[
https://issues.apache.org/jira/browse/FLINK-5554?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15829453#comment-15829453
]
ramkrishna.s.vasudevan commented on FLINK-5554:
-----------------------------------------------
Thanks for starting this thread via JIRA [~tonycox].
[~fhueske]
I was acutally asking the same thing in the other JIRA FLINK-2168 as how does
the family model will get translated into calcite. In Apache Phoenix projects
it understands the language of Calcite and HBase and it tries to act as the
wrapper because the family name is actually created and maintained by Phoenix.
So for the SQL layer it is purely relational query.
But here in Flink that is not the case. We know there is a table in HBase but
the data model is purely non relational.
bq.What should work straightforward is to have a TableSource which maps the
schema of an HBase table into a relational schema.
Not sure on that.
bq.I think the HBaseTableSource should return a composite type (with column
family and qualifier), and we can get columns by composite type accessing.
Sorry I am not getting this. Can some one help throw more light on this. I am
interested to work here and get the entire thing working. Thanks [~jark].
My other suggestion would be will it make sense to add a PhoenixTableSource so
that it becomes easy to take to relational data model of Phoenix?
> Add sql operator to table api for getting columns from HBase
> ------------------------------------------------------------
>
> Key: FLINK-5554
> URL: https://issues.apache.org/jira/browse/FLINK-5554
> Project: Flink
> Issue Type: Improvement
> Components: Table API & SQL
> Reporter: Anton Solovev
>
> example of select query
> {code}
> table.select("f1:q1, f1:q2, f1:q3");
> {code}
> or/and
> {code}
> table.select('f1:'q1, 'f1:'q2, 'f1:'q3);
> {code}
> let's discuss how to provide better api for selecting from HBase
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
