[jira] [Updated] (PHOENIX-1807) Support UNION queries in subquery

["Maryann Xue (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/PHOENIX-1807?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Maryann Xue updated PHOENIX-1807:
---------------------------------
    Attachment: PHOENIX-1807.patch

1. Fix union query top-level "select" nodes with a list of adjusted aliases from a non-wildcard sub-select if any:
    a) No check the number of select nodes among all sub-selects, since it will be done later at compile stage. 
    b) We simply create a temporary alias if no column name or alias is present, since it cannot be referred by outer queries.

2. Remove constructor of SelectStatement without "sub-selects", in order to force awareness of sub-selects when constructing a SelectStatement.
    There are quite a few different signatures of select() method in ParseNodeFactory now, and I think we should follow a convention for calling these select() methods:
    1) To rewrite a SelectStatement by replacing some part of it or adding a new clause to it, we should use methods like select(Statement select, ...), which copies every part of the old SelectStatement to the new one unless specified otherwise, including the *sub-selects*. For example, filter push-down or order-by push-down.
    2) To construct a new SelectStatement using none or some parts of an existing SelectStatement, we should call select(TableNode from, HintNode hint, ...) with the full parameter list. For example, construction of a SelectStatement in DeleteCompiler.

> Support UNION queries in subquery
> ---------------------------------
>
>                 Key: PHOENIX-1807
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-1807
>             Project: Phoenix
>          Issue Type: New Feature
>            Reporter: Maryann Xue
>            Assignee: Maryann Xue
>             Fix For: 4.4.0
>
>         Attachments: PHOENIX-1807.patch
>
>   Original Estimate: 48h
>  Remaining Estimate: 48h
>
> Example:
> select * from aTable where pk in (select pk from idx1 where q1 = 'x' union all select pk from idx2 where q2 = 'y')



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