Bug: Thread cleanup exception during interpreter shutdown when using VS Code debugger

Inizia riproducendo lo script fornito con il debugger di VS Code e confrontalo con una normale esecuzione di python script.py. Analizza i thread gestiti dal debugger e la pulizia durante l’arresto dell’interprete relativa a _DeleteDummyThreadOnDel.__del__; il lavoro è completato quando l’esecuzione con il debugger termina senza il TypeError riportato.

📌 Environment

• Python Version: 3.13.5
• OS: Windows 11
• Debugger: VS Code (debugpy)

---

📝 Description

When running code under the VS Code debugger, an exception is raised during interpreter shutdown:

Exception ignored in: <function _DeleteDummyThreadOnDel.__del__>
Traceback (most recent call last):
  File ".../threading.py", line 1385, in __del__
TypeError: 'NoneType' object does not support the context manager protocol

---

🔍 Key Observation

• ❌ Does NOT occur when running normally (python script.py)
• ✅ Only occurs when running via VS Code debugger

---

🧪 Steps to Reproduce

from strands import Agent, tool
from strands_tools import http_request
import boto3
from strands.models import BedrockModel

session = boto3.Session(region_name="us-east-1")

model = BedrockModel(
    model_id="your-model-id",
    boto_session=session
)

@tool
def dummy_tool(text: str) -> str:
    return text

agent = Agent(
    system_prompt="Test agent",
    tools=[http_request, dummy_tool],
    model=model
)

agent("Test request")

print("DONE EXECUTION")

---

📤 Observed Output (Debugger Only)

DONE EXECUTION
Exception ignored in: <function _DeleteDummyThreadOnDel.__del__>
Traceback (most recent call last):
  File ".../threading.py", line ...
TypeError: 'NoneType' object does not support the context manager protocol

---

✅ Expected Behavior

No exception during interpreter shutdown in debugger mode.

---

❌ Actual Behavior

Thread cleanup triggers a TypeError during interpreter shutdown when running under debugger.

---

📎 Notes

• Likely related to debugger-managed threads and interpreter shutdown order
• Possibly occurs when threads outlive global state during teardown
• Does not impact runtime correctness

---

🛠️ Workarounds

• Running script outside debugger avoids issue
• Adding delays (time.sleep) reduces occurrence but does not fully eliminate

---

📌 Summary

This appears to be a debugger-related threading cleanup issue during interpreter shutdown in Python 3.13.