nebu 0.1.6__tar.gz → 0.1.7__tar.gz

{nebu-0.1.6/src/nebu.egg-info → nebu-0.1.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nebu
-Version: 0.1.6
+Version: 0.1.7
 Summary: A globally distributed container runtime
 Requires-Python: >=3.10.14
 Description-Content-Type: text/markdown

{nebu-0.1.6 → nebu-0.1.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "nebu"
-version = "0.1.6"
+version = "0.1.7"
 description = "A globally distributed container runtime"
 readme = "README.md"
 requires-python = ">=3.10.14"

{nebu-0.1.6 → nebu-0.1.7}/src/nebu/containers/container.py

@@ -26,7 +26,7 @@ class Container:
         self,
         name: str,
         image: str,
-        namespace: str = "default",
+        namespace: Optional[str] = None,
         platform: Optional[str] = None,
         env: Optional[List[V1EnvVar]] = None,
         command: Optional[str] = None,
@@ -208,8 +208,8 @@ class Container:
 
         # Save constructor params to `self` for reference, like you do in ReplayBuffer.
         self.kind = "Container"
-        self.namespace = namespace
-        self.name = name
+        self.namespace = self.container.metadata.namespace
+        self.name = self.container.metadata.name
         self.platform = platform
         self.metadata = meta_request
         self.image = image
@@ -290,7 +290,7 @@ class Container:
     def load(
         cls,
         name: str,
-        namespace: str = "default",
+        namespace: Optional[str] = None,
         config: Optional[GlobalConfig] = None,
     ):
         """

nebu-0.1.7/src/nebu/containers/decorator.py

@@ -0,0 +1,85 @@
+import base64
+import inspect
+import pickle
+import time
+from typing import Any, Callable, List, Optional
+
+import requests
+
+from nebu.containers.container import Container
+
+
+def container(
+    image: str,
+    name: Optional[str] = None,
+    namespace: Optional[str] = None,
+    accelerators: Optional[List[str]] = None,
+    platform: str = "runpod",
+    python_cmd: str = "python",
+):
+    def decorator(func: Callable):
+        nonlocal name
+        if name is None:
+            name = func.__name__
+
+        def wrapper(*args: Any, **kwargs: Any):
+            nonlocal name
+            # Create your container with the server script
+            cont = Container(
+                name=name,  # type: ignore
+                namespace=namespace,
+                platform=platform,
+                image=image,
+                accelerators=accelerators,
+                # Command to start our function execution server
+                command=f"{python_cmd} -m nebu.containers.server",  # TODO: need to get the server code into the container
+                proxy_port=8080,
+            )
+
+            # Wait for container to be running
+            while (
+                cont.container.status
+                and cont.container.status.status
+                and cont.container.status.status.lower() != "running"
+            ):
+                print(
+                    f"Container '{cont.container.metadata.name}' not running yet; waiting..."
+                )
+                time.sleep(1)
+
+            # Get function source code
+            func_code = inspect.getsource(func)
+
+            # Serialize arguments using pickle for complex objects
+            serialized_args = base64.b64encode(pickle.dumps(args)).decode("utf-8")
+            serialized_kwargs = base64.b64encode(pickle.dumps(kwargs)).decode("utf-8")
+
+            # Prepare payload
+            payload = {
+                "function_code": func_code,
+                "args": serialized_args,
+                "kwargs": serialized_kwargs,
+            }
+
+            # Get container URL
+            container_url = (
+                cont.status.tailnet_url
+                if cont.status and hasattr(cont.status, "tailnet_url")
+                else "http://localhost:8080"
+            )
+
+            # Send to container and get result
+            response = requests.post(f"{container_url}/execute", json=payload)
+
+            if response.status_code != 200:
+                raise RuntimeError(f"Function execution failed: {response.text}")
+
+            # Deserialize the result
+            serialized_result = response.json()["result"]
+            result = pickle.loads(base64.b64decode(serialized_result))
+
+            return result
+
+        return wrapper
+
+    return decorator

