[jira] [Commented] (TAJO-509) Cannot use TajoResultSet.getString() if the identifier has dots in it

["Hyunsik Choi (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/TAJO-509?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13874569#comment-13874569 ] 

Hyunsik Choi commented on TAJO-509:
-----------------------------------

It's a Tajo SQL parser problem. The current parse deals with alias name as an identifier as follows:
{code}
derived_column
  : value_expression as_clause?
  ;

column_reference
  : (tb_name=Identifier DOT)? name=Identifier
  ;

as_clause
  : (AS)? Identifier
  ;
{code}

As far as I know, some DBMSs (like PostgreSQL) do so as the current Tajo. Therefore, in my point of view, its a decidable issue. If its necessary is reasonable or there is any volunteer, we can support identifier including dots.

> Cannot use TajoResultSet.getString(<identifier>) if the identifier has dots in it
> ---------------------------------------------------------------------------------
>
>                 Key: TAJO-509
>                 URL: https://issues.apache.org/jira/browse/TAJO-509
>             Project: Tajo
>          Issue Type: Bug
>            Reporter: Daniel Einspanjer
>            Priority: Minor
>
> If you submit via the TajoClient a select statement similar to "SELECT foo as 'a.b' FROM bar" and then attempt to get the 'a.b' column value by name, Tajo cannot find it because it is using code in Catalog that splits the identifier into a qualifier and name.
> I don't believe the label applied to a SELECT clause expression should be tokenized in such a way.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)