Bug in Environment Initialization with Direct Role Assignment Causes Error

[iphysresearch]

Bug description

我看到Environment中也初始化参数也可以直接定义Role进去，所以我就试了下（因为我目标要实现多个不同的环境下的不同Team要协作），结果发现直接Bug报错。目测Bug的来源我还不太确定，因为我不太熟悉这个类的构造方式。根据下面截图的报错，可能是在 self.roles 赋值的时候，触发了什么机制保护？

Bug solved method

如上。

Environment information

• LLM type and model name:
  • gpt-4-1106-preview
• System version:

# Inside a docker container
Linux e24f4ba5e757 6.2.0-26-generic #26~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Jul 13 16:27:29 UTC 2 x86_64 x86_64 x86_64 GNU/Linux

• Python version:
  • Conda Python 3.10.13
• MetaGPT version or branch:
  • 0.8.0

• packages version:
• installation method:

pip install -e . && pip install -e .[rag]

Screenshots or logs

[better629]

Maybe you can ref to examples/debate_simple.py, firstly init environment and then add roles into team. or, Environment(desc="xxx", roles={assistant1.profile: assistant1, assistant2.profile: assistant2}) which due to the rolesinside env is a dict that keys areprofileofrole`.

BTW, I see you define a android assistant, what's your demands?