fix: registered timestamp on timer node

timer_node.py

@@ -5,20 +5,20 @@ Measures workflow execution time from start to finish
 
 import time
 
-# Global storage for workflow timers
-_TIMER_START_TIMES = {}
-
 
 class MetaHubTimerNode:
     """
-    Timer node that measures workflow execution time.
+    Timer node that records a timestamp for workflow execution timing.
 
     Usage:
-    1. Place this node at the START of your workflow
-    2. Connect any input (image, latent, conditioning, etc.) to trigger it early
+    1. Place this node where you want to START measuring time
+    2. Connect any input (clip, image, latent, conditioning) to trigger it
     3. Connect the 'elapsed_time' output to MetaHub Save Node's 'generation_time_override'
 
-    The timer starts when this node first executes in a workflow run.
+    How it works:
+    - This node records a timestamp when it executes
+    - The Save Node calculates elapsed time from this timestamp
+    - Place multiple timers to measure different workflow stages
     """
 
     @classmethod
@@ -41,8 +41,8 @@ class MetaHubTimerNode:
 
     def measure_time(self, clip=None, image=None, latent=None, conditioning=None):
         """
-        Measures elapsed time since this timer node was first executed.
-        Each timer node instance tracks its own independent timing.
+        Records the current timestamp when this node executes.
+        The Save Node will calculate elapsed time from this timestamp.
 
         Args:
             clip: Optional CLIP input (passed through unchanged)
@@ -51,30 +51,14 @@ class MetaHubTimerNode:
             conditioning: Optional CONDITIONING input (passed through unchanged)
 
         Returns:
-            (clip, image, latent, conditioning, elapsed_time): Original inputs and elapsed time in seconds
+            (clip, image, latent, conditioning, start_timestamp): Original inputs and start timestamp
         """
-        current_time = time.time()
+        # Return current timestamp - Save Node will calculate elapsed time
+        start_timestamp = time.time()
+        print(f"[MetaHub Timer] Recording start timestamp: {start_timestamp}")
 
-        # Use this node instance's ID as unique timer identifier
-        # This allows multiple independent timers in the same workflow
-        timer_id = id(self)
-
-        # Start timer on first execution of this specific timer instance
-        if timer_id not in _TIMER_START_TIMES:
-            _TIMER_START_TIMES[timer_id] = current_time
-            print(f"[MetaHub Timer] Started timer #{timer_id}")
-
-            # Cleanup old timers (keep only last 50 to support multiple timers)
-            if len(_TIMER_START_TIMES) > 50:
-                oldest_key = min(_TIMER_START_TIMES.keys(), key=lambda k: _TIMER_START_TIMES[k])
-                del _TIMER_START_TIMES[oldest_key]
-
-        # Calculate elapsed time
-        elapsed = current_time - _TIMER_START_TIMES[timer_id]
-        print(f"[MetaHub Timer] Elapsed time: {elapsed:.2f}s")
-
-        # Passthrough the inputs unchanged + return elapsed time
-        return (clip, image, latent, conditioning, elapsed)
+        # Passthrough the inputs unchanged + return start timestamp
+        return (clip, image, latent, conditioning, start_timestamp)
 
 
 NODE_CLASS_MAPPINGS = {