Github user aljoscha commented on the issue:
https://github.com/apache/flink/pull/5573
@yanghua I mean having two separate ways of querying: the server should
have two endpoints (or differentiate based on a query parameter) and return
either the stringified accumulators or the serialized accumulators. For now,
the former would be used for the web frontend while the latter would be used by
the `RestClusterClient`. @GJL maybe has a better opinion on how this should be
done on the server side, though.
---
