[ 
https://issues.apache.org/jira/browse/FLINK-19098?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17187607#comment-17187607
 ] 

Brian Zhou edited comment on FLINK-19098 at 8/31/20, 9:48 AM:
--------------------------------------------------------------

Thanks Jark and Danny, I can pick up this change and will take care of the 
annotation. can you assign this task to me? I noticed csv format is using the 
similar jsonNode converter with json format, if we do as [~danny0405] 
suggested, is it fine to keep this in json and let csv has the flink-json 
dependency? Or should we need a "flink-format-common" module to deal with this?


was (Author: brian zhou):
Thanks Jark and Danny, I can pick up this change and will take care of the 
annotation. can you assign this task to me? I noticed csv format is using the 
similar jsonNode converter with json format, if we do as [~danny0405] 
suggested, is it fine to keep this in json and let csv has the flink-json 
dependency?

> Make Rowdata converters public
> ------------------------------
>
>                 Key: FLINK-19098
>                 URL: https://issues.apache.org/jira/browse/FLINK-19098
>             Project: Flink
>          Issue Type: Improvement
>          Components: Formats (JSON, Avro, Parquet, ORC, SequenceFile), Table 
> SQL / Ecosystem
>    Affects Versions: 1.11.0
>            Reporter: Brian Zhou
>            Priority: Major
>
> Some of the Rowdata converters(SeDer between Rowdata and format objects like 
> GenericRecord/JsonNode) are private or package-private (like Json), this is 
> not easy for other third-party connector projects to utilize to implement its 
> own format factory in Table API.
> For example, Pravega connector is now developing a schema-registry-based 
> format factory. The [Pravega schema 
> registry|https://github.com/pravega/schema-registry] is a rest service 
> similar with confluent registry , but it can help to serialize/deserialize 
> json/avro/protobuf/custom format data. It will help a lot if these converters 
> are public.
> Noticed in FLINK-16048, we have already moved the avro converters out and 
> made them public. Similarly, it should be safe to make at least json and csv 
> format converters public.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to