[
https://issues.apache.org/jira/browse/PARQUET-1991?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17312470#comment-17312470
]
Antoine Pitrou commented on PARQUET-1991:
-----------------------------------------
Several things:
1) this is a Parquet C++ implementation blunder that should not have happened;
it doesn't sound very desirable to add a placeholder in the format to guard
against misinterpretations of those files
2) converted types are superceded by logical types, so it sounds unlikely that
the spec would add new converted types and end up specifying a converted type
numbered 24
3) interpretation of said converted type is not necessary, since the NullType
logical type should be there as well
I would personally recommend closing as won't fix. [~gszadovszky] Do you have
an opinion on this?
> Reserve ConvertedType==24 due to bug in parquet-cpp implementation
> ------------------------------------------------------------------
>
> Key: PARQUET-1991
> URL: https://issues.apache.org/jira/browse/PARQUET-1991
> Project: Parquet
> Issue Type: Improvement
> Components: parquet-format
> Reporter: Micah Kornfield
> Priority: Major
>
> https://issues.apache.org/jira/browse/PARQUET-1990 sometimes a converted type
> of 24 is written out because we are leaking a C++ enum value that doesn't
> exist in the parquet thrift specification. We should reserve this so if any
> other converted type elements are added future readers of older files don't
> accidentally misinterpret this value.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
