[
https://issues.jboss.org/browse/TEIID-1473?page=com.atlassian.jira.plugin...
]
Howard Abrams commented on TEIID-1473:
--------------------------------------
Completely agree about best practices.
This came up in our project and seemed like something the engine could make a better guess
at, so I figured it was worth mentioning and logging even though we are making sure all
our translators now implement limit and offset.
Regarding batch size assumptions, doesn't the translator already get the batch size in
the ExecutionContext? I would think most "smart" translators would make
assumptions based on that size, not their total size. If this enhancement was added, I was
assuming the batch size sent in the context could be set to the limit, and therefore the
translator would get the hints it needed.
Don't batch more rows than is needed when query contains limit
but translator does not support limit
----------------------------------------------------------------------------------------------------
Key: TEIID-1473
URL:
https://issues.jboss.org/browse/TEIID-1473
Project: Teiid
Issue Type: Enhancement
Reporter: Howard Abrams
Assignee: Steven Hawkins
Priority: Minor
Teiid shouldn't try to batch more rows than is needed when query contains limit and
translator does not support limit.
For example, say there is a translator that exposes a table "A", but does not
support limit.
select * from A limit 2
will currently select an entire batch of rows (1024?) from A, instead of just stopping
after 2 rows.
Likewise, if the translator supports neither offset nor limit, and the query contains
both, Teiid should not try to batch more that (offset + limit) rows.
This is marked as "Minor" because the workaround of naively supporting limit
and offset within a translator is not too hard.
--
This message is automatically generated by JIRA.
For more information on JIRA, see:
http://www.atlassian.com/software/jira