v1.0.4; fix server miss eval bug

UPDATE.md

@@ -1,4 +1,7 @@
 # UPDATE
+## 20260106
+### v1.0.4
+Fix bug: 修复高频评估会将之前的评估冲掉问题
 ## 20260101
 ### v1.0.3
 Fix bug: 修复多进程中修改全局参数配置的问题, 同一个进程池中修改的全局变量会被保留, 因此task_register中需要实例化对象

robogauge/scripts/server.py

@@ -83,7 +83,9 @@ def run_api_server(input_queue: multiprocessing.Queue, result_dict: dict, port=9
     def get_result(task_id: str):
         if task_id not in result_dict:
             return {"status": ResponseStatus.NOT_FOUND}
-        result = result_dict.pop(task_id)
+        result = result_dict[task_id]
+        if result["status"] == ResponseStatus.FINISHED:
+            result_dict.pop(task_id)
         return result
 
     print(f"📡 API Server listening on port {port}...")

setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name="robogauge",  # 包名
-    version="1.0.1",   # 版本号
+    version="1.0.4",   # 版本号
     author="Wu Tianyang", # 你的名字
     author_email="993660140@qq.com",
     description="A generic robot RL model evaluation library based on MuJoCo",