Merge

.github/workflows/ghcr_runtime.yml

@@ -145,8 +145,7 @@ jobs:
         run: make install-python-dependencies
       - name: Run runtime tests
         run: |
-          # We install pytest-xdist in order to run tests across CPUs. However, tests start to fail when we run
-          # then across more than 2 CPUs for some reason
+          # We install pytest-xdist in order to run tests across CPUs
           poetry run pip install pytest-xdist
 
           # Install to be able to retry on failures for flaky tests
@@ -158,10 +157,10 @@ jobs:
           SKIP_CONTAINER_LOGS=true \
           TEST_RUNTIME=eventstream \
           SANDBOX_USER_ID=$(id -u) \
-          SANDBOX_BASE_CONTAINER_IMAGE=$image_name \
+          SANDBOX_RUNTIME_CONTAINER_IMAGE=$image_name \
           TEST_IN_CI=true \
           RUN_AS_OPENHANDS=false \
-          poetry run pytest -n 3 --reruns 1 --reruns-delay 3 --cov=agenthub --cov=openhands --cov-report=xml -s ./tests/runtime
+          poetry run pytest -n 3 -raR --reruns 1 --reruns-delay 3 --cov=agenthub --cov=openhands --cov-report=xml -s ./tests/runtime
       - name: Upload coverage to Codecov
         uses: codecov/codecov-action@v4
         env:
@@ -207,8 +206,7 @@ jobs:
         run: make install-python-dependencies
       - name: Run runtime tests
         run: |
-          # We install pytest-xdist in order to run tests across CPUs. However, tests start to fail when we run
-          # then across more than 2 CPUs for some reason
+          # We install pytest-xdist in order to run tests across CPUs
           poetry run pip install pytest-xdist
 
           # Install to be able to retry on failures for flaky tests
@@ -220,10 +218,10 @@ jobs:
           SKIP_CONTAINER_LOGS=true \
           TEST_RUNTIME=eventstream \
           SANDBOX_USER_ID=$(id -u) \
-          SANDBOX_BASE_CONTAINER_IMAGE=$image_name \
+          SANDBOX_RUNTIME_CONTAINER_IMAGE=$image_name \
           TEST_IN_CI=true \
           RUN_AS_OPENHANDS=true \
-          poetry run pytest -n 3 --reruns 1 --reruns-delay 3 --cov=agenthub --cov=openhands --cov-report=xml -s ./tests/runtime
+          poetry run pytest -n 3 -raR --reruns 1 --reruns-delay 3 --cov=agenthub --cov=openhands --cov-report=xml -s ./tests/runtime
       - name: Upload coverage to Codecov
         uses: codecov/codecov-action@v4
         env:
@@ -275,7 +273,7 @@ jobs:
 
           TEST_RUNTIME=eventstream \
           SANDBOX_USER_ID=$(id -u) \
-          SANDBOX_BASE_CONTAINER_IMAGE=$image_name \
+          SANDBOX_RUNTIME_CONTAINER_IMAGE=$image_name \
           TEST_IN_CI=true \
           TEST_ONLY=true \
           ./tests/integration/regenerate.sh

.gitignore

@@ -233,4 +233,5 @@ containers/runtime/project.tar.gz
 trajectory.json
 event_history.json
 sc.py
-t.py
+t.py
+containers/runtime/code

containers/runtime/README.md

@@ -1,11 +1,12 @@
-# Dynamic constructed Dockerfile
+# Dynamically constructed Dockerfile
 
-This folder builds runtime image (sandbox), which will use a `Dockerfile` that is dynamically generated depends on the `base_image` AND a [Python source distribution](https://docs.python.org/3.10/distutils/sourcedist.html) that's based on the current commit of `openhands`.
+This folder builds a runtime image (sandbox), which will use a dynamically generated `Dockerfile`
+that depends on the `base_image` **AND** a [Python source distribution](https://docs.python.org/3.10/distutils/sourcedist.html) that is based on the current commit of `openhands`.
 
-The following command will generate Dockerfile for `ubuntu:22.04` and the source distribution `.tar` into `containers/runtime`.
+The following command will generate a `Dockerfile` file for `nikolaik/python-nodejs:python3.11-nodejs22` (the default base image), an updated `config.sh` and the runtime source distribution files/folders into `containers/runtime`:
 
 ```bash
 poetry run python3 openhands/runtime/utils/runtime_build.py \
-    --base_image ubuntu:22.04 \
+    --base_image nikolaik/python-nodejs:python3.11-nodejs22 \
     --build_folder containers/runtime
 ```

evaluation/utils/shared.py

@@ -6,7 +6,6 @@
 import subprocess
 import time
 import traceback
-from concurrent.futures import ProcessPoolExecutor, as_completed
 from typing import Any, Awaitable, Callable, TextIO
 
 import pandas as pd
@@ -302,6 +301,11 @@ def _process_instance_wrapper(
             time.sleep(5)
 