{nebu-0.1.6 → nebu-0.1.7/src/nebu.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nebu
-Version: 0.1.6
+Version: 0.1.7
 Summary: A globally distributed container runtime
 Requires-Python: >=3.10.14
 Description-Content-Type: text/markdown

nebu-0.1.6/src/nebu/containers/decorator.py

@@ -1,174 +0,0 @@
-import base64
-import inspect
-import pickle
-import time
-from typing import Any, Callable, List, Optional
-
-import requests
-
-from nebu.containers.container import Container
-from nebu.containers.models import V1ContainerRequest, V1EnvVar, V1ResourceMetaRequest
-
-
-def container(
-    image: str,
-    name: Optional[str] = None,
-    namespace: str = "default",
-    accelerators: Optional[List[str]] = None,
-    platform: str = "runpod",
-    python_cmd: str = "python",
-):
-    def decorator(func: Callable):
-        nonlocal name
-        if name is None:
-            name = func.__name__
-
-        def wrapper(*args: Any, **kwargs: Any):
-            nonlocal name
-            # Create your container with the server script
-            cont = Container(
-                name=name,  # type: ignore
-                namespace=namespace,
-                platform=platform,
-                image=image,
-                accelerators=accelerators,
-                # Command to start our function execution server
-                command=f"{python_cmd} -m nebu.containers.server",  # TODO: need to get the server code into the container
-                proxy_port=8080,
-            )
-
-            # Wait for container to be running
-            while (
-                cont.container.status
-                and cont.container.status.status
-                and cont.container.status.status.lower() != "running"
-            ):
-                print(
-                    f"Container '{cont.container.metadata.name}' not running yet; waiting..."
-                )
-                time.sleep(1)
-
-            # Get function source code
-            func_code = inspect.getsource(func)
-
-            # Serialize arguments using pickle for complex objects
-            serialized_args = base64.b64encode(pickle.dumps(args)).decode("utf-8")
-            serialized_kwargs = base64.b64encode(pickle.dumps(kwargs)).decode("utf-8")
-
-            # Prepare payload
-            payload = {
-                "function_code": func_code,
-                "args": serialized_args,
-                "kwargs": serialized_kwargs,
-            }
-
-            # Get container URL
-            container_url = (
-                cont.status.tailnet_url
-                if cont.status and hasattr(cont.status, "tailnet_url")
-                else "http://localhost:8080"
-            )
-
-            # Send to container and get result
-            response = requests.post(f"{container_url}/execute", json=payload)
-
-            if response.status_code != 200:
-                raise RuntimeError(f"Function execution failed: {response.text}")
-
-            # Deserialize the result
-            serialized_result = response.json()["result"]
-            result = pickle.loads(base64.b64decode(serialized_result))
-
-            return result
-
-        return wrapper
-
-    return decorator
-
-
-def on_feedback(
-    human: Human,
-    accelerators: Optional[List[str]] = None,
-    platform: str = "runpod",
-    python_cmd: str = "python",
-    timeout: Optional[str] = None,
-    env: Optional[List[V1EnvVar]] = None,
-):
-    def decorator(func: Callable):
-        nonlocal name
-        if name is None:
-            name = func.__name__
-
-        # Get function source code
-        func_code = inspect.getsource(func)
-
-        command = """
-
-"""
-
-        # Create the container request
-        container_request = V1ContainerRequest(
-            kind="Container",
-            platform=platform,
-            metadata=V1ResourceMetaRequest(
-                name=name,
-                namespace=namespace,
-            ),
-            image=image,
-            env=env,
-            command=f"{python_cmd} -m nebu.containers.server",
-            accelerators=accelerators,
-            timeout=timeout,
-            proxy_port=8080,
-            restart="Never",  # Jobs should not restart
-        )
-
-        def run(*args: Any, **kwargs: Any):
-            # Create a container from the request
-            cont = Container.from_request(container_request)
-
-            # Wait for container to be running
-            while (
-                cont.status
-                and cont.status.status
-                and cont.status.status.lower() != "running"
-            ):
-                print(f"Job '{cont.metadata.name}' not running yet; waiting...")
-                time.sleep(1)
-
-            # Serialize arguments using pickle for complex objects
-            serialized_args = base64.b64encode(pickle.dumps(args)).decode("utf-8")
-            serialized_kwargs = base64.b64encode(pickle.dumps(kwargs)).decode("utf-8")
-
-            # Prepare payload
-            payload = {
-                "function_code": func_code,
-                "args": serialized_args,
-                "kwargs": serialized_kwargs,
-            }
-
-            # Get container URL
-            container_url = (
-                cont.status.tailnet_url
-                if cont.status and hasattr(cont.status, "tailnet_url")
-                else "http://localhost:8080"
-            )
-
-            # Send to container and get result
-            response = requests.post(f"{container_url}/execute", json=payload)
-
-            if response.status_code != 200:
-                raise RuntimeError(f"Function execution failed: {response.text}")
-
-            # Deserialize the result
-            serialized_result = response.json()["result"]
-            result = pickle.loads(base64.b64decode(serialized_result))
-
-            return result
-
-        # Attach the run method to the container request
-        container_request.run = run  # type: ignore
-
-        return container_request
-
-    return decorator