[
https://issues.apache.org/jira/browse/CALCITE-724?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14542515#comment-14542515
]
Julian Hyde commented on CALCITE-724:
-------------------------------------
Is it correct to call this a hash join? I get the idea that you take one row
from the left and then use it to lookup rows on the right.
The distinction is important. With a hash join the "key" you get from the left
would have to be a single value. If the data structure on the right is, say, a
b-tree, then you the condition could be a single value, but it could also be a
range, or even a collection of ranges, say "x is null or x between 100 and 200
or x = 500 or x > 1000". If we allow a fairly general expression then we will
be able to fully exploit whatever search capabilities the index has.
We used to have support for "sargs" in Calcite but I obsoleted them. See
https://github.com/apache/incubator-calcite/tree/8e480759fc25d7a1e10f7f20f9eddc7556beded8/core/src/main/java/org/eigenbase/sarg.
We could revive them.
> Add support for hash join using an index
> ----------------------------------------
>
> Key: CALCITE-724
> URL: https://issues.apache.org/jira/browse/CALCITE-724
> Project: Calcite
> Issue Type: Improvement
> Affects Versions: 1.3.0-incubating
> Reporter: Andy Grove
> Assignee: Julian Hyde
>
> I would like to implement a hash join where the left input is a stream or a
> table scan and for each incoming tuple I would like Calcite to perform an
> index-based lookup against a table to find matching tuples.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
