[Feature] Compatibility with ML Flow's session tracing

[Jiayang-21]

What feature would you like to see?

Description

Seems like DSPy'sautolog uses a callback-based tracing: after DSPy's invocation it'd finalize trace，causing
mlflow.update_current_trace() to not with active trace, thus not able to utilize the session tracing capability

Or did I miss something? This tripped my ClaudeCode for a while and I current best guess is above.

Setup

• DSpy on server side; Client side determine if a call is a new session or not (Task-turn structure)

Would you like to contribute?

• Yes, I'd like to help implement this.
• No, I just want to request it.

Additional Context

No response

[chenmoneygithub]

@Jiayang-21 Thanks for reporting the issue!

i took a deeper look at the trace grouping feature of MLflow, and it looks like you need to manually store the session id at the client side, and include it in the request:

@app.post("/chat")
def handle_chat(request: Request, chat_request: ChatRequest):
    session_id = request.headers.get("X-Session-ID", "default-session")
    user_id = request.headers.get("X-User-ID", "default-user")
    response_text = process_chat(chat_request.message, user_id, session_id)
    return {"response": response_text}

[Jiayang-21]

Thanks Chen, this helped solve my issue :)