[ 
https://issues.apache.org/jira/browse/KAFKA-6684?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16458262#comment-16458262
 ] 

Ewen Cheslack-Postava commented on KAFKA-6684:
----------------------------------------------

The idea here seems quite similar to the SimpleHeaderConverter included with 
KIP-145. We might want to consider a) standardizing all these 
casts/transformations across the different mechanisms (Cast transformation and 
SimpleHeaderConverter) and b) reusing as much code as possible. For (b), we 
might be able to just check whether something is a logical type and then always 
delegate to shared code from SimpleHeaderConverter so we get a minimal 
implementation w/ all the support for different types.

> Support casting values with bytes schema to string 
> ---------------------------------------------------
>
>                 Key: KAFKA-6684
>                 URL: https://issues.apache.org/jira/browse/KAFKA-6684
>             Project: Kafka
>          Issue Type: Improvement
>          Components: KafkaConnect
>            Reporter: Amit Sela
>            Priority: Critical
>             Fix For: 2.0.0
>
>
> Casting from BYTES is not supported, which means that casting LogicalTypes is 
> not supported.
> This proposes to allow casting anything to a string, kind of like Java's 
> {{toString()}}, such that if the object is actually a LogicalType it can be 
> "serialized" as string instead of bytes+schema.
>  
> {noformat}
> Examples:
> BigDecimal will cast to the string representation of the number.
> Timestamp will cast to the string representation of the timestamp, or maybe 
> UTC yyyymmddTHH:MM:SS.f format?
> {noformat}
>  
> Worst case, bytes are "casted" to whatever the {{toString()}} returns - its 
> up to the user to know the data.
> This would help when using a JSON sink, or anything that's not Avro.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to