[jira] [Commented] (LENS-813) For multifact queries, having clauses are getting added to both sub queries.

["Rajat Khandelwal (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/LENS-813?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15088806#comment-15088806 ] 

Rajat Khandelwal commented on LENS-813:
---------------------------------------

Posting the thought process here:

* Having clause has a boolean expression
* Hence, having AST is the AST of a boolean expression
* For multi-fact queries, the AST is copied to both facts and then both facts independently resolve expression in the AST recursively.
  ** If a fact is unable to resolve an expression, it just leaves it as it is.
* Multi-fact queries are by-design queries involving measures not present in a single fact. 
* The measures in multi-fact queries are split over the facts. 
* Having clause only involves conditions over measures and expressions over them. 
* For a multi-fact query, the whole having clause makes sense only after all the measures are in context, which is in the outer query. 
* Special cases of having clause can be pushed down. Necessary conditions:
  ** The boolean expression should be *and* over multiple boolean expressions. 
  ** Each single boolean expression should be evaluable by a single fact. 


For the sake of correctness, I'm planning to implement the easier version first, which is to have *having* at the outer query level and the ingredients at inner query level.

> For multifact queries, having clauses are getting added to both sub queries. 
> -----------------------------------------------------------------------------
>
>                 Key: LENS-813
>                 URL: https://issues.apache.org/jira/browse/LENS-813
>             Project: Apache Lens
>          Issue Type: Bug
>          Components: cube
>            Reporter: Rajat Khandelwal
>




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)