[
https://issues.apache.org/jira/browse/PARQUET-758?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17598185#comment-17598185
]
ASF GitHub Bot commented on PARQUET-758:
----------------------------------------
gszadovszky commented on PR #184:
URL: https://github.com/apache/parquet-format/pull/184#issuecomment-1232491783
I've came up with this ordering thing because we specify it for every
logical types. (Unfortunately we don't do this for primitive types.) Therefore,
I would expect to have the order specified for this new logical type as well
which is not trivial and requires to solve
[PARQUET-1222](https://issues.apache.org/jira/browse/PARQUET-1222). At least we
should add a note about this issue.
> It seems this would require parquet implementations to null out statistics
for logical types that they don't support, does parquet-mr do that today?
I do not have the proper environment to test it but based on the code we do
not handle unknown logical types well in parquet-mr. I think it handles unknown
logical types as if they were not there at all which is fine from the data
point of view but we would blindly use the statistics which may cause data
loss. Created
[PARQUET-2182](https://issues.apache.org/jira/browse/PARQUET-2182) to track
this.
> [Format] HALF precision FLOAT Logical type
> ------------------------------------------
>
> Key: PARQUET-758
> URL: https://issues.apache.org/jira/browse/PARQUET-758
> Project: Parquet
> Issue Type: Improvement
> Components: parquet-format
> Reporter: Julien Le Dem
> Priority: Minor
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
