v5.0.3: harden embedded-node restart handling + version bump

On top of Lucas's #119 (--db-sync-mode safe):
- DaemonManager: age out the crash-retry budget (m_restarts) so legit node
  toggles/restarts don't trip "maximum retries exceeded" after ~4 spawns over a
  session; the cap now only bites a genuine rapid crash-loop.
- DaemonManager: clear the m_stopping latch on start(), so the crash auto-restart
  keeps working after the node has been toggled off then on (it was a one-shot
  before).
- bump version 5.0.2 -> 5.0.3.

CMakeLists.txt

@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 3.18)
 
 project(feather
-  VERSION "5.0.2"
+  VERSION "5.0.3"
   DESCRIPTION "A free Monero desktop wallet"
   LANGUAGES CXX C ASM
 )

src/utils/DaemonManager.cpp

@@ -4,6 +4,7 @@
 #include "utils/DaemonManager.h"
 
 #include <QCoreApplication>
+#include <QDateTime>
 #include <QDir>
 #include <QDirIterator>
 #include <QEventLoop>
@@ -62,6 +63,11 @@ void DaemonManager::start() {
     if (state == QProcess::ProcessState::Running || state == QProcess::ProcessState::Starting)
         return;
 
+    // wowlet: a deliberate (re)start clears the stop latch, so a later UNEXPECTED exit can auto-restart
+    // again. Without this, m_stopping stays true after the first node toggle-off and the crash
+    // auto-restart is silently dead for the rest of the session.
+    m_stopping = false;
+
     if (!this->unpackBins()) {
         this->setErrorMessage("Unable to unpack the embedded wownerod binary.");
         return;
@@ -85,6 +91,14 @@ void DaemonManager::start() {
         // and surfaces a clear error via handleProcessError().
     }
 
+    // wowlet: the retry cap guards against a rapid crash-loop, not legit restarts. If the node had been
+    // up a while before this (re)start, clear the budget — otherwise a few node toggles over a session
+    // trip "maximum retries exceeded" and it won't start again until wowlet is relaunched.
+    const qint64 nowMs = QDateTime::currentMSecsSinceEpoch();
+    if (m_lastSpawnMs != 0 && (nowMs - m_lastSpawnMs) > 60000)
+        m_restarts = 0;
+    m_lastSpawnMs = nowMs;
+
     m_restarts += 1;
     if (m_restarts > 4) {
         this->setErrorMessage("wownerod failed to start: maximum retries exceeded");
@@ -108,6 +122,14 @@ void DaemonManager::start() {
     arguments << "--no-igd";                 // no UPnP port mapping
     arguments << "--out-peers" << "16";
     arguments << "--log-file" << QDir(dataDir).filePath("wownerod.log");
+    // wowlet: disable LMDB batch write mode so every block is committed in its own transaction.
+    // Without this, wownero accumulates blocks in a single large write transaction that is ABORTED
+    // (not committed) when the daemon shuts down mid-sync — causing all progress since the last
+    // batch boundary to be silently lost. This is visible in wownerod.log: sessions consistently
+    // revert to a stale height despite a full graceful shutdown sequence. Using "safe" mode is
+    // slower for initial sync but is the only correct choice for an embedded node that starts and
+    // stops with the wallet; the wownero chain is small enough that the overhead is acceptable.
+    arguments << "--db-sync-mode" << "safe";
 
     // Full archival by default (wownero's chain is small); prune is an explicit opt-in.
     if (conf()->get(Config::pruneBlockchain).toBool())

src/utils/DaemonManager.h

@@ -73,6 +73,7 @@ private:
     bool m_synced = false;
     bool m_stopping = false;
     int m_restarts = 0;
+    qint64 m_lastSpawnMs = 0;   // wowlet: wall-clock of last spawn — ages out the crash-retry budget
 
     static QPointer<DaemonManager> m_instance;
 };