0 {
+			out = append(out, "pskr/filter/v2/"+b+"/#")
+			continue
+		}
+		for _, g := range w.cfg.RxGrids {
+			out = append(out, "pskr/filter/v2/"+b+"/+/+/+/+/"+strings.ToUpper(g)+"/+/+")
+		}
+	}
+	return out
+}
+
 // Start connects and subscribes. Safe to call when already running.
 func (w *Watcher) Start() error {
 	w.mu.Lock()
@@ -143,10 +178,7 @@ func (w *Watcher) Start() error {
 
 	opts.OnConnect = func(c mqtt.Client) {
 		w.cfg.Logf("pskr: connected to %s", w.cfg.Broker)
-		for _, b := range w.cfg.Bands {
-			// Every mode, every pair of stations, on this band. That firehose IS
-			// the point: the detector's job is to find the shape in it.
-			topic := "pskr/filter/v2/" + b + "/#"
+		for _, topic := range w.topics() {
 			if tok := c.Subscribe(topic, 0, w.handle); tok.Wait() && tok.Error() != nil {
 				w.cfg.Logf("pskr: subscribe %s failed: %v", topic, tok.Error())
 				continue
@@ -206,6 +238,13 @@ func (w *Watcher) handle(_ mqtt.Client, m mqtt.Message) {
 		return
 	}
 
+	// The locator store takes every transmitter, before any of the geography
+	// below. What it wants is "which square is this callsign in", and that is
+	// true whoever happened to hear the report.
+	if w.cfg.OnGrid != nil {
+		w.cfg.OnGrid(call, grid[:4])
+	}
+
 	// THE RECEIVER HAS TO BE NEAR THE OPERATOR. This is the whole difference
 	// between a useful feed and a world map.
 	//

internal/pskr/topics_test.go

@@ -0,0 +1,46 @@
+package pskr
+
+import (
+	"strings"
+	"testing"
+)
+
+// The receiver square is a level the BROKER can filter on, which is the whole
+// point: measured on the live feed, the four opening bands unfiltered are 83
+// messages a second of which about one in a hundred survives the NearKm test.
+// One ring of squares is under two a second, and the same for every operator —
+// where filtering by DXCC ranged from 1.2 (OH) to 72.5 (K) on one band.
+func TestTopicsFilterOnTheReceiverSquare(t *testing.T) {
+	w := New(Config{Bands: []string{"6m"}, RxGrids: []string{"jn36", "JN37"}})
+	got := w.topics()
+	if len(got) != 2 {
+		t.Fatalf("want one subscription per band × square, got %v", got)
+	}
+	// Level order: band/mode/txcall/rxcall/txgrid/RXGRID/txdxcc/rxdxcc
+	want := "pskr/filter/v2/6m/+/+/+/+/JN36/+/+"
+	if got[0] != want {
+		t.Errorf("topic = %q, want %q", got[0], want)
+	}
+	if !strings.Contains(got[1], "/JN37/") {
+		t.Errorf("square not upper-cased into the topic: %q", got[1])
+	}
+}
+
+// With no squares the old behaviour stands: receive the band and decide here.
+func TestTopicsWithoutSquares(t *testing.T) {
+	w := New(Config{Bands: []string{"10m", "2m"}})
+	got := w.topics()
+	if len(got) != 2 || got[0] != "pskr/filter/v2/10m/#" {
+		t.Errorf("unfiltered topics = %v", got)
+	}
+}
+
+// Grid chasing wants every band. "+" is the MQTT single-level wildcard, so one
+// subscription per square covers the lot instead of one per band per square.
+func TestTopicsAllBands(t *testing.T) {
+	w := New(Config{Bands: []string{"+"}, RxGrids: []string{"JN36"}})
+	got := w.topics()
+	if len(got) != 1 || got[0] != "pskr/filter/v2/+/+/+/+/+/JN36/+/+" {
+		t.Errorf("all-band topic = %v", got)
+	}
+}