chore: trim comments added in the coverage work

cmd/eraser/cmd_auto.go

@@ -85,11 +85,9 @@ func runAutoLoop(every time.Duration) error {
 	}
 }
 
-// waitForNextCycle sleeps for every and reports whether the user asked to
-// stop. Signals are caught only while waiting: during a cycle Ctrl+C keeps
-// its default and kills the process, which is safe mid-send (history is
-// written per broker, the OS drops the lock) and doesn't make the user wait
-// out a 15-minute send. A var so tests can end the loop.
+// waitForNextCycle reports whether to stop. Signals are caught only here:
+// mid-cycle Ctrl+C just kills the process, which is safe (history is per
+// broker, the OS drops the lock). A var so tests can end the loop.
 var waitForNextCycle = func(every time.Duration) (stop bool) {
 	ctx, cancel := signal.NotifyContext(context.Background(), os.Interrupt, syscall.SIGTERM)
 	defer cancel()

cmd/eraser/cmd_send.go

@@ -93,10 +93,8 @@ func countWithEmail(brokers []broker.Broker) int {
 	return n
 }
 
-// capSends keeps brokers up to the n-th one with an email. Email-less brokers
-// are only skipped with a note, so they mustn't use up the daily budget: they
-// never get a history row, sort first on every run, and would otherwise eat
-// the same slots each time.
+// capSends keeps brokers up to the n-th one with an email. Email-less ones
+// never get a history row and sort first every run; they mustn't eat budget.
 func capSends(brokers []broker.Broker, n int) []broker.Broker {
 	for i, b := range brokers {
 		if strings.TrimSpace(b.Email) == "" {

cmd/eraser/cmd_watch_test.go

@@ -1,8 +1,7 @@
 //go:build !race
 
-// go-imap v1.2.1 writes IDLE's DONE from its own goroutine onto the writer
-// the next command uses, which -race flags; that's upstream, so this test
-// only runs without the race detector.
+// go-imap v1.2.1 writes IDLE's DONE from its own goroutine; -race flags it
+// (upstream), so this only runs without the race detector.
 
 package main
 

internal/browser/browser.go

@@ -133,10 +133,8 @@ func (b *Browser) NavigateAndFill(url string, brokerID string, autoSubmit bool)
 		}
 	}
 
-	// The first Run starts Chrome and ties it to that Run's context. Start it
-	// on the long-lived b.ctx: started on the per-call timeout context below,
-	// cancelling that at return closed the browser for every later call
-	// (`eraser fill --pending` fills several forms with one Browser).
+	// Start Chrome on b.ctx: the first Run's context owns the browser, so the
+	// per-call timeout ctx below would close it after one form (fill --pending).
 	if err := chromedp.Run(b.ctx); err != nil {
 		result.ErrorMessage = fmt.Sprintf("browser failed to start: %v", err)
 		return result, err

internal/browser/browser_chrome_test.go

@@ -107,13 +107,10 @@ func TestSubmitButtonTextPattern_MirrorsJSRegex(t *testing.T) {
 	}
 }
 
-// Ubuntu 24.04 runners block the unprivileged user namespaces Chrome's
-// sandbox needs. The pages under test are local fixtures, so CI runs
-// without it.
+// Ubuntu 24.04 runners can't give Chrome its sandbox; pages are local fixtures.
 func init() {
 	if os.Getenv("GITHUB_ACTIONS") == "true" {
-		// A cold Chrome start under -race on a shared runner can outlast
-		// chromedp's 20s default for the DevTools URL.
+		// cold start under -race can outlast chromedp's 20s default
 		extraAllocatorOptions = append(extraAllocatorOptions, chromedp.NoSandbox, chromedp.WSURLReadTimeout(90*time.Second))
 	}
 }

internal/inbox/imap_watch_test.go

@@ -1,8 +1,7 @@
 //go:build !race
 
-// go-imap v1.2.1's client writes IDLE's DONE from its own goroutine onto the
-// writer the next command uses, which -race flags; that's upstream, so this
-// test only runs without the race detector.
+// go-imap v1.2.1 writes IDLE's DONE from its own goroutine; -race flags it
+// (upstream), so this only runs without the race detector.
 
