[
https://issues.apache.org/jira/browse/PARQUET-214?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ryan Blue updated PARQUET-214:
------------------------------
Description: Older version of parquet-avro evidently didn't correctly use
the UTF8 annotation and converted a string to binary. This worked because after
the problem was fixed, parquet-avro used the schema based on the requested Avro
projection schema, which correctly had the UTF8 annotation. The problem was
that this assumed that the requested projection schema was satisfied by the
file schema, which was fixed in PARQUET-139. However, that used the fields from
the file schema that can be missing UTF8 annotations.
> Avro: Regression caused by correctly projecting schema
> ------------------------------------------------------
>
> Key: PARQUET-214
> URL: https://issues.apache.org/jira/browse/PARQUET-214
> Project: Parquet
> Issue Type: Bug
> Components: parquet-mr
> Affects Versions: 1.6.0
> Reporter: Ryan Blue
> Fix For: 1.6.0
>
>
> Older version of parquet-avro evidently didn't correctly use the UTF8
> annotation and converted a string to binary. This worked because after the
> problem was fixed, parquet-avro used the schema based on the requested Avro
> projection schema, which correctly had the UTF8 annotation. The problem was
> that this assumed that the requested projection schema was satisfied by the
> file schema, which was fixed in PARQUET-139. However, that used the fields
> from the file schema that can be missing UTF8 annotations.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
