add selfplay test

openrl/selfplay/callbacks/selfplay_api.py

@@ -50,7 +50,7 @@ def _init_callback(self) -> None:
         )
 
         self.bind = SelfplayAPIServer.bind()
-        serve.run(self.bind)
+        serve.run(self.bind, route_prefix="/selfplay")
         success = False
         try_time = 10
         while not success:

openrl/selfplay/selfplay_api/selfplay_api.py

@@ -33,7 +33,7 @@
 from openrl.selfplay.selfplay_api.opponent_model import BattleResult
 
 
-@serve.deployment(route_prefix="/selfplay")
+@serve.deployment()
 @serve.ingress(app)
 class SelfplayAPIServer(BaseSelfplayAPIServer):
     @app.post("/set_sample_strategy")

setup.py

@@ -70,9 +70,14 @@ def get_extra_requires() -> dict:
             "datasets==2.13",
             "evaluate",
         ],
-        "selfplay": ["ray[default]", "ray[serve]", "pettingzoo[classic]", "trueskill"],
+        "selfplay": [
+            "ray[default]>=2.7",
+            "ray[serve]",
+            "pettingzoo[classic]",
+            "trueskill",
+        ],
         "selfplay_test": [
-            "ray[default]",
+            "ray[default]>=2.7",
             "ray[serve]",
             "fastapi",
             "pettingzoo[mpe]",

tests/test_selfplay/test_train_selfplay.py

@@ -34,7 +34,7 @@ def config(request):
     cfg_parser = create_config_parser()
     cfg = cfg_parser.parse_args(["--config", "./examples/selfplay/selfplay.yaml"])
     cfg.selfplay_api.port = request.param["port"]
-    print("port:",request.param["port"])
+    print("port:", request.param["port"])
     for i, c in enumerate(cfg.callbacks):
         if c["id"] == "SelfplayCallback":
             c["args"][