 package inbox
 
@@ -20,10 +19,8 @@ import (
 	"github.com/drumandbytes/eraser/internal/config"
 )
 
-// New mail during IDLE must not wedge the connection: the follow-up SELECT
-// draws an unsolicited EXISTS, which deadlocked when the watch loop was also
-// the only reader of the client's update channel. Scripted, because go-imap's
-// own server races when it broadcasts updates.
+// New mail during IDLE: the follow-up SELECT's unsolicited EXISTS must not
+// wedge the watch. Scripted, since go-imap's server races on updates.
 func TestWatchForNewEmailsSurvivesNewMail(t *testing.T) {
 	reIdled := make(chan struct{})
 	addr := fakeIMAPServer(t, func(conn net.Conn, br *bufio.Reader) error {

internal/inbox/monitor.go

@@ -396,12 +396,9 @@ func (m *Monitor) WatchForNewEmails(ctx context.Context, callback func(Email)) e
 		return fmt.Errorf("failed to select mailbox: %w", err)
 	}
 
-	// The client delivers unilateral responses (EXISTS after the SELECT in
-	// FetchBrokerEmails, for one) on Updates and blocks until they're read,
-	// so draining it here in the loop deadlocked on the first new mail. A
-	// separate reader keeps the connection moving and just flags new mail.
-	// ponytail: the drain goroutine lives as long as the connection; fine
-	// for one watch per Monitor.
+	// Updates blocks until read, and FetchBrokerEmails' SELECT emits EXISTS
+	// into it, so it needs its own reader or the loop deadlocks.
+	// ponytail: drain goroutine lives as long as the connection.
 	updates := make(chan client.Update, 16)
 	newMail := make(chan struct{}, 1)
 	go func() {
@@ -446,8 +443,7 @@ func (m *Monitor) WatchForNewEmails(ctx context.Context, callback func(Email)) e
 				callback(email)
 			}
 
-			// The fetch's own SELECT reports EXISTS too; that's covered by
-			// the fetch just done, so don't let it bounce IDLE straight away.
+			// the fetch's own SELECT flagged new mail too; drop it
 			select {
 			case <-newMail:
 			default:

internal/web/handlers_send_test.go

@@ -18,9 +18,8 @@ import (
 	"github.com/drumandbytes/eraser/internal/history"
 )
 
-// fakeRelay is a plaintext SMTP relay that accepts any number of
-// connections. A recipient in reject gets a 550; authFail answers MAIL FROM
-// with a 535, the way a provider that has locked the account does.
+// fakeRelay is plaintext SMTP. reject -> 550 on RCPT; authFail -> 535 on
+// MAIL FROM, like a provider that locked the account.
 type fakeRelay struct {
 	addr     string
 	mu       sync.Mutex

internal/web/server.go

@@ -323,8 +323,7 @@ func (s *Server) parseTemplates() (map[string]*template.Template, error) {
 func (s *Server) Start() error {
 	router := s.setupRouter()
 
-	// lifeMu: serve's Ctrl+C handler may call Shutdown before or while this
-	// runs; it used to read httpServer unsynchronised (nil = panic).
+	// serve's Ctrl+C handler can call Shutdown before or during this
 	s.lifeMu.Lock()
 	if s.shutDown {
 		s.lifeMu.Unlock()
@@ -342,9 +341,7 @@ func (s *Server) Start() error {
 	srv := s.httpServer
 	s.lifeMu.Unlock()
 
-	// Listen before opening the browser, so it opens only once there's a
-	// server to load (it used to open after a blind 500ms, even on a
-	// taken port).
+	// listen first: open the browser only once there's a server to load
 	ln, err := net.Listen("tcp", srv.Addr)
 	if err != nil {
 		cancel()
@@ -364,8 +361,7 @@ func (s *Server) Start() error {
 	return nil
 }
 
-// Shutdown gracefully shuts down the server. Before Start it just makes
-// Start return straight away.
+// Shutdown stops the server; called before Start, Start returns at once.
 func (s *Server) Shutdown(ctx context.Context) error {
 	s.lifeMu.Lock()
 	defer s.lifeMu.Unlock()