[
https://issues.apache.org/jira/browse/HUDI-5191?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
zouxxyy updated HUDI-5191:
--------------------------
Description:
After spark3.2, avro was upgraded to 1.10.2+, some functions changed.
E.g:
If the Key is not found, `GenericData#get` will throw a
AvroRuntimeException("Not a valid schema field: " + key), while it will return
null before.
And, `Schema.Field` will check the default value internally, which can cause
failures such as
`TestOverwriteNonDefaultsWithLatestAvroPayload#testNullColumn`
At this point, some test cases will fail in spark3.2+ scenarios.
was:
After spark3.2, avro was upgraded to 1.10.2+, some functions changed.
E.g:
If the Key is not found, `GenericData#get` will throw a
AvroRuntimeException("Not a valid schema field: " + key), while it will return
null before.
At this point, some test cases will fail in spark3.2+ scenarios.
> avro compatibility with spark3.2+
> ---------------------------------
>
> Key: HUDI-5191
> URL: https://issues.apache.org/jira/browse/HUDI-5191
> Project: Apache Hudi
> Issue Type: Bug
> Reporter: zouxxyy
> Priority: Major
>
> After spark3.2, avro was upgraded to 1.10.2+, some functions changed.
> E.g:
> If the Key is not found, `GenericData#get` will throw a
> AvroRuntimeException("Not a valid schema field: " + key), while it will
> return null before.
> And, `Schema.Field` will check the default value internally, which can cause
> failures such as
> `TestOverwriteNonDefaultsWithLatestAvroPayload#testNullColumn`
> At this point, some test cases will fail in spark3.2+ scenarios.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