 
+def _process_instance_wrapper_mp(args):
+    """Wrapper for multiprocessing, especially for imap_unordered."""
+    return _process_instance_wrapper(*args)
+
+
 def run_evaluation(
     dataset: pd.DataFrame,
     metadata: EvalMetadata | None,
@@ -328,20 +332,13 @@ def run_evaluation(
 
     try:
         if use_multiprocessing:
-            with ProcessPoolExecutor(num_workers) as executor:
-                futures = [
-                    executor.submit(
-                        _process_instance_wrapper,
-                        process_instance_func=process_instance_func,
-                        instance=instance,
-                        metadata=metadata,
-                        use_mp=True,
-                        max_retries=max_retries,
-                    )
+            with mp.Pool(num_workers) as pool:
+                args_iter = (
+                    (process_instance_func, instance, metadata, True, max_retries)
                     for _, instance in dataset.iterrows()
-                ]
-                for future in as_completed(futures):
-                    result = future.result()
+                )
+                results = pool.imap_unordered(_process_instance_wrapper_mp, args_iter)
+                for result in results:
                     update_progress(result, pbar, output_fp)
         else:
             for _, instance in dataset.iterrows():

frontend/package.json

@@ -64,7 +64,7 @@
     "@testing-library/jest-dom": "^6.5.0",
     "@testing-library/react": "^16.0.1",
     "@testing-library/user-event": "^14.5.2",
-    "@types/node": "^22.5.5",
+    "@types/node": "^22.6.1",
     "@types/react": "^18.3.8",
     "@types/react-dom": "^18.3.0",
     "@types/react-highlight": "^0.12.8",

frontend/src/components/AgentStatusBar.tsx

@@ -19,6 +19,7 @@ function AgentStatusBar() {
   const { curAgentState, currentStep } = useSelector(
     (state: RootState) => state.agent,
   );
+  const { curStatusMessage } = useSelector((state: RootState) => state.status);
 
   const AgentStatusMap: {
     [k: string]: { message: string; indicator: IndicatorColor };
@@ -81,17 +82,38 @@ function AgentStatusBar() {
   // - Agent is thinking
   // - Agent is ready
   // - Agent is not available
+  useEffect(() => {
+    if (
+      curAgentState === AgentState.AWAITING_USER_INPUT ||
+      curAgentState === AgentState.ERROR ||
+      curAgentState === AgentState.INIT
+    ) {
+      if (document.cookie.indexOf("audio") !== -1) beep();
+    }
+  }, [curAgentState]);
+
+  const [statusMessage, setStatusMessage] = React.useState<string>("");
+
+  React.useEffect(() => {
+    const trimmedCustomMessage = curStatusMessage.message.trim();
+    if (trimmedCustomMessage) {
+      setStatusMessage(t(trimmedCustomMessage));
+    } else {
+      setStatusMessage(AgentStatusMap[curAgentState].message);
+    }
+  }, [curAgentState, curStatusMessage.message]);
+
   return (
-    <div className="flex items-center">
-      <div
-        className={`w-3 h-3 mr-2 rounded-full animate-pulse ${AgentStatusMap[curAgentState].indicator}`}
-      />
-      <span className="text-sm text-stone-400">
-        {AgentStatusMap[curAgentState].message}
-      </span>
-      {currentStep && (
-        <span className="text-sm text-stone-400 ml-2">{currentStep}</span>
-      )}
+    <div className="flex flex-col items-center">
+      <div className="flex items-center">
+        <div
+          className={`w-3 h-3 mr-2 rounded-full animate-pulse ${AgentStatusMap[curAgentState].indicator}`}
+        />
+        <span className="text-sm text-stone-400">{statusMessage}</span>
+        {currentStep && (
+          <span className="text-sm text-stone-400 ml-2">{currentStep}</span>
+        )}
+      </div>
     </div>
   );
 }

frontend/src/components/modals/settings/SettingsForm.tsx

@@ -52,13 +52,17 @@ function SettingsForm({
   const [enableAdvanced, setEnableAdvanced] =
     React.useState(advancedAlreadyInUse);
 
+  const handleAdvancedChange = (value: boolean) => {
+    setEnableAdvanced(value);
+  };
+
   return (
     <>
       <Switch
         data-testid="advanced-options-toggle"
         aria-checked={enableAdvanced}
         isSelected={enableAdvanced}
-        onValueChange={(value) => setEnableAdvanced(value)}
+        onValueChange={handleAdvancedChange}
       >
         Advanced Options
       </Switch>