[
https://issues.apache.org/jira/browse/PHOENIX-2802?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
James Taylor updated PHOENIX-2802:
----------------------------------
Attachment: PHOENIX-2802.patch
Please review, [~samarthjain]. It turns out we usually don't need a merge sort
if we're doing a ORDERED aggregation. There's a bit more optimization we can do
that I'll do in a follow up JIRA.
> Reverse scan of in-place, ordered aggregation may produce incorrect results
> ---------------------------------------------------------------------------
>
> Key: PHOENIX-2802
> URL: https://issues.apache.org/jira/browse/PHOENIX-2802
> Project: Phoenix
> Issue Type: Improvement
> Reporter: James Taylor
> Assignee: James Taylor
> Fix For: 4.8.0
>
> Attachments: PHOENIX-2802.patch
>
>
> Try something like the following query:
> {code}
> select pod, count(*) as orgs from organization group by pod order by pod desc;
> {code}
> versus this one:
> {code}
> select pod, count(*) as orgs from organization group by pod order by pod asc;
> {code}
> There may be some missing logic in how we setup the client-side order by in
> this case.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
