[
https://issues.apache.org/jira/browse/KAFKA-20155?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18059967#comment-18059967
]
Matthias J. Sax commented on KAFKA-20155:
-----------------------------------------
I was just digging some code, and I am wondering if we can actually do this,
w/o also being lazy for the value deserialization? To deserialize the value
itself, we need to pass in `Headers` into `deserialize(...)` method, what
implies we need to deserialize the headers, too. Or do I miss something? Or
course, not all deserializers required access to headers, but not sure how we
could decide what "type" of value deserializer we have, to apply this
optimization for header-agnostic deserializers?
Also not sure if lazy value deserialization by itself would make sense? If not,
it seems there is no opportunity to deserialize the headers lazily either.
Thoughts? \cc [~alisa23] [~frankvicky] (and [~lucasbru] who proposed lazy
deserialization on the KIP). If we strip out lazy header deserialization, also
wondering if we would want to change the byte format and update the KIP?
> Implement lazy deserialization mechanism
> ----------------------------------------
>
> Key: KAFKA-20155
> URL: https://issues.apache.org/jira/browse/KAFKA-20155
> Project: Kafka
> Issue Type: Sub-task
> Components: streams
> Reporter: TengYao Chi
> Assignee: TengYao Chi
> Priority: Major
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
