[
https://issues.apache.org/jira/browse/NIFI-14734?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
David Handermann updated NIFI-14734:
------------------------------------
Issue Type: Improvement (was: Bug)
> Schema Inference For Numeric Types
> ----------------------------------
>
> Key: NIFI-14734
> URL: https://issues.apache.org/jira/browse/NIFI-14734
> Project: Apache NiFi
> Issue Type: Improvement
> Components: Extensions
> Affects Versions: 2.4.0
> Reporter: Dariusz Seweryn
> Assignee: Samkit Jain Ksolves Nifi Expertise
> Priority: Major
>
> Record Readers, when inferring schema from numeric types, e.g. JSON \{ "key":
> 100 }, will infer a schema with Field("key", Integer).
> This may pose a problem for streaming processors such as ConsumeKafka or
> ConsumeKinesisStream where a batch of records is retrieved at once — the
> FlowFile writer gets created with the schema of the first received record,
> which is some cases may infer an Integer sized field, where the next records
> would contain Long values. Such situation would end up with exceptions being
> thrown on record write to FlowFile.
> This feature could be implemented as an additional property like "Numeric
> Values Inferrence": ["Narrowest" (default, as it is now), "Widest"]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
