[ 
https://issues.apache.org/jira/browse/HIVE-15891?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15872349#comment-15872349
 ] 

Wei Zheng commented on HIVE-15891:
----------------------------------

I think it should, because as long as query involves update operations on acid 
table we have no guarantee how the AST will be after the rewrite. So we should 
throw the error. Current row filtering implementation has a vague distinction 
of "not supporting acid". I think this should fall in the "unsupported" 
category for now, until the AST manipulation is consistent.

> Detect query rewrite scenario for UPDATE/DELETE/MERGE and fail fast
> -------------------------------------------------------------------
>
>                 Key: HIVE-15891
>                 URL: https://issues.apache.org/jira/browse/HIVE-15891
>             Project: Hive
>          Issue Type: Bug
>          Components: Transactions
>    Affects Versions: 2.2.0
>            Reporter: Wei Zheng
>            Assignee: Wei Zheng
>         Attachments: HIVE-15891.1.patch, HIVE-15891.2.patch, 
> HIVE-15891.3.patch
>
>
> Currently ACID UpdateDeleteSemanticAnalyzer directly manipulates the AST tree 
> but it's different from the general approach of modifying the token stream 
> and thus will cause AST tree mismatch if there is any rewrite happening after 
> UpdateDeleteSemanticAnalyzer.
> The long term solution will be to rewrite the AST handling logic in 
> UpdateDeleteSemanticAnalyzer, to make it consistent with the general approach.
> This ticket will for now detect the error prone cases and fail early. 



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to