[
https://issues.apache.org/jira/browse/PHOENIX-2315?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
James Taylor updated PHOENIX-2315:
----------------------------------
Attachment: PHOENIX-2315.patch
Please review, [~samarthjain]. We were lucky (or unlucky) not to see this issue
earlier. When an expression is used to group by, the expression.getMaxLength()
will be set to the precision of the calculated value (in this case it ends up
being the max precision for a long which is 38). We, instead, need to use
expression.getMaxLength() only if getDataType().getByteSize() returns null.
> Group by fields of primary key columns together with non-primary key columns
> can give wrong result
> --------------------------------------------------------------------------------------------------
>
> Key: PHOENIX-2315
> URL: https://issues.apache.org/jira/browse/PHOENIX-2315
> Project: Phoenix
> Issue Type: Bug
> Reporter: Alicia Ying Shu
> Assignee: James Taylor
> Fix For: 4.7.0
>
> Attachments: PHOENIX-2315.patch
>
>
> create table tgb_counter(tgb_id integer NOT NULL,utc_date_epoch integer NOT
> NULL,tgb_name varchar(40),ack_success_count integer,ack_success_one_ack_count
> integer, CONSTRAINT pk_tgb_counter PRIMARY KEY(tgb_id, utc_date_epoch));
> SELECT tgb_id, tgb_name, (utc_date_epoch/10)*10 AS
> utc_epoch_hour,SUM(ack_success_count + ack_success_one_ack_count) AS
> ack_tx_sum FROM tgb_counter GROUP BY tgb_id, tgb_name, utc_epoch_hour;
> tgb_name returns NULL.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
