[
https://issues.apache.org/jira/browse/HIVE-28750?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
László Bodor updated HIVE-28750:
--------------------------------
Description:
1. we set HiveOperation.QUERY without any explanation, this might cause
problems on codepaths that rely on the real hive operation
https://github.com/apache/hive/blob/35589d78ca428f6274e8fb0f6399d70b947581bb/ql/src/java/org/apache/hadoop/hive/ql/parse/SemanticAnalyzer.java#L12955
SemanticAnalyzerFactory tries to set up operation (into QueryState) according
to the analyzer class, so every overwrite later implies a workaround/hacky, and
here it's not even clear why
2. ANALYZE TABLE is HiveOperation.QUERY but ANALYZE TABLE ... FOR COLUMNS is
HiveOperation.ANALYZE_TABLE, which is confusing. Fixing this would involve
changing hundreds of q.outs
was:
1. we set HiveOperation.QUERY without any explanation, this might cause
problems on codepaths that rely on the real hive operation
https://github.com/apache/hive/blob/35589d78ca428f6274e8fb0f6399d70b947581bb/ql/src/java/org/apache/hadoop/hive/ql/parse/SemanticAnalyzer.java#L12955
2. ANALYZE TABLE is HiveOperation.QUERY but ANALYZE TABLE ... FOR COLUMNS is
HiveOperation.ANALYZE_TABLE, which is confusing. Fixing this would involve
changing hundreds of q.outs
> Fix HiveOperation.QUERY usage in semantic analyzer and ANALYZE TABLE
> statements
> -------------------------------------------------------------------------------
>
> Key: HIVE-28750
> URL: https://issues.apache.org/jira/browse/HIVE-28750
> Project: Hive
> Issue Type: Improvement
> Reporter: László Bodor
> Priority: Major
>
> 1. we set HiveOperation.QUERY without any explanation, this might cause
> problems on codepaths that rely on the real hive operation
> https://github.com/apache/hive/blob/35589d78ca428f6274e8fb0f6399d70b947581bb/ql/src/java/org/apache/hadoop/hive/ql/parse/SemanticAnalyzer.java#L12955
> SemanticAnalyzerFactory tries to set up operation (into QueryState) according
> to the analyzer class, so every overwrite later implies a workaround/hacky,
> and here it's not even clear why
> 2. ANALYZE TABLE is HiveOperation.QUERY but ANALYZE TABLE ... FOR COLUMNS is
> HiveOperation.ANALYZE_TABLE, which is confusing. Fixing this would involve
> changing hundreds of q.outs
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
