Issue with Playground Rendering After Updating Langserve Version

[estebancatano]

I updated my langserve version from 0.2.3 to 0.3.0, and I'm experiencing an issue.

If I pass config_keys=("run_id", "configurable", "metadata") as a parameter in the add_routes method, the playground is not displayed correctly.

This sample code also produces the error:
add_routes(
app,
ChatOpenAI(model="gpt-4o-2024-08-06"),
path="/openai",
config_keys=("run_id", "configurable", "metadata"),
)

The key run_id may be causing the error because if I remove it, it does render, but a warning appears for the metadata key.

The installed versions is as follows:

• langchain-openai 0.2.1
• langserve 0.3.0

[nicho2]

I obtain this too

config = {
    "configurable": {
        # le user_id donne les droits d'accès aux espaces
        "user_id": "3",
        # Checkpoints are accessed by thread_id
        "thread_id": thread_id

    },
    "run_name": "spaceUpChat"
}
runnable = graph.with_types(input_type=ChatInputType, output_type=dict)
runnable = runnable.with_config(config)
add_routes(app, runnable)

[eyurtsev]

Thanks for reporting. I'll need to investigate

[fhkingma]

Any pre-lim results from the investigation? Any way we can assist?
We have to upgrade to pydantic>=2.0.0 based on a business requirement, and would like to understand if we can hack our way around this to enable the playground again, until a fix is released.