Github user tomatophantastico commented on the issue:
https://github.com/apache/metamodel/pull/178
Hi,
I am not sure, if i get the targeted use case here right, so intuitively i
would hesitate to use this MM module for querying the Kafka, because:
1) In Kafka message values are binary values (right?) and will often carry
complex messages. While you can choose for simple/primitive types appropriate
deserializers, for more complex (json, xml, avro, kryo, ....) this is not
possible. So actually an nesting of datacontexts would be great here, though i
do not think this is possible.
2) W.r.t. the subscriber issue:
While it may work, your approach materializes all message data in main
memory, for each query, each time.
This can get very expensive, for both the Kafka server and the consumer.
If continually used, this should definitely by a subscriber / database.
---
