Handle dashboard promises and stop shadowing min/max builtins

go.mod

@@ -1,6 +1,6 @@
 module github.com/larslaskowski/pimonitor
 
-go 1.26.7
+go 1.26.9
 
 require gopkg.in/yaml.v3 v3.0.1
 

internal/alert/notify.go

@@ -414,14 +414,15 @@ func formatMessage(ev Event) string {
 		ev.Metric, resource, ev.Kind, ev.From, ev.To, ev.Value)
 }
 
-// eventReaches reports whether an event's severity reaches at least min. An
-// event is relevant to a webhook when either side of the transition is at or
-// above the webhook's minimum level, so a "min_level: crit" webhook still
-// receives the cleared event when a metric drops out of crit.
-func eventReaches(ev Event, min Level) bool {
+// eventReaches reports whether an event's severity reaches at least
+// minLevel. An event is relevant to a webhook when either side of the
+// transition is at or above the webhook's minimum level, so a
+// "min_level: crit" webhook still receives the cleared event when a metric
+// drops out of crit.
+func eventReaches(ev Event, minLevel Level) bool {
 	peak := severity(ev.From)
 	if s := severity(ev.To); s > peak {
 		peak = s
 	}
-	return peak >= severity(min)
+	return peak >= severity(minLevel)
 }

internal/httpapi/handlers.go

@@ -26,7 +26,7 @@ func writeJSON(w http.ResponseWriter, log func(msg string, args ...any), v any)
 // Config{} directly; production always sets it from
 // config.Config.HealthzMaxStaleness(...), which never returns zero.
 func (s *Server) handleHealthz(w http.ResponseWriter, _ *http.Request) {
-	if max := s.cfg.HealthzMaxStaleness; max > 0 && time.Since(s.metrics.Snapshot().Timestamp) > max {
+	if maxStale := s.cfg.HealthzMaxStaleness; maxStale > 0 && time.Since(s.metrics.Snapshot().Timestamp) > maxStale {
 		http.Error(w, "stale: latest snapshot is older than the configured healthz_max_staleness_seconds", http.StatusServiceUnavailable)
 		return
 	}

internal/web/assets/app.js

@@ -1197,10 +1197,12 @@
         return;
       }
       // Refresh immediately on return so the user never looks at a stale
-      // card while waiting for the first interval to elapse.
-      pollMetrics();
-      pollHistory();
-      pollAlerts();
+      // card while waiting for the first interval to elapse. The three
+      // refreshes are independent and handle their own fetch errors, so they
+      // are fired concurrently and deliberately not awaited.
+      void pollMetrics();
+      void pollHistory();
+      void pollAlerts();
       startPolling();
     });
   }
@@ -1228,5 +1230,5 @@
     await reloadAll();
   }
 
-  main();
+  main().catch((e) => console.error('failed to initialise dashboard', e));
 })();

internal/web/polling_test.go

@@ -1,6 +1,7 @@
 package web
 
 import (
+	"regexp"
 	"strings"
 	"testing"
 )
@@ -61,3 +62,30 @@ func TestAppJS_PollFunctionsGuardInFlightRequests(t *testing.T) {
 		}
 	}
 }
+
+// TestAppJS_AsyncCallsAreNotFloating guards issue #168 (SonarCloud
+// javascript:S9383): a call to one of app.js's own async functions written
+// as a bare statement drops the returned promise, so any rejection surfaces
+// only as an anonymous "unhandled rejection". Every such call must be
+// awaited, explicitly discarded with `void` (fire-and-forget refreshes), or
+// chained to a rejection handler (`main().catch(...)`) — all of which keep
+// the call from starting the statement.
+func TestAppJS_AsyncCallsAreNotFloating(t *testing.T) {
+	data, err := assetsFS.ReadFile("assets/app.js")
+	if err != nil {
+		t.Fatalf("read app.js: %v", err)
+	}
+	js := string(data)
+
+	decls := regexp.MustCompile(`async function (\w+)\(`).FindAllStringSubmatch(js, -1)
+	if len(decls) == 0 {
+		t.Fatal("app.js: expected at least one async function declaration")
+	}
+	for _, d := range decls {
+		bare := regexp.MustCompile(`(?m)^[ \t]*` + d[1] + `\([^)]*\);`)
+		for _, loc := range bare.FindAllStringIndex(js, -1) {
+			line := strings.Count(js[:loc[0]], "\n") + 1
+			t.Errorf("app.js:%d: %s() is called as a bare statement; await it, prefix it with void, or attach .catch", line, d[1])
+		}
+	}
+}