You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In aggregate operator, we use the function name frist_value_respect_nulls to construct a aggregate function which bases on any, and the intermediate result type is AggregateFunction(any, nullable(string)). That is too bad, becasue there are several aggregate functions have the same intermediate result type AggregateFunction(any, nullable(string)), but their data structure may be different. For frist_value_respect_nulls, the intemediate result doesn't containt null flags.
ShuffleReader builds a deserializer from the type info AggregateFunction(any, nullable(string)). CH will construct a aggregate function any combined with AggregateFunctionNullBase to deserialize the data. The null flag takes one byte, that is why 3 bytes remained for any.
exmy
changed the title
[CH] Exception with aggerate function first query
[CH] Exception with aggerate function first/last
Nov 15, 2023
Backend
CH (ClickHouse)
Bug description
create table test_data(key int, value string) stored as orc
When enable
replaceSortAggWithHashAgg
,select first(value) from test_data group by key
throws exception:Spark version
None
Spark configurations
No response
System information
No response
Relevant logs
No response
The text was updated successfully, but these errors were encountered: