[
https://issues.apache.org/jira/browse/NIFI-6088?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16779976#comment-16779976
]
Matt Burgess commented on NIFI-6088:
------------------------------------
I also noticed this with a BIGINT type, currently all integral values are
inferred as LONGs. A similar approach should apply
> JSON Record Readers sometimes infer types too narrow for the values, leading
> to truncation
> ------------------------------------------------------------------------------------------
>
> Key: NIFI-6088
> URL: https://issues.apache.org/jira/browse/NIFI-6088
> Project: Apache NiFi
> Issue Type: Bug
> Components: Extensions
> Reporter: Matt Burgess
> Priority: Major
>
> When using the "Infer Schema" strategy on JSON Record Readers
> (JsonPathReader, JsonTreeReader, e.g.), all fractional numbers are inferred
> as floats, meaning there can be some truncation if the value can only truly
> fit in a double. The inference code is meant to fit the field to the widest
> of its type, so it should infer fractional fields as doubles rather than
> floats.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
