[
https://issues.apache.org/jira/browse/DRILL-4572?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15222606#comment-15222606
]
Jinfeng Ni commented on DRILL-4572:
-----------------------------------
This seems to be a bug in the query planner. I could re-produce this
problem, with similar query, even on 1.4 and 1.5.
select rtrim(s.name2) as name3
from (
select
ltrim(n_name) as `name2`,
count(*) as num_creates
from cp.`tpch/nation.parquet`
group by ltrim(n_name)
order by name2 ASC
) s limit 100
Error: SYSTEM ERROR: AssertionError: Internal error: Type
'RecordType(ANY $f0, BIGINT num_creates)' has no field 'name2'
> Column Alias Not Available in Subquery
> --------------------------------------
>
> Key: DRILL-4572
> URL: https://issues.apache.org/jira/browse/DRILL-4572
> Project: Apache Drill
> Issue Type: Bug
> Components: Query Planning & Optimization
> Affects Versions: 1.4.0, 1.5.0, 1.6.0
> Reporter: John Omernik
>
> In the query and error below, I have a "inside" aggregate query where the
> groupped field is the result of a function, but I use an alias (`cdate`) to
> reference it. In the outer part of the query, I try to reference that alias
> and the query fails with the error. The error seems to imply it can't see the
> aliased field.
> Jinfeng Ni was also able to reproduce based on the mailing list.
> select date_part('year', s.cdate) as ydate from (
> select
> to_date(createdon, 'yyyy-MM-dd HH:mm:ss') as `cdate`, count(*) as
> num_creates
> from
> view_mytable
> where
> createdon <> '0000-00-00 00:00:00' and createdon >= '2010-01-10'
> group by
> to_date(createdon, 'yyyy-MM-dd HH:mm:ss')
> order by cdate ASC
> ) s limit 100
> SYSTEM ERROR: AssertionError: Internal error: Type 'RecordType(ANY $f0,
> BIGINT num_creates)' has no field 'cdate'
> [Error Id: 50caacaf-0b9b-4c1c-a3e1-53248bc385c0 on
> hadoopmapr6.brewingintel.com:31010]
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
