I have a message queue, with messages that have a well defined schema. Up until very recently, there has been only one message type, and therefore I've only needed one schema.

A recent change to the queue now allows for multiple different messages, with different schemas on the same queue.

Up to now, I've been handling the messages like:

for msg in queue:
    loaded_msg = MessageSchema.Schema().load(msg)  # if the message doesn't meet the schema, that's bad, and the test should fail immediately
    # do stuff

Now that there are two, this fails, as the new schema isn't a sub, or superset of the current model.

My current approach to handle this is:

for msg in queue:
    try:
        loaded_msg = MessageSchema.Schema().load(msg)
    except ValidationError as e:
        try:
            _ = NewMessageSchema.schema().load(msg)
            continue    # Don't care about NewMessages, but it should still match one of the schemas
        except ValidationError:
            raise e  # message doesn't meet either schema, this is bad
    # do stuff

Obviously, this isn't scalable if new message schemas are allowed on the queue. What's the best way to handle it?