[
https://issues.apache.org/jira/browse/HUDI-5572?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
HunterXHunter updated HUDI-5572:
--------------------------------
Description:
When we use spark to initialize the hudi table,
.hoodie#hoodie.properties#hoodie.table.create.schema will carry information
'name=$tablename_record' and 'namespace'='hoodie.$tablename'.
But Flink will not carry this information when writing,
so there will be incompatibilities when doing `validateSchema`.
Here I think we should skip check the compatibility of Schema#name when using
flink write.
!image-2023-01-18-11-51-12-914.png|width=851,height=399!
was:
When we use spark to initialize the hudi table,
.hoodie#hoodie.properties#hoodie.table.create.schema will carry information
'name=$tablename_record' and 'namespace'='hoodie.$tablename'.
But Flink will not carry this information when writing,
so there will be incompatibilities when doing `validateSchema`.
Here I think we should skip check the compatibility of Schema#name when using
flink write.
> Flink write need to skip check the compatibility of Schema#name
> ---------------------------------------------------------------
>
> Key: HUDI-5572
> URL: https://issues.apache.org/jira/browse/HUDI-5572
> Project: Apache Hudi
> Issue Type: Bug
> Reporter: HunterXHunter
> Priority: Major
> Attachments: image-2023-01-18-11-51-12-914.png
>
>
> When we use spark to initialize the hudi table,
> .hoodie#hoodie.properties#hoodie.table.create.schema will carry information
> 'name=$tablename_record' and 'namespace'='hoodie.$tablename'.
> But Flink will not carry this information when writing,
> so there will be incompatibilities when doing `validateSchema`.
> Here I think we should skip check the compatibility of Schema#name when using
> flink write.
> !image-2023-01-18-11-51-12-914.png|width=851,height=399!
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
