[
https://issues.apache.org/jira/browse/KAFKA-9993?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17107726#comment-17107726
]
Boyang Chen commented on KAFKA-9993:
------------------------------------
Thanks [~vvcephei] An example is from the recent implementation for Raft:
[https://github.com/confluentinc/kafka/pull/319/files#r424848803] a couple of
common fields just as error code, leader id and later on cluster.id are
attached to all the generated RPCs like vote, fetch, begin/end quorum, etc, but
you have to manually extract them one by one without a common ancestor. It
would be good to consolidate them.
Similar cases for group coordinator and txn coordinator code are definitely
possible to be simplified, as they separate share a lot of common fields for
sure.
> Think about inheritance in the protocol generation framework
> ------------------------------------------------------------
>
> Key: KAFKA-9993
> URL: https://issues.apache.org/jira/browse/KAFKA-9993
> Project: Kafka
> Issue Type: Improvement
> Reporter: Boyang Chen
> Priority: Major
>
> We have seen that there are a lot of common fields inside the
> request/response templates that could be extracted as a super class for auto
> generated classes. For example most response contains a top level error code.
> Currently to build a service receiving multiple RPCs, the code template
> produces a lot of redundant error code extraction logic which is far from
> ideal.
> What we want to discuss is whether to enable the general inheritance
> mechanism in this framework, what's the trade-off and complexity increase,
> and if there is any workaround just to make less boiler templates.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
