When the subscriber is receiving UADP messages from multiple groups and writers on the same connection, how can the subscriber recognize that the message is a heartbeat message - just from the message itself, without necessarily having access to the configuration?

The spec says "If the key frame is a heartbeat DataSetMessage, only the header is encoded but the following
structure shall not be encoded into the DataSetMessage.". I think it means the DataSetMessage Header. But, a generic UADP parser (such as Wireshark dissector) will just see a corrupted key frame, where the DataSetMessage Header is not followed by the Data Key Frame DataSetMessage data.

I could only find one place in the spec which could help distinguishing the heartbeat messages - "For heartbeat DataSetMessages the majorVersion and
minorVersion in the configurationVersion shall always be 0 in the configuration and in the DataSetMessages.".

However, these fields are not always present in the DataSetMessage Header. Maybe the spec is trying to say that in order for the parser to recognize the heartbeat messages, the ConfigurationVersionMajorVersion and ConfigurationVersionMinorVersion fields must be enabled in the message? And without them, the heartbeat messages are not possible? That would be OK, if that is the intent, I think it should be expressed more clearly in the spec.